阅读说明：本技术 基于多源信号融合的高压断路器特性参数预测方法及系统 (High-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion ) 是由 钟声 豆龙江 王国驹 于 2021-06-25 设计创作，主要内容包括：本发明提供基于多源信号融合的高压断路器特性参数预测方法,包括下列步骤：判断所述高压断路器是否处于故障状态,若是,则采集高压断路器在故障状态下第T时刻的故障数据,所述故障数据包括振动信号、声音信号、温度信号、电流信号、操作次数；提取所述振动信号、声音信号、温度信号、电流信号、操作次数的特征向量,并将所述特征向量分成训练集以及待处理数据集,将所述训练集作为预设的深度学习模型的输入,同时将故障数据的预测值作为深度学习模型输出,对所述深度学习模型进行学习训练；将待处理数据集输入至训练结束后的深度学习模型中,从而获得待处理数据集在T+n时刻的预测值,所述n为任一正自然数。(The invention provides a high-voltage circuit breaker characteristic parameter prediction method based on multi-source signal fusion, which comprises the following steps: judging whether the high-voltage circuit breaker is in a fault state, if so, acquiring fault data of the high-voltage circuit breaker at the T-th moment in the fault state, wherein the fault data comprises vibration signals, sound signals, temperature signals, current signals and operation times; extracting the feature vectors of the vibration signals, the sound signals, the temperature signals, the current signals and the operation times, dividing the feature vectors into a training set and a data set to be processed, taking the training set as the input of a preset deep learning model, outputting the predicted value of fault data as the deep learning model, and performing learning training on the deep learning model; and inputting the data set to be processed into the deep learning model after training is finished, so as to obtain a predicted value of the data set to be processed at the time of T + n, wherein n is any positive natural number.)

1. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on multi-source signal fusion is characterized by comprising the following steps of:

judging whether the high-voltage circuit breaker is in a fault state, if so, acquiring fault data of the high-voltage circuit breaker at the T-th moment in the fault state, wherein the fault data comprises vibration signals, sound signals, temperature signals, current signals and operation times;

extracting the feature vectors of the vibration signals, the sound signals, the temperature signals, the current signals and the operation times, dividing the feature vectors into a training set and a data set to be processed, taking the training set as the input of a preset deep learning model, outputting the predicted value of fault data as the deep learning model, and performing learning training on the deep learning model;

and inputting the data set to be processed into the deep learning model after training is finished, so as to obtain a predicted value of the data set to be processed at the time of T + n, wherein n is any positive natural number.

2. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on the multi-source signal fusion as claimed in claim 1, wherein the step of judging whether the high-voltage circuit breaker is in a fault state comprises the following steps: if the Hall sensor collects a switching-on and switching-off coil current signal of the high-voltage circuit breaker, judging that the high-voltage circuit breaker is in a fault state;

acquiring vibration data of the high-voltage circuit breaker in the opening and closing process through a vibration sensor;

acquiring collision sound data of the high-voltage circuit breaker by using a sound sensor;

acquiring ambient temperature data around the high-voltage circuit breaker through a temperature and humidity sensor;

and measuring the number of times of high-voltage open-close operation through the proximity switch.

3. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on the multi-source signal fusion as claimed in claim 2, wherein the method further comprises the following steps: the method comprises the following steps of extracting vibration data and sound characteristic parameters in sound data by a short-time energy-variable method, wherein the method comprises the following specific steps:

setting time domain signals of vibration data and sound data as x (n), and carrying out windowing and framing processing on the time domain signals to obtain an ith frame signal x_i(m), the energy per frame is expressed as:

by introducing a logarithmic relation, the sharp change of energy is reduced, which is specifically expressed as:

LEN_i＝log₉(1+EN_i/10)

for the ith frame signal x_i(m) performing a center cropping process, which is specifically represented as:

wherein δ is a fixed value;

the waveform rate of change for each frame is calculated after the process of center clipping:

wherein:

finally, obtaining an expression of the sound characteristic parameter: EZR_i＝LEN_i×(BXR_i+0.1)。

4. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on the multi-source signal fusion as claimed in claim 1, wherein the method further comprises the following steps: and extracting energy characteristic parameters in the vibration data and the sound data by an energy entropy method.

5. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on the multi-source signal fusion as claimed in claim 1, wherein the method further comprises the following steps: and extracting an extreme point in the current signal by a cubic spline difference method, and taking the extreme point as a characteristic parameter of the current signal.

6. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on the multi-source signal fusion is characterized in that the deep learning model comprises an RBM (radial basis function) network, the RBM network comprises an input layer, a plurality of hidden layers, a fully-connected layer and an output layer, the number of input nodes of the input layer is at least 5, the number of output nodes of the output layer is at least 5, and the RBM network is trained layer by adopting a supervised learning method.

7. The method for predicting the characteristic parameters of the high-voltage circuit breaker based on the multi-source signal fusion is characterized by further comprising the steps of obtaining a predicted value of a data set to be processed at the moment T, comparing the predicted value of the data set to be processed at the moment T + n with an actual parameter value collected at the moment T + n, and if an error between the predicted value and the actual parameter value meets a threshold value, indicating that the RBM learning training is finished, otherwise, readjusting the parameters of the RBM to perform the learning training again.

8. Multisource signal fusion's high voltage circuit breaker characteristic parameter prediction system, its characterized in that includes data acquisition device, data processing system, data acquisition device includes sound sensor, vibration sensor, hall sensor, temperature and humidity sensor, proximity switch and wireless communication unit, wireless communication unit with data processing system signal links to each other, data processing system includes:

the data receiving module is used for receiving the high-voltage circuit breaker parameters sent by the wireless communication unit;

the characteristic extraction module is used for extracting a characteristic vector of the high-voltage circuit breaker data;

the deep learning module is preset with an RBM network, the RBM network comprises an input layer, a plurality of hidden layers, a full-connection layer and an output layer, the number of input nodes of the input layer is at least 5, and the number of output nodes of the output layer is at least 5;

and the comparison module is used for comparing and judging the prediction result output by the deep learning module with the actual parameter value of the high-voltage circuit breaker.

Technical Field

The invention relates to the technical field of high-voltage circuit breaker state monitoring, in particular to a high-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion.

Background

The high-voltage circuit breaker plays a very important role in protecting and controlling the power system by opening, closing and bearing normal or abnormal current of an operating line, and the operating state of the high-voltage circuit breaker has important significance on the stability and reliability of the power system. According to the statistical data of the occurrence condition of the breaker faults, 63.2 percent of the breaker faults in China are caused by operating mechanisms.

With the rapid development of ultrahigh voltage lines in China, the requirement on the stability of characteristic parameters of a high-voltage circuit breaker is more and more strict, and in order to effectively solve the transient impact problems such as inrush current, overvoltage and the like generated by the switching-on and switching-off operation of a switch and improve the service life of power transmission and transformation equipment and the stability of a power system, the high-voltage circuit breaker must keep good mechanical characteristics. The mechanical service life influencing factors of the electric power equipment are mainly related to the components of the electric power mechanical mechanism, and mainly comprise aging abrasion of all elements in the mechanism, control voltage, system pressure, structural process parameters and the like, the change of the parameters can influence the opening and closing speed and time of the high-voltage circuit breaker, and as the operation times increase increasingly serious, the problem that how to effectively detect the characteristic parameters of the high-voltage circuit breaker becomes the primary concern of operation and maintenance personnel is solved.

At present, the traditional detection method can only measure the high-voltage switch dynamic characteristic by using a high-voltage switch dynamic characteristic tester during power failure maintenance, and cannot carry out online test.

Disclosure of Invention

The invention aims to provide a high-voltage circuit breaker characteristic parameter prediction method based on multi-source signal fusion so as to solve the problems in the background technology.

The invention is realized by the following technical scheme: the invention discloses a high-voltage circuit breaker characteristic parameter prediction method based on multi-source signal fusion, which comprises the following steps:

judging whether the high-voltage circuit breaker is in a fault state, if so, acquiring fault data of the high-voltage circuit breaker at the T-th moment in the fault state, wherein the fault data comprises vibration signals, sound signals, temperature signals, current signals and operation times;

extracting the feature vectors of the vibration signals, the sound signals, the temperature signals, the current signals and the operation times, dividing the feature vectors into a training set and a data set to be processed, taking the training set as the input of a preset deep learning model, outputting the predicted value of fault data as the deep learning model, and performing learning training on the deep learning model;

and inputting the data set to be processed into the deep learning model after training is finished, so as to obtain a predicted value of the data set to be processed at the time of T + n, wherein n is any positive natural number.

Optionally, the determining whether the high-voltage circuit breaker is in a fault state includes: if the Hall sensor collects a switching-on and switching-off coil current signal of the high-voltage circuit breaker, judging that the high-voltage circuit breaker is in a fault state;

acquiring vibration data of the high-voltage circuit breaker in the opening and closing process through a vibration sensor;

acquiring collision sound data of the high-voltage circuit breaker by using a sound sensor;

acquiring ambient temperature data around the high-voltage circuit breaker through a temperature and humidity sensor;

and measuring the number of times of high-voltage open-close operation through the proximity switch.

Optionally, the method further includes: the method comprises the following steps of extracting vibration data and sound characteristic parameters in sound data by a short-time energy-variable method, wherein the method comprises the following specific steps:

setting time domain signals of vibration data and sound data as x (n), and carrying out windowing and framing processing on the time domain signals to obtain an ith frame signal x_i() The energy per frame is expressed as:

by introducing a logarithmic relation, the sharp change of energy is reduced, which is specifically expressed as:

LEN_i＝log₉(1+EN_i/10)

for the ith frame signal x_i() And performing center cropping treatment, which is specifically expressed as:

wherein δ is a fixed value;

the waveform rate of change for each frame is calculated after the process of center clipping:

wherein:

finally, obtaining an expression of the sound characteristic parameter: EZR_i＝LEN_i×(BXR_i+0.1)。

Optionally, the method further includes: and extracting energy characteristic parameters in the vibration data and the sound data by an energy entropy method.

Optionally, the method further includes: and extracting an extreme point in the current signal by a cubic spline difference method, and taking the extreme point as a characteristic parameter of the current signal.

Optionally, the deep learning model includes an RBM network, the RBM network includes an input layer, a plurality of hidden layers, a full connection layer, and an output layer, the input nodes of the input layer are at least 5, the output nodes of the output layer are at least 5, and the RBM network is trained layer by using a supervised learning method.

Optionally, the method further includes obtaining a predicted value of the to-be-processed data set at time T, comparing the predicted value of the to-be-processed data set at time T + n with an actual parameter value acquired at time T + n, if an error between the predicted value and the actual parameter value meets a threshold, indicating that the RBM network learning training is completed, otherwise, readjusting parameters of the RBM network to perform the learning training again.

The invention discloses a multisource signal fused high-voltage circuit breaker characteristic parameter prediction system, which comprises a data acquisition device and a data processing system, wherein the data acquisition device comprises a sound sensor, a vibration sensor, a Hall sensor, a temperature and humidity sensor, a proximity switch and a wireless communication unit, the wireless communication unit is in signal connection with the data processing system, and the data processing system comprises:

the data receiving module is used for receiving the high-voltage circuit breaker parameters sent by the wireless communication unit;

the characteristic extraction module is used for extracting a characteristic vector of the high-voltage circuit breaker data;

the deep learning module is preset with an RBM network, the RBM network comprises an input layer, a plurality of hidden layers, a full-connection layer and an output layer, the number of input nodes of the input layer is at least 5, and the number of output nodes of the output layer is at least 5;

and the comparison module is used for comparing and judging the prediction result output by the deep learning module with the actual parameter value of the high-voltage circuit breaker.

Compared with the prior art, the invention has the following beneficial effects:

according to the high-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion, data acquisition is achieved by detecting current signals of switching-on and switching-off coils, when the current signals are detected, the acquisition system is automatically triggered to complete data acquisition of all signals and transmit the data to the intelligent diagnosis terminal in a wireless mode, and the intelligent diagnosis terminal completes prediction of high-voltage circuit breaker characteristic parameters through data analysis.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only preferred embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without inventive exercise.

FIG. 1 is a flow chart of a high-voltage circuit breaker characteristic parameter prediction method based on multi-source signal fusion provided by the invention;

fig. 2 is a schematic connection diagram of a data acquisition device according to an embodiment of the present invention;

fig. 3 is a schematic connection diagram of a data processing system according to an embodiment of the present invention.

In the figure, a sound sensor 1, a vibration sensor 2, a Hall sensor 3, a temperature and humidity sensor 4, a proximity switch 5, a wireless communication unit 6, a data processing system 7, a data receiving module 101, a feature extraction module 102, a deep learning module 103 and a comparison module 104 are arranged.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. It is to be understood that the described embodiments are merely a subset of embodiments of the invention and not all embodiments of the invention, with the understanding that the invention is not limited to the example embodiments described herein. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the invention described herein without inventive step, shall fall within the scope of protection of the invention.

In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present invention. It will be apparent, however, to one skilled in the art, that the present invention may be practiced without one or more of these specific details. In other instances, well-known features have not been described in order to avoid obscuring the invention.

It is to be understood that the present invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the singular forms "a", "an" 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. As used herein, the term "and/or" includes any and all combinations of the associated listed items.

In order to provide a thorough understanding of the present invention, a detailed structure will be set forth in the following description in order to explain the present invention. Alternative embodiments of the invention are described in detail below, however, the invention may be practiced in other embodiments that depart from these specific details.

Referring to fig. 1, the first aspect of the invention discloses a high-voltage circuit breaker characteristic parameter prediction method based on multi-source signal fusion, which comprises the following steps:

step S1, judging whether the high-voltage circuit breaker is in a fault state, if so, collecting fault data of the high-voltage circuit breaker at the T-th moment in the fault state, wherein the fault data comprise vibration signals, sound signals, temperature signals, current signals and operation times;

step S2, extracting the feature vectors of the vibration signals, the sound signals, the temperature signals, the current signals and the operation times, dividing the feature vectors into a training set and a data set to be processed, taking the training set as the input of a preset deep learning model, simultaneously taking the predicted value of fault data as the output of the deep learning model, and performing learning training on the deep learning model;

and step S3, inputting the data set to be processed into the deep learning model after training is finished, so as to obtain a predicted value of the data set to be processed at the time of T + n, wherein n is any positive natural number.

As an example, in step S1, fault data of the high-voltage circuit breaker at time T is collected by the sound sensor 1, the vibration sensor 2, the hall sensor 3, the temperature and humidity sensor 4, and the proximity switch 5.

As an alternative embodiment of step S1, the determining whether the high voltage circuit breaker is in the fault state includes: if the Hall sensor 3 acquires a switching-on/off coil current signal of the high-voltage circuit breaker, judging that the high-voltage circuit breaker is in a fault state;

vibration data in the opening and closing processes of the high-voltage circuit breaker are collected through a vibration sensor 2;

acquiring collision sound data of the high-voltage circuit breaker by using a sound sensor 1;

acquiring ambient temperature data around the high-voltage circuit breaker through a temperature and humidity sensor 4;

the number of high-voltage open-close operations is measured by the proximity switch 5.

The collected fault data is converted into common standard signals by a signal conditioner, and then the signals received by the signal conditioner are converted into digital signals output one by an analog-to-digital converter, but the digital signals do not have any meaning at the moment and only represent the relative quantity of one by one. These relative quantities, i.e. the feature parameters, are then subjected to the construction of a feature vector by the data processing system 7.

In an alternative embodiment of step S2, it extracts the sound characteristic parameters in the vibration data and the sound data by a short-time variable method, and its specific steps include:

setting time domain signals of vibration data and sound data as x (n), and carrying out windowing and framing processing on the time domain signals to obtain an ith frame signal x_i() The energy per frame is expressed as:

by introducing a logarithmic relation, the sharp change of energy is reduced, which is specifically expressed as:

LEN_i＝log₉(1+EN_i/10)

for the ith frame signal x_i() And performing center cropping treatment, which is specifically expressed as:

wherein δ is a fixed value;

the waveform rate of change for each frame is calculated after the process of center clipping:

wherein:

finally, obtaining an expression of the sound characteristic parameter: EZR_i＝LEN_i×(BXR_i+0.1)。

In an alternative embodiment of step S2, the energy characteristic parameter in the vibration data and the sound data is extracted by an energy entropy method, and the extraction of the energy characteristic parameter by the energy entropy method is a conventional method in the art, and the embodiment is not specifically described here.

In an alternative embodiment of step S2, the extreme point in the current signal is extracted by a cubic spline difference method, and the extreme point is used as the characteristic parameter of the current signal, and the cubic spline difference method is a conventional method in the art, and the embodiment is not specifically described here.

In an optional implementation manner of step S2, since the data collected by the temperature and humidity sensor 4 and the proximity switch 5 are numerical signals, the data collected by the temperature and humidity sensor 4 and the proximity switch 5 can be used as corresponding characteristic parameters after being normalized.

In an alternative embodiment of step S2, the deep learning model includes a RBM (restricted boltzmann machine) network, the restricted boltzmann machine is a generative random neural network, the network is composed of visible units (corresponding to visible variables, i.e. data samples) and hidden units (corresponding to hidden variables), and the visible variables and the hidden variables are binary variables, i.e. their states take {0,1 }. Only between the visible units and the hidden units, edges exist in the whole network, and no edge connection exists between the visible units and between the hidden units.

The RBM network disclosed in this embodiment includes an input layer, a plurality of hidden layers, a fully connected layer, and an output layer, where the number of input nodes of the input layer is at least 5, and the number of output nodes of the output layer is at least 5, where the input of the input layer is a sound characteristic parameter, an energy characteristic parameter, a current characteristic parameter, a temperature characteristic parameter, and an operation number characteristic parameter, and the output of the output layer is a predicted value of sound, a predicted value of energy, a predicted value of current, a predicted value of temperature, and a predicted value of operation number.

Optionally, after the sound characteristic parameter, the energy characteristic parameter, the current characteristic parameter, the temperature characteristic parameter, and the operation frequency characteristic parameter are input into the RBM network, the embodiment adopts a supervised learning method to train the RBM network layer by layer, and the specific steps include:

setting the energy of the joint configuration of the visible variable v and the hidden variable h of the RBM network as follows:

wherein, the parameter { W, a, b } of RBM at the time of theta, W is the weight of the edge between the visible unit and the hidden unit, and b and a are the bias of the visible unit and the hidden unit respectively;

after the energy after the joint configuration of v and h, the joint probability of v and h can be obtained:

where Z (θ) is a normalization factor, the above equation can be further written as:

maximizing the likelihood function P (v) of the observed data, which can be obtained from the edge distribution of P (v, h) to h in the above formula:

the parameters of RBM are derived by maximizing p (v), which is equivalent to maximizing log ((v) ═ θ)):

optionally, the step S3 further includes: and comparing the predicted value of the data set to be processed at the moment of T + n with the actual parameter value acquired at the moment of T + n, if the error between the predicted value and the actual parameter value meets a threshold value, indicating that the RBM network learning training is finished, and realizing the parameter prediction of the high-voltage circuit breaker by using the RBM network after the training is finished, otherwise, readjusting the parameters of the RBM network for learning training again.

Referring to fig. 2-3, a second aspect of the present invention discloses a multi-source signal fused high-voltage circuit breaker characteristic parameter prediction system, which includes a data acquisition device and a data processing system 7, wherein the data acquisition device includes a sound sensor 1, a vibration sensor 2, a hall sensor 3, a temperature and humidity sensor 4, a proximity switch 5 and a wireless communication unit 6, the wireless communication unit 6 is in signal connection with the data processing system 7, and the data processing system 7 includes:

the data receiving module 101 is used for receiving the high-voltage circuit breaker parameters sent by the wireless communication unit 6;

a feature extraction module 102, configured to extract a feature vector of the high-voltage breaker data;

the deep learning module 103 is preset with an RBM network, the RBM network comprises an input layer, a plurality of hidden layers, a full-connection layer and an output layer, the number of input nodes of the input layer is at least 5, and the number of output nodes of the output layer is at least 5;

and the comparison module 104 is used for comparing and judging the prediction result output by the deep learning module 103 with the actual parameter value of the high-voltage circuit breaker.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus embodiment, since it is substantially similar to the method embodiment, it is relatively simple to describe, and reference may be made to some descriptions of the method embodiment for relevant points. The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.