阅读说明：本技术 风速预测模型训练方法、预测方法、装置及电子设备 (Wind speed prediction model training method, prediction method and device and electronic equipment ) 是由 丁明月 俞海国 董凌 金金 于 2020-06-30 设计创作，主要内容包括：本申请实施例提供了一种风速预测模型训练方法、预测方法、装置及电子设备。本申请实施例提供的风速预测模型的训练方法,包括下列步骤：获取目标电场的目标实测样本风速,以及目标电场对应的上游区域电场的样本风速数据；根据目标电场与上游区域之间的风程时间间隔,确定包括若干个上游样本风速数据的样本风速组；根据目标实测样本风速和样本风速组,通过对初始的风速预测模型进行参数调整来进行训练,当达到预设的训练条件时结束训练,得到风速预测模型。本申请提供的风速预测模型的训练方法利用了大量的目标实测样本风速和目标电场对应的上游区域电场的样本风速数据,能够获取到精确的风速预测模型。(The embodiment of the application provides a wind speed prediction model training method, a prediction device and electronic equipment. The training method of the wind speed prediction model provided by the embodiment of the application comprises the following steps: acquiring target actual measurement sample wind speed of a target electric field and sample wind speed data of an electric field in an upstream area corresponding to the target electric field; determining a sample wind speed group comprising a plurality of upstream sample wind speed data according to a wind path time interval between a target electric field and an upstream area; and training by adjusting parameters of the initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and finishing the training when a preset training condition is reached to obtain the wind speed prediction model. According to the training method of the wind speed prediction model, a large amount of target actual measurement sample wind speed and sample wind speed data of an upstream area electric field corresponding to the target electric field are utilized, and the accurate wind speed prediction model can be obtained.)

1. A training method of a wind speed prediction model is characterized by comprising the following steps:

acquiring a target actual measurement sample wind speed of a target electric field and sample wind speed data of an upstream electric field in an upstream area corresponding to the target electric field;

determining a sample wind speed group comprising a plurality of upstream sample wind speed data according to a wind path time interval between the target electric field and an upstream electric field in the upstream region;

and training by adjusting parameters of an initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and finishing the training when a preset training condition is reached to obtain a wind speed prediction model.

2. A method of training a wind speed prediction model according to claim 1, wherein the upstream electric field and the wind path time interval are determined by a method comprising:

acquiring a first measured wind speed of a target electric field in a first time period;

acquiring second measured wind speeds of all electric fields except the target electric field in each wind speed acquisition time period at preset time intervals;

calculating a correlation value of the first measured wind speed and each second measured wind speed according to a preset incidence relation;

and if the correlation value is larger than a preset correlation threshold value, determining that the electric field except the target electric field corresponding to the correlation value is an upstream electric field and the wind path time interval corresponding to the upstream electric field.

3. A method of training a wind speed prediction model according to claim 2, wherein the upstream region is determined by a method comprising the steps of:

determining all upstream electric fields of the target electric field;

clustering all the upstream electric fields of the target electric field into a plurality of upstream areas according to a preset clustering method; the upstream electric field in each upstream area is within a preset geographical position range;

an upstream zone is determined among the plurality of upstream zones.

4. The method for training a wind speed prediction model according to claim 1, wherein the determining a sample wind speed group comprising a number of sample wind speed data according to a wind path time interval between the target electric field and an upstream electric field in the upstream region comprises:

acquiring a minimum wind path time interval and a maximum wind path time interval corresponding to an upstream electric field in an upstream area;

and according to a preset time step, determining the upstream sample wind speed data one by one between the minimum wind path time interval and the maximum wind path time interval.

5. The method for training a wind speed prediction model according to claim 1, wherein the step of obtaining a target measured sample wind speed of a target electric field comprises: and taking the numerical prediction wind speed obtained by numerical prediction as the target actual measurement sample wind speed.

6. A method of wind speed prediction, comprising:

acquiring measured wind speed data of all upstream electric fields in an upstream area of a target electric field;

obtaining a wind speed prediction model determined by a wind speed prediction model training method provided by any one of claims 1-5;

and inputting the actually measured wind speed data into the wind speed prediction model to obtain the predicted wind speed of the target electric field.

7. The method of claim 6, wherein after determining the predicted wind speed for the target electric field, further comprising:

acquiring an actual measurement wind speed of a target electric field corresponding to the predicted wind speed;

if the difference value between the actually measured wind speed of the target electric field and the predicted wind speed is larger than a preset error condition, re-acquiring sample wind speed data of all electric fields in other upstream areas of the target electric field, and training the wind speed prediction model; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

8. The method of claim 6, wherein after determining the predicted wind speed for the target electric field, further comprising:

acquiring an actual measurement wind speed of a target electric field corresponding to the predicted wind speed;

if the difference value between the actually measured wind speed of the target electric field and the predicted wind speed is larger than a preset error condition, other wind speed prediction models are replaced, and the wind speed prediction models are trained; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

9. A wind speed prediction model training device is characterized by comprising:

the wind speed acquisition module is used for acquiring a target actual measurement sample wind speed of a target electric field and sample wind speed data of an upstream electric field in an upstream area corresponding to the target electric field;

the grouping module is used for determining a sample wind speed group comprising a plurality of sample wind speed data according to the wind path time interval between the target electric field and the upstream electric field in the upstream area;

and the training module is used for carrying out model training by carrying out parameter adjustment on the initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and finishing the training when a preset training condition is reached to obtain the wind speed prediction model.

10. The training device of the wind speed prediction model according to claim 9, wherein the wind speed obtaining module comprises a wind speed selecting unit, a calculating unit and an output unit;

the wind speed selecting unit is used for acquiring a first measured wind speed of a target electric field in a first time interval; acquiring second measured wind speeds of all electric fields except the target electric field in each wind speed acquisition time period at preset time intervals;

the calculation unit is used for calculating a correlation value of the first measured wind speed and each second measured wind speed according to a preset incidence relation;

the output unit is used for determining that the electric field except the target electric field corresponding to the correlation value is an upstream electric field and a wind path time interval corresponding to the upstream electric field if the correlation value is larger than a preset correlation threshold.

11. A wind speed prediction device, comprising:

the data acquisition module is used for acquiring the actually measured wind speed data of all the upstream electric fields in one upstream area of the target electric field;

the model acquisition module is used for acquiring a wind speed prediction model;

and the data prediction module is used for inputting the actually measured wind speed data into the wind speed prediction model and operating to determine the predicted wind speed of the target electric field.

12. An electronic device, comprising:

a processor, a memory, and a bus;

the bus is used for connecting the processor and the memory;

the memory is used for storing operation instructions;

the processor is configured to implement the wind speed prediction model training method according to any one of claims 1 to 5 or implement the wind speed prediction method according to any one of claims 6 to 8 by calling the operation instruction.

13. A computer-readable storage medium, characterized in that the computer storage medium is adapted to store a computer program for implementing a wind speed prediction model training method according to any of the claims 1-5 or for implementing a wind speed prediction method according to any of the claims 6-8, when the computer program is run in an electronic device.

Technical Field

The application relates to the technical field of wind power, in particular to a wind speed prediction model training method, a prediction device and electronic equipment.

Background

The numerical weather forecast (numerical weather prediction) refers to a method for predicting an atmospheric motion state and a weather phenomenon in a future period by performing numerical calculation through a large-scale computer under certain initial value and side value conditions according to atmospheric actual conditions and solving a hydrodynamics and thermodynamics equation set describing a weather evolution process. That is, the numerical weather forecast uses data such as wind speed and wind direction as input quantities, and the output power of the wind farm can be predicted from the forecasted meteorological elements by a prediction algorithm. The accurate forecasting of the numerical weather forecast data can provide important decision support for power dispatching, and therefore the accuracy of the numerical weather forecast data is one of important decision factors of the prediction accuracy of the new energy power generation.

In the field of wind power generation, the wind speed change condition of a wind field needs to be predicted through numerical weather prediction, however, power prediction of a wind power plant requires that the wind speed is predicted once every 15min (minutes), namely, the wind speed is predicted once every 15min through the numerical weather prediction, and at present, the numerical weather prediction has the following two difficulties aiming at the wind speed prediction of the wind power plant:

1. since the numerical weather forecast leads or lags the forecast of the weather system movement, it is difficult to accurately forecast the time point of the sudden rising and falling of the wind speed.

2. The numerical weather forecast is called a mesoscale numerical weather forecast, a weather system is predicted only on a mesoscale (the horizontal scale is generally 15-300 kilometers), but the phenomenon of small-scale strong wind gusts at the position of a wind power plant is difficult to capture.

Disclosure of Invention

Aiming at the defects of the existing mode, the application provides a wind speed prediction model training method, a prediction device and electronic equipment, and aims to solve the technical problems that the wind speed prediction accuracy is insufficient or the prediction is limited by the scale size of a prediction mode in the prior art.

In a first aspect, an embodiment of the present application provides a training method for a wind speed prediction model, including the following steps:

acquiring a target actual measurement sample wind speed of a target electric field and sample wind speed data of an upstream electric field in an upstream area corresponding to the target electric field;

determining a sample wind speed group comprising a plurality of upstream sample wind speed data according to a wind path time interval between the target electric field and an upstream electric field in an upstream region;

and training by adjusting parameters of the initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and finishing the training when a preset training condition is reached to obtain the wind speed prediction model.

In certain implementations of the first aspect, the upstream electric field is determined by a method comprising:

acquiring a first measured wind speed of a target electric field in a first time period;

acquiring second measured wind speeds of all electric fields except the target electric field in a wind speed acquisition period at preset time intervals;

calculating a correlation value of the first measured wind speed and each second measured wind speed according to a preset incidence relation;

and if the correlation value is larger than the preset correlation threshold value, determining that the electric field except the target electric field corresponding to the correlation value is an upstream electric field and determining the wind path time interval corresponding to the upstream electric field.

With reference to the first aspect and the implementations described above, in some implementations of the first aspect, the upstream region is determined by a method comprising:

determining all upstream electric fields of the target electric field;

clustering all upstream electric fields of the target electric field into a plurality of upstream areas according to a preset clustering method; the upstream electric field in each upstream region is within a preset geographical location range;

an upstream zone is determined among the plurality of upstream zones.

With reference to the first aspect and the foregoing implementations, in some implementations of the first aspect, determining a sample wind speed group including a number of sample wind speed data according to a wind path time interval between the target electric field and an upstream electric field in an upstream region includes:

acquiring a minimum wind path time interval and a maximum wind path time interval corresponding to an electric field of an upstream area;

and according to the preset time step, determining the upstream sample wind speed data one by one between the minimum wind path time interval and the maximum wind path time interval.

With reference to the first aspect and the foregoing implementation manners, in some implementation manners of the first aspect, the step of acquiring the target measured sample wind speed of the target electric field includes: and taking the numerical prediction wind speed obtained by numerical prediction as the target actual measurement sample wind speed.

In a second aspect, the present application provides a wind speed prediction method, comprising:

acquiring measured wind speed data of all upstream electric fields in an upstream area of a target electric field;

acquiring a wind speed prediction model, wherein the wind speed prediction model is determined by a wind speed prediction model training method provided by the first aspect of the application;

and inputting the actually measured wind speed data into a wind speed prediction model to obtain the predicted wind speed of the target electric field.

In certain implementations of the first aspect, after determining the predicted wind speed for the target electric field, further comprising:

acquiring an actually measured wind speed of a target electric field corresponding to the predicted wind speed;

if the difference value between the actually measured wind speed and the predicted wind speed of the target electric field is larger than the preset error condition, re-acquiring sample wind speed data of all electric fields in other upstream areas of the target electric field, and training a wind speed prediction model; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

With reference to the second aspect and the foregoing implementations, in some implementations of the second aspect, after determining the predicted wind speed of the target electric field, further includes:

acquiring an actually measured wind speed of a target electric field corresponding to the predicted wind speed;

if the difference value of the actually measured wind speed and the predicted wind speed of the target electric field is larger than the preset error condition, other wind speed prediction models are repeated, and the wind speed prediction models are trained; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

In a third aspect, the present application provides a wind speed prediction model training apparatus, including:

the wind speed acquisition module is used for acquiring a target actual measurement sample wind speed of a target electric field and sample wind speed data of an upstream electric field in an upstream area corresponding to the target electric field;

the grouping module is used for determining a sample wind speed group comprising a plurality of sample wind speed data according to the wind path time interval between the target electric field and the upstream electric field in the upstream area;

and the training module is used for carrying out model training by carrying out parameter adjustment on the initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and finishing the training when the preset training condition is reached to obtain the wind speed prediction model.

In certain implementations of the third aspect, the wind speed acquisition module includes a wind speed selection unit, a calculation unit, and an output unit;

the wind speed selecting unit is used for acquiring a first measured wind speed of a target electric field in a first time interval; acquiring second measured wind speeds of all electric fields except the target electric field in each wind speed acquisition time period at preset time intervals;

the calculating unit is used for calculating a correlation value of the first measured wind speed and each second measured wind speed according to a preset incidence relation;

the output unit is used for determining that the electric field except the target electric field corresponding to the correlation value is an upstream electric field and a wind path time interval corresponding to the upstream electric field if the correlation value is larger than a preset correlation threshold.

In a fourth aspect, the present application provides a wind speed prediction apparatus, comprising:

the data acquisition module is used for acquiring the actually measured wind speed data of all the upstream electric fields in one upstream area of the target electric field;

the model acquisition module is used for acquiring a wind speed prediction model;

and the data prediction module is used for inputting the actually measured wind speed data into the wind speed prediction model and operating to determine the predicted wind speed of the target electric field.

In a fifth aspect, the present application provides an electronic device, comprising:

a processor, a memory, and a bus;

a bus for connecting the processor and the memory;

a memory for storing operating instructions;

a processor for implementing the wind speed prediction model training method as described in the first aspect of the present application or implementing the wind speed prediction method as described in the second aspect of the present application by calling an operation instruction.

In a sixth aspect, the present application provides a computer-readable storage medium for storing a computer program for implementing a wind speed prediction model training method as described in the first aspect of the present application or for implementing a wind speed prediction method as described in the second aspect of the present application, when the computer program is run in an electronic device.

The technical scheme provided by the embodiment of the application has the following beneficial technical effects:

according to the training method of the wind speed prediction model, wind speed data with accurate upstream and downstream relations are screened out, a sample wind speed group of an upstream area electric field related to a downstream target electric field is selected, and the sample wind speed group is input into the determined wind speed prediction model to train the model.

According to the wind speed prediction method, the wind speed of the upstream electric field of the target electric field is monitored, and the trained wind speed prediction model is combined, so that the incidence relation between the upstream electric field and the wind speed in the downstream electric field is fully utilized, the wind speed of the downstream electric field is predicted accurately, and the wind speed prediction method is not limited by a scale range in numerical weather forecast.

Additional aspects and advantages of the present application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present application.

Drawings

The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

FIG. 1 is a schematic flow chart illustrating a method for training a wind speed prediction model according to an embodiment of the present disclosure;

FIG. 2 is a flowchart illustrating a method for determining an overall electric field in an upstream region of a target electric field according to an embodiment of the present disclosure;

FIG. 3 is a schematic flow chart illustrating a method for determining an upstream region of a target electric field according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of an upstream electric field distribution provided by an embodiment of the present application;

FIG. 5 is a schematic flow chart illustrating a wind speed prediction method according to an embodiment of the present disclosure;

FIG. 6 is a schematic structural framework diagram of a wind speed prediction model training device according to an embodiment of the present application

FIG. 7 is a schematic structural framework diagram of a wind speed prediction device according to an embodiment of the present disclosure;

FIG. 8 is a graph comparing predicted wind speed and measured wind speed obtained by different prediction devices according to an embodiment of the present application;

FIG. 9 is a comparison graph of the mean square error of the wind speed prediction device and the centralized prediction algorithm of the present application;

fig. 10 is a schematic structural framework diagram of an electronic device according to an embodiment of the present application.

Detailed Description

Reference will now be made in detail to the present application, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar parts or parts having the same or similar functions throughout. In addition, if a detailed description of the known art is not necessary for illustrating the features of the present application, it is omitted. The embodiments described below with reference to the drawings are exemplary only for the purpose of explaining the present application and are not to be construed as limiting the present application.

It will be understood by those within the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It should be understood that "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. As used herein, the term "and/or" includes all or any element and all combinations of one or more of the associated listed items.

The deployment of wind power generation depends heavily on windy weather, the existing weather forecast does not fully consider the geographical position distribution of a wind power generation field and the mutual influence among the wind power generation fields, the common numerical weather forecast predicts a weather system only on a medium scale, usually a horizontal scale of 15-300 kilometers, but the small-scale strong wind gust phenomenon of the position of the wind power generation field is difficult to capture. Therefore, the existing weather forecast cannot be accurate enough for forecasting the wind weather, and cannot provide accurate data reference for allocation of wind power generation.

The applicant of the present application considers that wind farms collecting wind energy are dispersed in areas rich in wind energy, the areas may be divided according to geographical administrative areas with different sizes, the geographical characteristics in the geographical administrative areas are usually fixed, and the influence of a certain day in the life cycle (from the time when the day enters a certain area to the time when the day leaves the same area) is usually stable, so that a method of predicting the wind speed of a downstream electric field through the wind speed of an upstream electric field can be adopted to accurately obtain the wind speed information of the electric field.

The following describes the technical solutions of the present application and how to solve the above technical problems with specific embodiments.

In order to accurately obtain wind speed data in different electric fields in a certain area, the wind speed data acquisition method needs to rely on a proper machine learning model, so that a wind speed prediction model based on the machine learning model which can be used is obtained firstly.

An embodiment of the first invention of the present application provides a method for training a wind speed prediction model, as shown in fig. 1, including the following steps:

s101: and acquiring target actual measurement sample wind speed of the target electric field and sample wind speed data of the upstream electric field in the upstream area corresponding to the target electric field.

S102: and determining a sample wind speed group comprising a plurality of upstream sample wind speed data according to the wind path time interval between the target electric field and the upstream electric field in the upstream area.

S103: and training by adjusting parameters of the initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and finishing the training when a preset training condition is reached to obtain the wind speed prediction model.

According to the training method of the wind speed prediction model provided by the embodiment of the application, the wind speed data with the accurate upstream and downstream relation are screened out for sufficient training, the machine learning model and a large amount of sample wind speed data with the target actual measurement sample wind speed and the upstream area electric field corresponding to the target electric field which are stably connected with each other are combined, and the accurate wind speed prediction model can be obtained through the training method based on the wind speed prediction model.

In S101, the target electric field used in S101 and the upstream electric field corresponding to the target electric field are a relative concept. Because the wind is very variable, the moving speed, moving direction and duration of the wind are problematic, the target electric field is to determine one electric field from a plurality of electric fields, and other electric fields having specific association with the electric field are used as upstream electric fields, which can also be called upstream area electric fields. For a certain pair of target and upstream electric fields, wind having a certain life cycle moves from the upstream electric field to the target electric field. The roles of the target and upstream zone farms are constantly changing for different dates of wind.

In addition, the target actual measurement sample wind speed and the sample wind speed data of the electric field in the upstream region corresponding to the target electric field mentioned in S101 are actually derived from the historical wind speed data of each electric field in a certain time period, and for the convenience of description, a distinction is made, and it should be understood that there is no essential difference between the two. A plurality of wind power plants may be included in a certain geographic area, and when one wind power plant is taken as a target electric field, the following determination needs to be carried out through a certain method: which are the electric fields of the upstream region of the determined target electric field during some same historical period. In this geographic region, not all electric fields outside the target electric field are upstream region electric fields that have an interaction.

Alternatively, as shown in fig. 2, the upstream farm and the wind interval in S101 are determined by a method including the steps of:

s1011: acquiring a target electric field in a first period t₀Is measured at a first measured wind speed v₀。

S1012: all electric fields except the target electric field are acquired at a preset time interval delta t and are respectively acquired in each wind speed acquisition time period t_iSecond measured wind speed v_i，t_i＝t₀+ n · Δ t, n is a positive integer.

S1013: calculating a first measured wind speed v according to a preset incidence relation₀With each second measured wind speed v_iThe correlation value of (2).

S1014: and if the correlation value is larger than the preset correlation threshold value, determining that the electric field except the target electric field corresponding to the correlation value is an upstream electric field and determining the wind path time interval corresponding to the upstream electric field.

The foregoing step S1011 is to determine a target electric field a in a certain geographic area, and then determine a time period, such as a first time period t, in the geographic area₀The average wind speed in 1 hour is taken from 9 to 10 points in 23 evening of 12 months and 23 months in 2017. Of course, the average wind speed in a finer time period, such as the above-mentioned average wind speed from 9 points to 9 points, may be obtained according to the processing capability of the equipment. For B, C, D, E, F, etc. electric fields other than the a electric field, since some of these electric fields are likely to be upstream electric fields of the a electric field, it is necessary to find the electric fields of these upstream areas.

S1012 is acquiring wind speed data of the electric field of the possible upstream area. Determining that the A electric field is a target electric field, and assuming that the B electric field is an upstream electric field of the A electric field, the wind speed of the A electric field is v within the time period from 9 to 10 points in 23 evening in 12 and 23 months in 2017 and 12 months₀In this time period, the wind blowing through the B electric field does not reach the a electric field, but needs to move for a certain wind path (the distance between the a electric field and the B electric field) and can correspond to the wind speed of the a electric field after a certain wind path time interval. Therefore, the wind speed of the A electric field is v within the time period from 9 to 10 points in the evening of 12 and 23 days in 2017 and 12 months in the A electric field₀In this case, the B-field may need to take a second measured wind speed v after 1 hour, or after 2 hours₁Namely, the second measured wind speed v of the B electric field is taken in the period from 10 to 11 points in the night of 12 and 23 days in 2017₂Or take 2Wind speed data v of 11 o ' clock to 12 o ' clock in 23 evening of 12 o ' clock of 017 years and 12 months₂. If the B field is upstream of the A field, then v₀And v₁And v₂There is a certain associative relationship.

In S1013, a correlation value between the two is calculated by presetting the correlation relationship, for example, a correlation relational expression, specifically an expression 1, may be adopted to obtain the correlation value.

Wherein r is a correlation value between the element X and the element Y, and the correlation value can judge the degree of correlation between the two elements. For example, the first measured wind speed of the target farm may be defined as element X, and the wind speed data of the farm in the upstream area after a certain interval of the wind path may be defined as element Y. To obtain an accurate correlation evaluation value, a first measured wind speed within a period of time and a second measured wind speed within a period of time, for example, the first wind speed and the second measured wind speed within a 1-hour period are adopted, and a plurality of instantaneous wind speeds exist within the 1 hour period, for example, an instantaneous wind speed is measured every 10 minutes, and X in expression 1_iI.e. the instantaneous wind speed data in the first measured wind speed, andis an average derived from instantaneous wind speed data in a plurality of first measured wind speeds, and accordingly, Y_iI.e. the instantaneous wind speed data in the second measured wind speed, andis an average derived from the instantaneous wind speed data in the plurality of first measured wind speeds. N in expression 1 is the number of instantaneous wind speed data.

Through statistical analysis on historical data, the electric fields corresponding to the correlation values exceeding a preset correlation threshold can be determined to have an upstream and downstream correlation relationship. For example, the predetermined correlation threshold may be 0.8, and all electric fields other than the a electric field having a correlation value greater than 0.8 are all used as the electric fields in the upstream region of the a electric field.

Under the condition that the upstream electric field of the target electric field is not known, the corresponding second measured wind speed v needs to be obtained by gradually increasing the preset time interval delta t, namely increasing the value of n_iContinuously checking the correlation between the second measured wind speed and the first measured wind speed to determine the correct second measured wind speed and the corresponding t_i＝t₀N in + n.DELTA.t. When the correlation values corresponding to different second measured wind speeds are calculated through S1013, and it is determined through S1014 that the correlation values meet the second measured wind speeds and the upstream electric field of the preset requirement, the value of n is also determined, how many preset time intervals need to pass are determined, and the wind path time interval corresponding to the upstream electric field is also correspondingly determined.

In the process of data processing, in order to make the data closer to reality, the electric field with the correlation value larger than the preset correlation threshold value needs to be further screened. For example, the second measured wind speed v of a certain electric field B at 1 hour, 2 hours and 3 hours apart in the interval of the wind path time_B1、v_B2And v_B3And second measured wind speeds v of the electric field C at 10 hours, 11 hours and 12 hours_C1、v_C2And v_C3And the correlation values with the first measured wind speed of the A electric field are respectively greater than 0.8, and the specific values are as follows in the following table 1:

TABLE 1 correlation numerical table of target electric field and upstream electric field

According to the table, the wind course time interval corresponding to the maximum value in the correlation values is selected, for example, the wind course time interval in the electric field B is 2 hours, and the wind course time interval in the electric field C is 11 hours.

Wind speed prediction is a very complicated task, and in order to acquire accurate prediction data as much as possible, factors such as different months and different seasons need to be considered. The electric field of the entire upstream area of the target electric field can be found according to the method described above, but in these fields, there are still differences, some of which are more accurate, in the degree of influence on the target electric field in different seasons and months.

Then, optionally, in a specific implementation manner of the embodiment of the present application, as shown in fig. 3, the upstream area in S101 is determined by a method including the following steps:

s1015: the total upstream electric field from the target electric field is determined.

S1016: clustering all upstream electric fields of the target electric field into a plurality of upstream areas according to a preset clustering method; the upstream electric field in each upstream region is within a preset geographic location range.

S1017: an upstream zone is determined among the plurality of upstream zones.

The specific method for determining all the upstream electric fields in S1015 can adopt the foregoing S1011 to S1014, and in fact, each electric field has a corresponding upstream electric field in the weather of a certain life cycle. In order to facilitate understanding of the contents of the schemes provided by the present implementation, a set of correlated upstream and downstream electric fields is still selected for analysis, and the set of upstream and downstream electric fields includes a target electric field and an upstream electric field other than the target electric field. S1016, clustering and dividing the data according to a preset clustering method, namely according to a certain commonality. Specifically, for example, k-means + + classification is performed according to latitude and longitude, and a plurality of regions are automatically divided.

Because each electric field is distributed on a specific geographic position, namely a specific longitude and latitude, and the distance between every two electric fields is far or near, an electric field layout chart clustered and distributed according to the longitude and latitude can be obtained. The upstream electric field in a certain area range is distributed in a plurality of upstream areas, the number of the electric field in each upstream area may be different, as shown in fig. 4, in which a pentagon represents a target electric field, and the upstream area in the figure includes a figure with 6 points different in number, such as a first upstream area including the upstream electric field represented by 3 triangles, a second upstream area including the upstream electric field represented by 6 circular rings, and so on. In different months or different seasons, the influence precision of different upstream regions on the target electric field is different, and therefore, an upstream region needs to be selected in S1017 to train the wind speed prediction model.

According to the foregoing description, the wind path time interval corresponding to each upstream electric field may be different, and therefore, when a plurality of upstream electric fields are included in the same upstream region, the wind path time interval has the maximum value. In a specific implementation manner of the foregoing embodiment of the present application, in S102, determining a sample wind speed group including several sample wind speed data according to a wind path time interval between the target electric field and the upstream electric field in the upstream region includes:

obtaining a minimum wind path time interval k corresponding to an upstream electric field_minAnd maximum wind path time interval k_max. According to a preset time step length T, at a minimum wind path time interval k_minAnd maximum wind path time interval k_maxAnd determining the upstream sample wind speed data one by one.

When a plurality of upstream regions of the target electric field are divided and one of the upstream regions is determined according to the foregoing implementation, the upstream regions generally include a plurality of upstream electric fields, and the wind speed data of all the upstream electric fields are used as training data. Each upstream electric field corresponds to a wind path time interval, so that a minimum wind path time interval k is obtained in the upstream area_minAnd maximum wind path time interval k_max. At a minimum wind path time interval k_minAnd maximum wind path time interval k_maxAccording to the method, a plurality of upstream sample wind speed data are determined in wind speed historical data according to a preset time step T, so that the wind speed prediction model can adopt more accurate and timely data during training, the limit of the scale size of the traditional numerical weather forecast is broken through when the wind speed prediction model carries out wind speed prediction, and the prediction result is obtained more quickly and accurately.

Following the foregoing example, for the six upstream electric fields in the second upstream region, it is assumed that the minimum wind path time interval is 2 hours, the maximum wind path time interval is 7 hours, and the preset time step is 15 minutes, respectivelyThe wind speed of the target electric field A in the time period from 9 to 10 points in 23 evening in 12 months in 2017 is determined as v₀Wind speed v of the upstream electric field B two hours after the above-mentioned period_B0Wind speed after 2 hours 15 minutes, wind speed after 2 hours 30 minutes, … …, wind speed at 6 hours, thus obtaining a set of sample wind speed groups. Similarly, for the upstream electric field C in the second upstream region, a set of sample wind speed sets can also be obtained.

After the training data is obtained, training of the wind speed prediction model is performed, and the specific training process in S103 is as follows:

training includes periodic outcome prediction and parameter modification.

The primary prediction comprises the following steps: a set of sample wind speed groups X is input to a wind speed prediction model, which outputs a target farm wind speed prediction Y1. And determining a deviation value between the predicted value Y1 of the wind speed of the target electric field and the measured value Y of the wind speed of the target electric field, and performing parameter correction according to the deviation value.

The parameter correction process comprises the following steps: and correcting the model parameters according to the deviation value. For example, when a neural network model is embodied, the deviation value is input into the model in a reverse direction, and each layer in the model calculates the partial derivative of the received parameter, and then the partial derivative is used as a weight to be multiplied by the original parameter of the layer.

And continuously predicting and correcting parameters according to the result until the deviation value obtained according to the model is smaller than a preset threshold value and/or training reaches preset times, and finishing training. More detailed model training details can be known by the relevant technical personnel and are not described in detail.

In order to obtain a more accurate prediction result, more abundant data needs to be input, and in a specific implementation manner of the embodiment of the present application, the step of obtaining the target measured sample wind speed of the target electric field in S101 includes: and taking the numerical prediction wind speed obtained by numerical prediction as the target actual measurement sample wind speed. The numerical forecast wind speed obtained through the numerical weather forecast is input into the wind speed forecasting model, and the numerical weather forecast data can be corrected. Of course, the input data are not limited to the historical observation data and the numerical weather forecast data, and may include other data affecting wind speed.

The method for predicting wind speed and correcting wind speed forecast obtained by numerical weather forecast mainly includes mode output statistics, Kalman filtering, BP (Back Propagation) neural network, adaptive partial least square method and the like. Among them, the most widely used method is the MOS (Model output statistics) method. Wind speed data obtained by predicting or correcting the prediction through the methods can be input into ensemble prediction, and training data of the model are enriched, so that the model is more accurate.

When the training of the wind speed prediction model is completed, the model can be applied in practice. An embodiment of the second aspect of the present application provides a wind speed prediction method, as shown in fig. 5, including the following steps:

s201: measured wind speed data of the entire electric field in an upstream area of the target electric field is acquired, and then step S203 is performed.

S202: a wind speed prediction model is obtained, which is determined by a wind speed prediction model training method as provided in the first aspect of the present application.

S203: and inputting the actually measured wind speed data into a wind speed prediction model to obtain the predicted wind speed of the target electric field.

The wind speed prediction method provided by the application can be used for predicting the wind speed of the downstream electric field more accurately by monitoring the obtained wind speed of the upstream electric field of the target electric field and combining the trained wind speed prediction model provided by the application and fully utilizing the incidence relation between the upstream electric field and the wind speed in the downstream electric field, and the scale range limitation in numerical weather forecast is not required to be considered.

In a practical implementation manner of the foregoing embodiment of the present application, after determining the predicted wind speed of the target electric field in S203, the method further includes:

and acquiring the actual measurement wind speed of the target electric field corresponding to the predicted wind speed.

If the difference value between the actually measured wind speed and the predicted wind speed of the target electric field is larger than the preset error condition, re-acquiring sample wind speed data of all electric fields in other upstream areas of the target electric field, and training a wind speed prediction model; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

Because the wind speed changes are very rich, the upstream areas corresponding to the target electric field are different, and the corresponding prediction dates may change in different upstream areas, the timely upstream areas need to be selected. For example, when wind speed prediction is performed on the mth day of the target electric field, it is necessary to predict the wind speed of the target electric field on the m-1 th day by using all the upstream regions, and select the upstream region with the best effect as the upstream region for the target electric field wind speed prediction on the mth day. That is, it is desirable to predict the target electric field wind speed on day m, but the upstream area(s) for which the prediction has been made do not necessarily fit. But the wind speed data of the m-1 th day is predicted, the m-1 th day becomes history, the actual wind speed of the m-1 th day can be obtained through actual measurement, the result of predicting through which upstream areas in the m-1 th day is the most accurate can be judged, and the wind speed prediction model trained through the data in the most accurate upstream areas is selected.

In addition to the specific upstream area determined by the above-mentioned means, the accurate upstream area may be analyzed by means of wind direction observation software, for example, GDAS (global data analysis system) meteorological data of a certain month of at least three years in history is input into the hysplit software, a wind energy source of the month is output every year, and the upstream area with the highest specific gravity of the wind energy source is selected as the upstream area for training of the wind speed prediction model.

The determination of the accuracy of the result obtained by the wind speed prediction model, in addition to the correlation with the used training data, is also related to the selected specific wind speed prediction model, and optionally, in a specific implementation manner of the foregoing embodiment of the present application, after determining the predicted wind speed of the target electric field, the method further includes:

and acquiring the actual measurement wind speed of the target electric field corresponding to the predicted wind speed. If the difference value of the actually measured wind speed and the predicted wind speed of the target electric field is larger than the preset error condition, other wind speed prediction models are selected again, and the wind speed prediction models are trained; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

Models that can be used as wind speed predictions may be:

the linear regression model is, for example, a linear regression model, a Ridge model, or an elastonet model.

Neural network models, such as the mlpragressor model.

The prediction effect of different wind speed prediction models changes according to the change of months and seasons. For example, it is desirable to predict the target farm wind speed on day s, and a wind speed prediction model that has already been predicted is not necessarily suitable for future day s wind speed predictions. The wind speed data of the s-1 th day is predicted, the s-1 th day becomes history, the actual wind speed of the s-1 th day can be obtained through actual measurement, the result predicted by which wind speed prediction model in the s-1 th day is the most accurate can be judged, and the most accurate wind speed prediction model is selected for training.

Based on the same inventive concept, the present application corresponds to an embodiment of the aforementioned training method for the wind speed prediction model, and in a third aspect, the present application provides a wind speed prediction model training device 10, as shown in fig. 6, including a wind speed obtaining module 11, a grouping module 12 and a training module 13.

The wind speed obtaining module 11 is configured to obtain a target measured sample wind speed of a target electric field and sample wind speed data of an upstream electric field in at least one upstream area of the target electric field.

The grouping module 12 is configured to determine a sample wind speed group including a number of sample wind speed data according to a wind path time interval between the target electric field and the upstream electric field in the upstream region.

The training module 13 is configured to perform model training by performing parameter adjustment on the initial wind speed prediction model according to the target actual measurement sample wind speed and the sample wind speed group, and when a preset training condition is reached, the training is finished to obtain the wind speed prediction model.

According to the training device for the wind speed prediction model, the method is executed through the modules, wind speed data with accurate upstream and downstream relations are screened out, and full training is carried out.

Optionally, the wind speed obtaining module 11 includes a wind speed selecting unit, a calculating unit and an output unit. The wind speed selecting unit is used for acquiring a target electric field in a first period t₀Is measured at a first measured wind speed v₀(ii) a All electric fields except the target electric field are acquired at a preset time interval delta t respectively in a wind speed acquisition time period t_iSecond measured wind speed v_i，t_i＝t₀+ i · Δ t, i is a positive integer. The calculation unit is used for calculating the first measured wind speed according to the preset incidence relation

v₀With each second measured wind speed v_iThe correlation value of (2). The output unit is used for determining that the electric field except the target electric field corresponding to the correlation value is an upstream electric field and a wind path time interval corresponding to the upstream electric field if the correlation value is larger than a preset correlation threshold.

Optionally, in the wind speed obtaining module 11, the upstream area is determined by a method including the following steps: the total upstream electric field of the target electric field is determined.

Clustering all upstream electric fields of the target electric field into a plurality of upstream areas according to a preset clustering method; the upstream electric field in each upstream region is within a preset geographic location range. Then, an upstream area is determined among the plurality of upstream areas.

Optionally, the grouping module 12 determines a sample wind speed group including a plurality of sample wind speed data according to a wind path time interval between the target electric field and the upstream electric field in the upstream region, including:

a minimum wind path time interval and a maximum wind path time interval corresponding to an upstream electric field in an upstream region are acquired. And according to the preset time step, determining the upstream sample wind speed data one by one between the minimum wind path time interval and the maximum wind path time interval.

Optionally, the step of acquiring the target measured sample wind speed of the target electric field by the wind speed acquiring module 11 further includes: and taking the numerical prediction wind speed obtained by numerical prediction as the target actual measurement sample wind speed.

In a fourth aspect, the present application provides a wind speed prediction device 20, as shown in fig. 7, comprising: a data acquisition module 21, a model acquisition module 22 and a data prediction module 23.

And the data acquisition module 21 is configured to acquire measured wind speed data of all the upstream electric fields in one upstream area of the target electric field.

The model obtaining module 22 is used for obtaining a wind speed prediction model.

The data prediction module 23 is configured to input the actually measured wind speed data into the wind speed prediction model and operate the model to determine the predicted wind speed of the target electric field.

Optionally, after determining the predicted wind speed of the target electric field, the data prediction module 23 further includes:

acquiring an actually measured wind speed of a target electric field corresponding to the predicted wind speed;

if the difference value between the actually measured wind speed and the predicted wind speed of the target electric field is larger than the preset error condition, re-acquiring sample wind speed data of all electric fields in other upstream areas of the target electric field, and training a wind speed prediction model; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

Optionally, after determining the predicted wind speed of the target electric field, the data prediction module 23 further includes:

acquiring an actually measured wind speed of a target electric field corresponding to the predicted wind speed;

if the difference value of the actually measured wind speed and the predicted wind speed of the target electric field is larger than the preset error condition, other wind speed prediction models are repeated, and the wind speed prediction models are trained; the preset error condition includes a preset data value and a sub-value satisfying the preset data value.

By using the wind speed prediction device provided by the present application, a prediction result closer to the actual wind conditions can be obtained, for example, by comparing the wind speed data predicted by the wind speed prediction device provided by the present application with the wind speed data obtained by the collective prediction algorithm in the prior art, as shown in fig. 8, the result obtained by the wind speed prediction device provided by the present application is closer to the actually measured wind speed than the result obtained by the collective prediction algorithm. As shown in fig. 9, the wind speed data obtained by the wind speed prediction apparatus of the present application using the algorithm is smaller than the statistical root mean square error result per day than the collective prediction algorithm, so that the wind speed prediction apparatus of the present application is used to perform wind speed prediction, and the result is more stable and accurate.

Based on the same inventive concept, an embodiment of the present application provides an electronic device, including:

a processor, a memory, and a bus;

a bus for connecting the processor and the memory;

a memory for storing operating instructions;

and the processor is used for realizing the wind speed prediction model training method or the wind speed prediction method described in the embodiment by calling the operation instruction.

Those skilled in the art will appreciate that the electronic devices provided by the embodiments of the present application may be specially designed and manufactured for the required purposes, or may comprise known devices in general-purpose computers. These devices have stored therein computer programs that are selectively activated or reconfigured. Such a computer program may be stored in a device (e.g., computer) readable medium or in any type of medium suitable for storing electronic instructions and respectively coupled to a bus.

Compared with the prior art, the wind speed of the downstream electric field can be more accurately predicted, and the scale range limitation in numerical weather forecast is not required to be considered.

The present application provides, in an alternative embodiment, an electronic device, as shown in fig. 10, the electronic device 1000 shown in fig. 10 including: a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are electrically coupled, such as by a bus 1002.

The Processor 1001 may be a CPU (Central Processing Unit), a general-purpose Processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array) or other Programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Which may implement or perform the various illustrative logical blocks, modules, and circuits described in connection with the disclosure. The processor 1001 may also be a combination of computing functions, e.g., comprising one or more microprocessors, DSPs and microprocessors, and the like.

Bus 1002 may include a path that transfers information between the above components. The bus 1002 may be a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus 1002 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in FIG. 10, but this is not intended to represent only one bus or type of bus.

The Memory 1003 may be a ROM (Read-Only Memory) or other type of static storage device that can store static information and instructions, a RAM (random access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read-Only Memory), a CD-ROM (Compact Disc Read-Only Memory) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), a magnetic Disc storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to these.

Optionally, the electronic device 1000 may also include a transceiver 1004. The transceiver 1004 may be used for reception and transmission of signals. The transceiver 1004 may allow the electronic device 1000 to communicate wirelessly or wiredly with other devices to exchange data. It should be noted that the transceiver 1004 is not limited to one in practical application.

Optionally, the electronic device 1000 may further include an input unit 1005. The input unit 1005 may be used to receive input numeric, character, image, and/or sound information, or to generate key signal inputs related to user settings and function control of the electronic apparatus 1000. The input unit 1005 may include, but is not limited to, one or more of a touch screen, a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, a camera, a microphone, and the like.

Optionally, the electronic device 1000 may further include an output unit 1006. Output unit 1006 may be used to output or show information processed by processor 1001. The output unit 1006 may include, but is not limited to, one or more of a display device, a speaker, a vibration device, and the like.

While fig. 10 illustrates an electronic device 1000 having various means, it is to be understood that not all illustrated means are required to be implemented or provided. More or fewer devices may alternatively be implemented or provided.

Optionally, the memory 1003 is used for storing application program codes for executing the scheme of the present application, and the processor 1001 controls the execution. The processor 1001 is configured to execute the application program code stored in the memory 1003 to implement any one of the wind speed prediction model training methods or wind speed prediction methods provided in the embodiments of the present application.

Based on the same inventive concept, embodiments of the present application provide a computer-readable storage medium for storing a computer program, which, when running in an electronic device, implements the wind speed prediction model training method or the wind speed prediction method as described above.

Those of skill in the art will appreciate that the various operations, methods, steps in the processes, acts, or solutions discussed in this application can be interchanged, modified, combined, or eliminated. Further, other steps, measures, or schemes in various operations, methods, or flows that have been discussed in this application can be alternated, altered, rearranged, broken down, combined, or deleted. Further, steps, measures, schemes in the prior art having various operations, methods, procedures disclosed in the present application may also be alternated, modified, rearranged, decomposed, combined, or deleted.

The terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present application, "a plurality" means two or more unless otherwise specified.

It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and may be performed in other orders unless explicitly stated herein. Moreover, at least a portion of the steps in the flow chart of the figure may include multiple sub-steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed alternately or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

The foregoing is only a partial embodiment of the present application, and it should be noted that, for those skilled in the art, several modifications and decorations can be made without departing from the principle of the present application, and these modifications and decorations should also be regarded as the protection scope of the present application.