阅读说明：本技术 一种用于间接空冷系统的智能水温控制方法 (Intelligent water temperature control method for indirect air cooling system ) 是由 李杨 于 2021-07-14 设计创作，主要内容包括：本发明公开了一种用于间接空冷系统的智能水温控制方法,涉及自动化控制技术领域,该方法包括：采集电厂DSC数据和散热器系统的温度数据并进行规整化处理；根据冬季工况设置百叶窗的最大开度限制以及每个扇区的目标冷水母管温度；预测下一周期每个执行器的开度变化量；确定是否需要更新执行器的开度,当需要更新时,记录预测开度与实际开度的差值,通过最大开度限制排除异常预测开度；分别计算下一周期每个扇区需要更新开度的执行器个数和控制顺序；DCS控制平台向间接空冷系统下发控制指令。该方法做到了扇区执行器开度的单独调控,还能选择出控制个数和顺序,使汽轮机在冬季做工最大化,达到降低背压以及增大能源型发电利用率的目的。(The invention discloses an intelligent water temperature control method for an indirect air cooling system, which relates to the technical field of automatic control and comprises the following steps: acquiring DSC data of a power plant and temperature data of a radiator system and carrying out regularization treatment; setting the maximum opening limit of the shutter and the target cold water main pipe temperature of each sector according to the working condition in winter; predicting the opening variation of each actuator in the next period; determining whether the opening degree of the actuator needs to be updated, recording the difference value between the predicted opening degree and the actual opening degree when the updating is needed, and eliminating the abnormal predicted opening degree through the maximum opening degree limitation; respectively calculating the number of actuators and the control sequence of the opening to be updated in each sector in the next period; and the DCS control platform issues a control instruction to the indirect air cooling system. The method realizes independent control of the opening of the sector actuator, and can select the control number and sequence to maximize the work of the steam turbine in winter, thereby achieving the purposes of reducing the back pressure and increasing the energy type power generation utilization rate.)

1. An intelligent water temperature control method for an indirect air cooling system is characterized in that the indirect air cooling system comprises a DCS control platform, a steam turbine, a condenser, a circulating water pump, a valve, an air cooling tower and a radiator system, wherein exhaust steam of the steam turbine enters the condenser, the condenser is connected with the air cooling tower through a circulating water inlet pipeline through the circulating water pump, and the air cooling tower is connected with the condenser through a circulating water outlet pipeline to form a circulating loop; the radiator system comprises a shutter and a cooling triangle which are arranged on the periphery of the air cooling tower, and an actuator for controlling the opening degree of the shutter, the DCS control platform is connected with the actuator, the shutter is installed on the air inlet side of the cooling triangle, each two cooling triangles share one shutter, and the cooling triangle is divided into a plurality of sectors;

the method comprises the following steps:

acquiring power plant DSC data and temperature data of a radiator system and carrying out regularization treatment, wherein the power plant DSC data comprises the actual cold water main pipe temperature of each sector;

setting the maximum opening limit of the shutter and the target cold water main pipe temperature of each sector according to the working condition in winter;

predicting the opening variation of each actuator in the next period according to the power plant DSC data, the temperature data of the radiator system and the target cold water main pipe temperature of each sector;

acquiring the temperature change trend of a cold water main pipe of each sector in the latest continuous set period, the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature and the temperature distribution, and determining whether the opening of the actuator needs to be updated;

when the opening degree needs to be updated, recording the difference value between the predicted opening degree and the actual opening degree by taking the maximum opening degree limit as an override condition, and eliminating abnormal predicted opening degree through the override condition, wherein the abnormal predicted opening degree comprises opening degrees with too small opening degree and too large opening degree jitter;

respectively calculating the number of actuators and a control sequence of the opening of each sector to be updated in the next period according to the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature of each sector and the temperature distribution and the sequencing of the cooling triangles;

the DCS control platform issues control instructions to the indirect air cooling system, wherein the control instructions comprise the number and the control sequence of actuators with opening degrees needing to be updated in each sector and the final opening degree of each actuator;

and when the actual cold water main pipe temperature reaches an error water temperature range allowed by the target cold water main pipe temperature, the actuator keeps the opening change of the previous period, otherwise, the step of acquiring the DSC data of the power plant and the temperature data of the radiator system and performing regularization processing is executed again.

2. The intelligent water temperature control method according to claim 1, wherein one unit is matched with all sectors, a temperature sensor is arranged at a water inlet and a water outlet of a cold water main pipe of each sector and a circulating water outlet, a grating array sensor is arranged at a temperature measuring point of a cooling triangle of each sector, and the DCS control platform is further connected with the temperature sensor and the grating array sensor;

the predicting of the opening variation of each actuator in the next period according to the power plant DSC data, the temperature data of the radiator system and the target cold water main pipe temperature of each sector comprises the following steps:

judging whether each sector has a freezing risk or not according to the temperature data of the radiator system;

if so, closing the opening of the corresponding shutter; otherwise, inputting the power plant DSC data and the temperature data of the radiator system into a polynomial model to calculate the temperature of a cold water main pipe of each sector in the next period;

inputting the temperature of the cold water main pipe of each sector in the next period, power plant DSC data and temperature data of a radiator system into an XGboost model to calculate the opening variation of each actuator in each sector in the next period;

the power plant DSC data also comprises the water inlet temperature of a cold water main pipe, the power generation load, the backpressure, the environmental temperature, the wind speed, the temperature of the circulating water out of the tower, the opening information of the upper period, the number of sectors used by a unit, the number of model switching sectors and the opening information of other actuators in the same sector; the temperature data of the radiator system comprises a sector temperature average (the temperature average of all cooling triangles in the sector), a cooling triangle temperature average, a sector temperature standard deviation and a cooling triangle temperature standard deviation.

3. The intelligent water temperature control method according to claim 1, wherein the calculating of the number of actuators and the control sequence of the opening degree of each sector in the next period, which need to be updated, according to the difference between the actual cold water main pipe temperature and the target cold water main pipe temperature of each sector and the temperature distribution and sequencing of the cooling triangles respectively comprises:

collecting client data and dividing the client data into a sample set and a test set, wherein the client data comprises a standard deviation of temperatures of a cooling triangle and a sector and a difference value between an actual cold water main pipe temperature and a target cold water main pipe temperature;

taking the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature as a water temperature difference value, and taking the temperature standard difference value between the cooling triangle and the sector as a uniformity difference value;

under the same coordinate system, fitting two first safety curves according to the sample set by taking the absolute value of the water temperature difference as an x axis and the uniformity difference as a y axis; fitting a second safety curve according to the sample set by taking the uniformity difference value as an x axis and the water temperature difference value as a y axis; taking the water temperature difference as an x axis, and taking x as 0.5 as a third safety straight line;

the first safety curve, the second safety curve and the third safety straight line are connected, and an intersection area is solved to serve as a safety area;

determining the number of the same sector in the safety area in the test set as a safety number, and subtracting the safety number from the total number of actuators needing to update the opening in the sector to obtain a derived value;

sequencing actuators needing opening updating according to the fact that the cooling triangles are high in mean temperature and small in variance, and sequentially opening the opening of the actuators; and otherwise, sequencing the actuators needing to update the opening according to the low average temperature and the large variance of the cooling triangle, and closing the openings of the actuators in sequence.

4. An intelligent water temperature control method according to claim 2, wherein the polynomial model has the expression:

wherein, a₀-a_mAssigning a coefficient, x, to the weight of each parameter_i(i-1, 2, …, n) represents the power plant DSC data and the temperature data of the radiator system, respectively, n represents the number of parameters, epsilon_iRepresenting the bias constant.

5. An intelligent water temperature control method according to claim 2, wherein the XGboost model has the expression:

wherein Obj represents the predicted opening variation, G represents the first derivative of the square error of the predicted value and the actual value of the model, H is the taylor formula expansion term, T represents the tree node of the model, γ represents the minimum limit value for learning the splitting of each tree, and λ is the blocking coefficient for limiting the derivative error interval step.

6. The intelligent water temperature control method according to claim 3, wherein after the step of subtracting a safety number from a total number of actuators requiring opening updating in the sector, the method further comprises:

under the same coordinate system, a fourth curve which takes the water temperature difference as an x axis and the number of actuators as a y axis is fitted according to artificial experience, and the expression is as follows: 5.678 log eT_errorL +3.8868, where T_errorRepresenting the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature;

and substituting the test set data into the fourth curve to obtain the number of actuators with the opening needing to be updated as a reference value, comparing the reference value with a derived value, and taking a value with a smaller value as the final number of actuators with the opening needing to be updated.

7. An intelligent water temperature control method according to claim 3, wherein the expression of the first safety curve is: x is-0.22 (T)_{sq_std}-T_{each_sq_std})²+1.1*|T_{sq_std}-T_{each_sq_std}|+1.428

The expression of the second safety curve is: y 1.11 (T)_error)²-|T_error|-4.452

Wherein, T_{sq_std}Denotes the temperature standard deviation, T, of the sector_{each_sq_std}Denotes the temperature standard deviation, T, of the cooling triangle_errorThe difference value of the actual cold water main pipe temperature and the target cold water main pipe temperature is represented.

8. The intelligent water temperature control method according to any one of claims 1 to 7, wherein the indirect air cooling system further comprises a local edge server, the edge server stores a model for predicting the opening variation of each actuator in the next period, and the edge server is in communication connection with the DCS control platform;

after the step of executing the step of issuing the control instruction to the indirect air cooling system by the DCS control platform, the method further comprises the following steps:

uploading the number and control sequence of the actuators with the opening to be updated in each sector and the final opening variation of each actuator to the edge server, recording and arranging the actuators into a log form, and storing historical power plant DSC data and temperature data of a radiator system in a database;

and after a certain period of control, extracting high-quality data after test operation as a training sample for updating the model, and downloading the updated model to the local.

Technical Field

The invention relates to the technical field of automatic control, in particular to an intelligent water temperature control method for an indirect air cooling system.

Background

In a modern power plant, steam generated by a steam turbine to do work needs to be cooled by a cooling system, and a mode of taking heat away by taking air as a medium for cooling is called air cooling for short, and the cooling system is also called an air cooling system. The air cooling system has the advantages of almost no water consumption and is very commonly applied in northern areas of China. In contrast, the control of the conventional cooling system generally adopts a PID control manner, and a control strategy for eliminating an error between a control target and an actual behavior of a controlled object is generated by the error, which is the essence of the PID control technology.

In winter environment, the heat exchanger is prevented from freezing by increasing the operation temperature, but the consumed coal is too large, and under the control of PID, an actuator for controlling air flow in an air cooling system is basically switched on and off in a whole sector, so that the most economical operation back pressure is missed.

Disclosure of Invention

The invention provides an intelligent water temperature control method for an indirect air cooling system aiming at the problems and technical requirements, wherein the outlet water temperature of a cold water main pipe is controlled to be +/-0.5 ℃ of the target cold water main pipe temperature of each sector, so that the work of a steam turbine is maximized in winter as much as possible, and the back pressure is reduced.

The technical scheme of the invention is as follows:

an intelligent water temperature control method for an indirect air cooling system is disclosed, wherein the indirect air cooling system comprises a DCS control platform, a steam turbine, a condenser, a circulating water pump and valve, an air cooling tower and a radiator system, exhaust steam of the steam turbine enters the condenser, the condenser is connected with the air cooling tower through the circulating water pump by a circulating water inlet pipeline, and the air cooling tower is connected with the condenser by a circulating water outlet pipeline to form a circulating loop; the radiator system comprises shutters and cooling triangles arranged on the periphery of the air cooling tower and an actuator for controlling the opening degree of the shutters, the DCS control platform is connected with the actuator, the shutters are arranged on the air inlet side of the cooling triangles, every two cooling triangles share one shutter, and the cooling triangles are divided into a plurality of sectors;

the method comprises the following steps:

acquiring power plant DSC data and temperature data of a radiator system and carrying out regularization treatment, wherein the power plant DSC data comprises the actual cold water main pipe temperature of each sector;

setting the maximum opening limit of the shutter and the target cold water main pipe temperature of each sector according to the working condition in winter;

predicting the opening variation of each actuator in the next period according to the DSC data of the power plant, the temperature data of the radiator system and the target cold water main pipe temperature of each sector;

acquiring the temperature change trend of a cold water main pipe of each sector in the latest continuous set period, the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature and the temperature distribution, and determining whether the opening of the actuator needs to be updated;

when the opening degree needs to be updated, taking the maximum opening degree limit as an override condition, recording the difference value between the predicted opening degree and the actual opening degree, and eliminating the abnormal predicted opening degree through the override condition, wherein the abnormal predicted opening degree comprises the opening degree with too small opening degree and too large opening degree jitter;

respectively calculating the number of actuators and a control sequence of the opening of each sector to be updated in the next period according to the difference value between the actual cold water main pipe temperature of each sector and the target cold water main pipe temperature and the temperature distribution and the sequencing of the cooling triangles;

the DCS control platform issues control instructions to the indirect air cooling system, wherein the control instructions comprise the number and the control sequence of actuators of which the opening is required to be updated in each sector and the final opening of each actuator;

and when the actual cold water main pipe temperature reaches the error water temperature range allowed by the target cold water main pipe temperature, the actuator keeps the opening change of the previous period, otherwise, the step of acquiring the DSC data of the power plant and the temperature data of the radiator system and performing regularization processing is executed again.

The technical scheme is that all sectors are matched with one unit, a water inlet and a water outlet of a cold water main pipe of each sector and a circulating water outlet are provided with temperature sensors, a temperature measuring point of a cooling triangle of each sector is provided with a grating array sensor, and a DCS control platform is also connected with the temperature sensors and the grating array sensors;

predicting the opening variation of each actuator in the next period according to the DSC data of the power plant, the temperature data of the radiator system and the target cold water main pipe temperature of each sector, and comprising the following steps:

judging whether each sector has a freezing risk or not according to the temperature data of the radiator system;

if so, closing the opening of the corresponding shutter; otherwise, inputting the DSC data of the power plant and the temperature data of the radiator system into a polynomial model to calculate the temperature of a cold water main pipe of each sector in the next period;

inputting the temperature of a cold water main pipe of each sector in the next period, power plant DSC data and temperature data of a radiator system into an XGboost model to calculate the opening variation of each actuator in each sector in the next period;

the DSC data of the power plant also comprises the water inlet temperature of a cold water main pipe, the power generation load, the backpressure, the environmental temperature, the wind speed, the temperature of the circulating water out of the tower, the opening information of the upper period, the number of sectors used by a unit, the number of model switching sectors and the opening information of other actuators in the same sector; the temperature data of the radiator system comprises a sector temperature average (the temperature average of all cooling triangles in the sector), a cooling triangle temperature average, a sector temperature standard deviation and a cooling triangle temperature standard deviation.

The further technical scheme is that the number of actuators of each sector in the next period, the opening of which needs to be updated, and the control sequence are respectively calculated according to the difference value between the actual cold water main pipe temperature of each sector and the target cold water main pipe temperature and the temperature distribution and the sequencing of a cooling triangle, and the method comprises the following steps:

collecting client data and dividing the client data into a sample set and a test set, wherein the client data comprises a standard deviation of temperatures of a cooling triangle and a sector and a difference value of an actual cold water main pipe temperature and a target cold water main pipe temperature;

taking the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature as a water temperature difference value, and taking the temperature standard difference value between the cooling triangle and the sector as a uniformity difference value;

under the same coordinate system, fitting two first safety curves according to a sample set by taking the absolute value of the water temperature difference as an x axis and the uniformity difference as a y axis; fitting a second safety curve according to the sample set by taking the uniformity difference value as an x axis and the water temperature difference value as a y axis; taking the water temperature difference as an x axis, and taking x as 0.5 as a third safety straight line;

the first safety curve, the second safety curve and the third safety straight line are connected, and the intersection area is solved to serve as a safety area;

determining the number of the same sector in the test set falling into a safety area as a safety number, and subtracting the safety number from the total number of actuators needing to update the opening in the sector to obtain a derived value;

sequencing actuators needing opening updating according to the fact that the cooling triangles are high in mean temperature and small in variance, and sequentially opening the opening of the actuators; and otherwise, sequencing the actuators needing to update the opening according to the low average temperature and the large variance of the cooling triangle, and closing the openings of the actuators in sequence.

The further technical scheme is that the expression of the polynomial model is as follows:

wherein, a₀-a_mAssigning a coefficient, x, to the weight of each parameter_i(i-1, 2, …, n) represents power plant DSC data and radiator system temperature data, respectively, n represents the number of entries, ε_iRepresenting the bias constant.

The further technical scheme is that the XGboost model has the expression as follows:

wherein Obj represents the predicted opening variation, G represents the first derivative of the square error of the predicted value and the actual value of the model, H is the taylor formula expansion term, T represents the tree node of the model, γ represents the minimum limit value for learning the splitting of each tree, and λ is the blocking coefficient for limiting the derivative error interval step.

The method further comprises the following step of subtracting the safety number from the total number of the actuators needing to update the opening degree in the sector, wherein the method comprises the following steps:

under the same coordinate system, a fourth curve which takes the water temperature difference as an x axis and the number of actuators as a y axis is fitted according to artificial experience, and the expression is as follows: 5.678 log_eT_errorL +3.8868, where T_errorRepresenting the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature;

and substituting the test set data into the fourth curve to obtain the number of actuators with the opening needing to be updated as a reference value, comparing the reference value with the derivation value, and taking the value with a smaller value as the final number of actuators with the opening needing to be updated.

The further technical scheme is that the expression of the first safety curve is as follows: x is-0.22 (T)_{sq_std}-T_{each_sq_std})²+1.1*|T_{sq_std}-T_{each_sq_std}|+1.428

The expression of the second safety curve is: y 1.11 (T)_error)²-|T_error|-4.452

Wherein, T_{sq_std}Denotes the temperature standard deviation, T, of the sector_{each_sq_std}Denotes the temperature standard deviation, T, of the cooling triangle_errorThe difference value of the actual cold water main pipe temperature and the target cold water main pipe temperature is represented.

The indirect air cooling system further comprises an edge server arranged locally, wherein a model for predicting the opening variation of each actuator in the next period is stored in the edge server, and the edge server is in communication connection with the DCS control platform;

after the step of executing the step of issuing the control instruction to the indirect air cooling system by the DCS control platform, the method further comprises the following steps:

uploading the number and control sequence of actuators of which the opening needs to be updated in each sector and the final opening variation of each actuator to an edge server, recording and arranging the actuators into a log form, and storing historical power plant DSC data and temperature data of a radiator system in a database;

and after a certain period of control, extracting high-quality data after test operation as a training sample for updating the model, and downloading the updated model to the local.

The beneficial technical effects of the invention are as follows:

aiming at the working conditions in winter, the anti-freezing detection and protection are carried out on the cooling triangle before the opening degree is predicted, the independent regulation and control of the opening degree of a sector actuator are carried out by combining a polynomial model and an XGboost model which are fit by integrating learning and a plurality of units, the traditional control method that the actuators in the same sector are simultaneously opened and closed is broken, the most economic output increment value opening degree variation is successfully learned by fine tuning and stabilizing the water temperature within the error range of a target value +/-0.5 ℃, the opening degree is controlled after the water temperature is stabilized, the excessive adjustment of the opening degree caused by inertia is prevented, and after the opening degree is determined to be updated, the abnormal opening degree is eliminated by taking the maximum opening degree as an override condition, so that the water temperature is prevented from being over-controlled; fitting three safety curves and a straight line through data provided by a client to determine a safety region, and sequentially controlling actuators needing opening updating in the same sector according to a specified control sequence of the actuators; and training the model in real time during the trial operation to obtain an optimal parameter model, so as to achieve the purposes of reducing the backpressure and increasing the energy type power generation utilization rate.

Drawings

Fig. 1 is a schematic view of an indirect air cooling system provided by the present application.

Fig. 2 is an overall flowchart of the intelligent water temperature control method provided by the present application.

Fig. 3 is a schematic flowchart of predicting the opening degree variation of the next cycle according to the present application.

Fig. 4 is a fitting graph of the number of actuators for updating the opening degree provided in the present application.

Detailed Description

The following further describes the embodiments of the present invention with reference to the drawings.

As shown in fig. 1, an indirect air cooling system includes a DCS control platform (not shown), a steam turbine 1, a condenser 2, a circulating water pump 3 and valves, an air cooling tower 4, a radiator system, and a local edge server (not shown), where exhaust steam of the steam turbine 1 enters the condenser 2, the condenser 2 is connected to the air cooling tower 4 through the circulating water pump 3 via a circulating water inlet pipeline 51, and the air cooling tower 4 is connected to the condenser 2 via a circulating water outlet pipeline 52 to form a circulation loop.

The radiator system completes the control of the temperature of circulating water and ensures reasonable temperature of the circulating water discharged from the tower, and comprises a shutter 6 and a cooling triangle which are arranged on the periphery of an air cooling tower 4 and an actuator (not shown in the figure) for controlling the opening degree of the shutter, wherein the shutter is arranged on the air inlet side of the cooling triangle, the cooling triangle comprises two vertically arranged radiating fins, every two cooling triangles share one shutter, the cooling triangle is divided into a plurality of sectors, and all the sectors are matched with one unit.

Temperature sensors are arranged at a water inlet and a water outlet of the cold water main pipe of each sector and a circulating water outlet, and a grating array sensor is arranged at a temperature measuring point of a cooling triangle of each sector; the edge server stores a model (namely a polynomial model and an XGboost model) for predicting the opening variation of each actuator in the next period.

The DCS control platform is respectively connected with the actuator, the temperature sensor, the grating array sensor and the edge server and used for acquiring DSC data of a power plant and temperature data of a radiator system.

As shown in fig. 2, an intelligent water temperature control method for an indirect air cooling system includes the following steps:

step 1: and acquiring DSC data of the power plant and temperature data of the radiator system in one period and carrying out regularization processing.

The DSC data of the power plant comprises the inlet water temperature of a cold water main pipe of each sector, the actual temperature of the cold water main pipe (namely the actual outlet water temperature of cooling water of the sector), the power generation load, the backpressure, the ambient temperature, the wind speed, the outlet temperature of circulating water from the tower, the opening information of the upper period, the number of sectors used by a unit, the number of model switching sectors and the opening information of other actuators in the same sector.

The temperature data of the radiator system comprises a sector temperature average (namely, the temperature average of all cooling triangles in the sector), a cooling triangle temperature average, a sector temperature standard deviation and a cooling triangle temperature standard deviation.

Alternatively, one period is generally set to 30 s.

Step 2: and (3) setting the maximum opening limit of the shutter and the target cold water main pipe temperature of each sector according to the winter working condition, and recording the actual cold water main pipe temperature in the step 1.

And step 3: and predicting the opening variation of each actuator in the next period according to the DSC data of the power plant, the temperature data of the radiator system and the target cold water main pipe temperature of each sector.

As shown in fig. 3, the method specifically includes the following steps:

step 31: and judging whether each sector has a freezing risk according to the temperature data of the radiator system, if so, reducing the opening of the corresponding shutter 6, and judging the freezing risk again, otherwise, entering the step 32.

Step 32: and inputting the DSC data of the power plant and the temperature data of the radiator system into a polynomial model to calculate the temperature of a cold water main pipe of each sector in the next period.

The expression of the polynomial model is:

wherein, a₀-a_mAssigning a coefficient, x, to the weight of each parameter_i(i-1, 2, …, n) represents power plant DSC data and radiator system temperature data, respectively, n represents the number of entries, ε_iThe bias constants are expressed, and in the model training process, the bias constants are different because the training sample sets are different.

Step 33: and inputting the temperature of the cold water main pipe of each sector in the next period, the DSC data of the power plant and the temperature data of the radiator system into an XGboost model to calculate the opening variation of each actuator in each sector in the next period.

The XGboost model has the expression:

wherein Obj represents the predicted opening variation, G represents the first derivative of the square error of the predicted value and the actual value of the model, H is the taylor formula expansion term, T represents the tree node of the model, γ represents the minimum limit value for learning the splitting of each tree, and λ is the blocking coefficient for limiting the derivative error interval step.

And 4, step 4: and acquiring the temperature change trend of the cold water main pipe of each sector in the latest continuous set period, the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature and the temperature distribution, and determining whether the opening of the actuator needs to be updated.

And when the actual temperature of the cold water main pipe does not reach the target temperature of the cold water main pipe, the cold water main pipe is in a rising or falling state, the actuator is kept still, the opening degree of the actuator is updated until the water temperature is kept stable and the heat dissipation capacity reaches a limit value, the opening degree is prevented from being excessively adjusted due to inertia, and then the step 5 of updating the opening degree of the actuator is executed according to a temperature protection logic.

Alternatively, the last consecutive set period may be set to the last three consecutive periods.

And 5: and recording the difference between the predicted opening (obtained by predicting the opening variation) and the actual opening in the period by taking the maximum opening limit as an override condition, and eliminating the abnormal predicted opening by the override condition, wherein the abnormal predicted opening comprises the opening with too small opening and too large opening jitter, so as to prevent the water temperature from being over-controlled.

Step 6: respectively calculating the number of actuators and a control sequence of the opening of each sector in the next period, which need to be updated, according to the difference value between the actual cold water main pipe temperature of each sector and the target cold water main pipe temperature and the temperature distribution and the sequencing of the cooling triangles, and specifically comprising the following steps:

step 61: and collecting client data and dividing the client data into a sample set and a test set, wherein the client data comprises the standard deviation of the temperature of a cooling triangle and a sector and the difference value of the actual cold water main pipe temperature and the target cold water main pipe temperature.

Step 62: and taking the difference value between the actual cold water main pipe temperature and the target cold water main pipe temperature as a water temperature difference value, and taking the temperature standard deviation between the cooling triangle and the sector as a uniformity difference value.

And step 63: as shown in fig. 4, in the same coordinate system, two first safety curves 71 are fitted according to a sample set with the absolute value of the water temperature difference as the x-axis and the uniformity difference as the y-axis, where the expression is:

x＝-0.22*(T_{sq_std}-T_{each_sq_std})²+1.1*|T_{sq_std}-T_{each_sq_std}|+1.428

and fitting a second safety curve 72 according to the sample set by taking the uniformity difference value as an x axis and the water temperature difference value as a y axis, wherein the expression is as follows: y 1.11 (T)_error)²-|T_error|-4.452

Wherein, T_{sq_std}Denotes the temperature standard deviation, T, of the sector_{each_sq_std}Denotes the temperature standard deviation, T, of the cooling triangle_errorThe difference value of the actual cold water main pipe temperature and the target cold water main pipe temperature is represented.

The water temperature difference is taken as the x-axis, and x is taken to be 0.5 as the third safety straight line 73.

Step 64: the first safety curve 71, the second safety curve 72 and the third safety line 73 are connected, and the intersection area is solved as a safety area (i.e., a shaded portion in the figure).

Step 65: and determining the number of the same sector in the test set falling into the safety area as a safety number, and subtracting the safety number from the total number of the actuators needing to update the opening in the sector to obtain a derived value.

And step 66: and fitting a fourth curve 74 taking the water temperature difference as an x axis and the number of actuators as a y axis according to artificial experience, wherein the expression is as follows: 5.678 log_eT_errorL +3.8868, where T_errorThe difference value of the actual cold water main pipe temperature and the target cold water main pipe temperature is represented.

Step 67: and substituting the test set data into the fourth curve 74 to obtain the number of actuators with the opening needing to be updated as a reference value, comparing the reference value with the derivation value, and taking the value with the smaller value as the final number of actuators with the opening needing to be updated.

Step 68: sequencing actuators needing opening updating according to the fact that the cooling triangles are high in mean temperature and small in variance, and sequentially opening the opening of the actuators; and otherwise, sequencing the actuators needing to update the opening according to the low average temperature and the large variance of the cooling triangle, and closing the openings of the actuators in sequence.

And 7: and the DCS control platform issues a control instruction to the indirect air cooling system, wherein the control instruction comprises the number and the control sequence of the actuators with the opening needing to be updated in each sector and the final opening of each actuator (namely the sum of the predicted opening variation and the opening of the current period).

And 8: uploading the number and control sequence of actuators of which the opening needs to be updated in each sector and the final opening variation of each actuator to an edge server, recording and arranging the actuators into a log form, and storing historical power plant DSC data and temperature data of a radiator system in a database.

And after a certain period of control, extracting high-quality data after test operation as a training sample for updating the model, and downloading the updated model to the local.

When the actual cold water main pipe temperature reaches the error water temperature range allowed by the target cold water main pipe temperature (namely the target cold water main pipe temperature is +/-0.5 ℃), the opening variation is not predicted any more, the actuator still keeps the opening variation of the previous period, and otherwise, the step 1 is executed again.

The method realizes independent regulation and control of the opening of the sector actuator, breaks through the traditional control method that the same sector actuator is opened and closed simultaneously, achieves fine adjustment and stabilization of the water temperature within the error range of a target value +/-0.5 ℃, successfully learns the most economical output increment value opening variation, trains the model in real time during the trial run to obtain the optimal parameter model, and achieves the purposes of reducing the backpressure and increasing the energy type power generation utilization rate.

What has been described above is only a preferred embodiment of the present application, and the present invention is not limited to the above embodiment. It is to be understood that other modifications and variations directly derivable or suggested by those skilled in the art without departing from the spirit and concept of the present invention are to be considered as included within the scope of the present invention.