CN113358157A - RST-PNN-GA-based power equipment temperature rise detection and early warning method - Google Patents

RST-PNN-GA-based power equipment temperature rise detection and early warning method Download PDF

Info

Publication number
CN113358157A
CN113358157A CN202110648478.5A CN202110648478A CN113358157A CN 113358157 A CN113358157 A CN 113358157A CN 202110648478 A CN202110648478 A CN 202110648478A CN 113358157 A CN113358157 A CN 113358157A
Authority
CN
China
Prior art keywords
power equipment
temperature rise
early warning
data
neural network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110648478.5A
Other languages
Chinese (zh)
Inventor
胡潇文
郭海龙
任伟
张斌
陈敏
李亚东
冉利利
陶冶
郑立
柯成军
王生鹏
靳夏
常雪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Lanzhou Power Supply Co Of State Grid Gansu Electric Power Co
Original Assignee
Lanzhou Power Supply Co Of State Grid Gansu Electric Power Co
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Lanzhou Power Supply Co Of State Grid Gansu Electric Power Co filed Critical Lanzhou Power Supply Co Of State Grid Gansu Electric Power Co
Priority to CN202110648478.5A priority Critical patent/CN113358157A/en
Publication of CN113358157A publication Critical patent/CN113358157A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G08—SIGNALLING
    • G08C—TRANSMISSION SYSTEMS FOR MEASURED VALUES, CONTROL OR SIMILAR SIGNALS
    • G08C17/00—Arrangements for transmitting signals characterised by the use of a wireless electrical link
    • G08C17/02—Arrangements for transmitting signals characterised by the use of a wireless electrical link using a radio link

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention provides a RST-PNN-GA-based power equipment temperature rise detection early warning method, and relates to the technical field of power equipment detection. According to the method, the temperature rise characteristic signals are analyzed by adopting a probabilistic neural network, the collected characteristic samples are simplified by utilizing a rough set, error back-propagation calculation is carried out by utilizing a genetic algorithm, and the temperature rise detection early warning method for the power equipment based on the RST-PNN-GA neural network algorithm is established. According to the invention, the temperature data of the power equipment is monitored in real time by the power equipment temperature rise online monitoring system, and the data is input into the power equipment temperature rise detection early warning model for analysis, so that a temperature rise early warning signal can be obtained.

Description

RST-PNN-GA-based power equipment temperature rise detection and early warning method
Technical Field
The invention relates to the technical field of power equipment detection, in particular to a RST-PNN-GA neural network algorithm-based power equipment temperature rise detection early warning method.
Background
In combination with the practical situation of power grid distribution in China, the connecting parts between contacts in the power equipment generate heat due to aging of materials or increase of contact resistance in long-term operation, and most of the heating parts are in the power equipment, so that the current monitoring means cannot effectively realize on-line monitoring. Meanwhile, due to the complexity of the power distribution network, measuring points are widely distributed, high voltage is provided around the measuring points, manual measurement is inconvenient, and various faults can often occur in the operation process of electrical equipment. According to statistics of relevant departments, when the electrical equipment is abnormally operated, the thermal fault is firstly shown. Thermal failures are caused by increased contact resistance due to poor connections to equipment, contacts, or oxidation, which in turn can lead to overheating, burning, or even shorting of the contacts. It can be said that thermal faults are closely related to the lifetime of the power equipment itself.
At present, an on-line monitoring system adopted by a transformer substation is mostly in a fixed threshold value alarming mode. Comparing the online monitoring value with a preset alarm threshold value, and if the real-time data is greater than the alarm threshold value, judging that the power system works abnormally; otherwise, the operation is normal. However, when the monitored value is larger than the set value, some of the failures have deteriorated. However, if the threshold is set too low, false alarm will be generated.
The neural network algorithm has good fault-tolerant capability, parallel processing capability and self-learning capability, can solve the problems of complex environmental information, unclear background knowledge and ambiguous inference rule, has high running speed, good self-adaptation performance and higher resolution, and particularly makes the weight sharing network more similar to a biological neural network, reduces the complexity of a network model and reduces the number of weights. And a neural network algorithm is adopted, so that the complex characteristic extraction and data reconstruction process in the traditional recognition algorithm is avoided.
Therefore, an online temperature rise monitoring system capable of displaying the temperature of each node of the power equipment in real time is researched and provided, the practicability and the reliability of the thermal fault early warning technology are met by combining an artificial intelligence technology, and the system has very important significance for stable operation of the conventional power system.
Disclosure of Invention
The invention aims to provide a temperature rise detection and early warning method for power equipment based on a RST-PNN-GA neural network algorithm.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a RST-PNN-GA-based power equipment temperature rise detection and early warning method comprises the following steps:
the method comprises the steps of constructing an electric power equipment temperature rise online monitoring system, then monitoring the temperature of electric power equipment in real time by using the electric power equipment temperature rise online monitoring system to obtain electric power equipment temperature data, and taking the electric power equipment temperature data obtained through real-time monitoring as an input variable;
secondly, constructing a power equipment temperature rise detection early warning model based on an RST-PNN-GA neural network algorithm, and inputting a part of power equipment temperature data obtained in the step into the constructed power equipment temperature rise detection early warning model for training;
inputting a part of temperature data of the electric power equipment obtained in the step into the trained temperature rise detection and early warning model of the electric power equipment, processing the input temperature data of the electric power equipment by the temperature rise detection and early warning model of the electric power equipment, and completing temperature rise detection and early warning of the electric power equipment.
Furthermore, the power equipment temperature rise on-line monitoring system comprises a plurality of monitoring slave machines, a plurality of groups of monitoring host machines and an upper computer monitoring center, wherein the monitoring slave machines are used for collecting the temperature of the power equipment temperature measurement nodes and uploading collected temperature data to the monitoring host machines; each group of monitoring host corresponds to a plurality of monitoring slave machines, and the monitoring host is used for summarizing, displaying and storing data of the corresponding monitoring slave machines and uploading the data to an upper computer monitoring center in a GPRS or RS485 mode; the upper computer monitoring center collects data collected by the monitoring hosts, realizes storage and display of the data and issuing of a threshold value, realizes fault early warning while carrying out data analysis, and can carry out fault early warning according to temperature data.
Preferably, the monitoring slave machine comprises a temperature and humidity sensor and a temperature collector, the temperature and humidity sensor monitors the temperature and the humidity of the working environment of the power equipment and the temperature of the monitoring node in real time, and the temperature collector receives temperature data from the temperature and humidity sensor and then uploads the temperature data to the monitoring host machine in a wireless communication mode; the monitoring host comprises a wireless module, a clock, an alarm indicator light, a touch screen and a single chip microcomputer, wherein the single chip microcomputer is electrically connected with the wireless module, the clock, the alarm indicator light and the touch screen respectively.
Preferably, the wireless module is a 433MHz wireless communication module, the temperature and humidity sensor is of the type SHT715, the single chip microcomputer is of the type STM8L151K4T6, and the touch screen is an industrial-grade resistive touch screen.
Further, the invention also provides a construction method of the power equipment temperature rise detection early warning model, which comprises the following steps:
the method comprises the steps of establishing a probabilistic neural network main model, inputting collected original fault characteristic sample data to perform training diagnosis, and forming a detection model based on the probabilistic neural network;
and secondly, optimizing the established detection model based on the probabilistic neural network by using the rough set and the genetic algorithm respectively, and establishing a RST-PNN-GA neural network-based power equipment temperature rise detection early warning model.
Further, the invention also provides a training method of the power equipment temperature rise detection early warning model, which is implemented according to the following specific steps:
inputting original fault characteristic information parameters acquired through a simulation test platform into a probabilistic neural network main body model for training;
performing attribute reduction on the original fault characteristic information parameters by using a rough set reduction theory, firstly determining condition attributes and decision attributes, forming a decision on fault type information, and finally forming a reduction attribute decision;
inputting the reduced fault characteristic parameter reduction decision table into the probabilistic neural network main body model;
fourthly, network initialization is carried out on a probabilistic neural network main body model and an improved genetic algorithm;
fifthly, performing forward feedback calculation by using a probabilistic neural network, firstly receiving the parameters input in the step three, determining the number of network neurons, and calculating the weight by using a basic algorithm of the probabilistic neural network to obtain a probability density value; the input and output of the probabilistic neural network are a 5-input 2-output model, 3 layers of neurons are preset in the middle layer, and the number of the neurons is 18;
sixthly, judging the output probability density value error and the iteration times thereof, and outputting a result according to the requirement;
if the errors judged in the step sixteenth are not in accordance with the requirements, performing error back propagation calculation by using a genetic algorithm, firstly calculating fitness, updating the population again, performing the operation of the genetic algorithm on the population again, updating new weight values, and recalculating the weight values again to the probabilistic neural network until the errors and the iteration times meet the requirements;
and outputting the diagnosis result.
Further, the invention also provides a training method of the probabilistic neural network body model, which is implemented according to the following steps:
1) normalization processing, namely initializing a network, processing input sample data and training the sample data into matrix samples as follows:
Figure BDA0003110137880000051
normalizing all sample data in the matrix X to obtain a normalization coefficient BTAnd the normalized learning sample C can be expressed as:
Figure BDA0003110137880000052
CM×N＝Bm×1[11...1]1×n·Xm×n
2) inputting the sample characteristic data processed in the first step into an input layer;
Figure BDA0003110137880000053
3) exerting the left and right of a mode layer, carrying out refinement calculation on the mode distance, and then carrying out normalization processing on sample data with processing characteristics formed by P n-dimensional vectors, wherein the matrix form is as follows:
Figure BDA0003110137880000054
the calculation of the pattern distance requires that each normalized matrix to be processed and each normalized feature sample matrix are used to calculate the euclidean distance.
Assume normalized training samples as ciI is 1,2 … m; the normalized matrix to be processed is djJ ═ 1,2.. p. The jth data to be processed (d)j) And the ith training sample (c)i) The euclidean distance of (a) may be expressed as: eij. The details are as follows:
Figure BDA0003110137880000061
4) activating gaussian function neurons of the pattern layer:
Figure BDA0003110137880000062
5) classifying the feature data samples which are not trained yet and solving the probability values of the data:
Figure BDA0003110137880000063
6) generalizing the probability value formula as follows:
Figure BDA0003110137880000064
in order to obtain the probability value of the finally output unprocessed feature signal data sample, learning and training are carried out one by one according to the aforementioned calculation steps, and meanwhile, the threshold value of the probabilistic neural network is set to search for the optimal probability value in a specified range, so that the output type corresponding to the obtained optimal probability value is the type of the feature signal data sample which is not processed, namely, the training and diagnosis process of the whole probabilistic neural network model is completed.
Further, the operation of the genetic algorithm comprises a selection operation, a crossover operation and a mutation operation.
Further, the rough reduction theory includes:
first, assume that the description of the information table S is:
S＝(U,C,D,V,F)；
wherein, U is discourse domain, C is condition attribute set, D is decision attribute set, V is Va∈C∪DVaIs the value range of the attribute, wherein is the attribute aVaU (coud) → V is an information decision function;
for equivalence relations, any subset of attributes has
Figure BDA0003110137880000071
Figure BDA0003110137880000072
A partition U/B may be formed for the discourse using equivalence relations. Wherein each small region divided is of an equivalent type and can be described as [ x ]]B＝{y∈U|(x，y)∈IND(B)}；
For each subset in any domain of discourse
Figure BDA0003110137880000073
There is an upper and lower approximation for B, which are described as respectively
Figure BDA0003110137880000074
Figure BDA0003110137880000075
In the formula, the B-based partition domain must be partitioned into the X class set of objects, which is called the upper approximation. The lower approximation refers to the set of objects that are likely to be classified into X classes in the B-based partitioned region.
Wherein collection of attributes
Figure BDA0003110137880000076
Further, the definition of the positive, negative regions (outer regions) and the boundary region with respect to the decision set D is as follows:
POSB(D)＝∪XeU/D B(X)；
NEGB(D)＝U-∪XeU/DB(X)；
Figure BDA0003110137880000077
in the formula, all the regions that are necessarily divided into one divided region based on D in the divided region of B are called positive regions, and mainly reflect the classification capability of the attribute B relative to D.
In the case of P, the number of P,
Figure BDA0003110137880000081
the dependency of Q on P can be defined as
k＝γP(Q)＝|POSP(Q)|/|U|；
For any subset of conditional attributes
Figure BDA0003110137880000082
If:
POSR(D)＝POSC(D)；
POSR(D)≠POSR-{a}(D)，a∈R；
r is a relative reduction of C.
Compared with the prior art, the invention has the beneficial technical effects that:
(1) the RST-PNN-GA neural network algorithm-based power equipment temperature rise detection and early warning method constructs a power equipment temperature rise detection and early warning model which takes a probabilistic neural network as a main body and is optimized by a rough set theory and a genetic algorithm; the method has the advantages that the neurons of the probabilistic neural network are updated again by utilizing the operations of selection, intersection, variation and the like of the genetic algorithm, so that a new optimal solution is formed for threshold discrimination, the diagnosis accuracy of the high-voltage circuit breaker fault diagnosis model is improved, and a reliable early warning result is obtained; the simplest attribute table of the fault characteristic sample is updated again by using the attribute reduction operation of the rough set, so that the time of a diagnosis process is reduced, the diagnosis rate is improved, and the diagnosis rate of the temperature rise detection early warning model of the power equipment is improved;
(2) the temperature rise online monitoring system of the power equipment is constructed to monitor the temperature data of the power equipment in real time, and the data is input into a temperature rise detection early warning model of the power equipment for analysis to obtain a temperature rise early warning signal;
(3) the RST-PNN-GA neural network algorithm-based power equipment temperature rise detection early warning method can eliminate the influence of the change of factors irrelevant to the learning task in input data on the learning performance in a data representation mode, simultaneously retains useful information in the learning task, and performs real-time reverse transmission updating on errors of the learning task while learning training; the method is applied to the temperature rise detection and early warning of the power equipment, and can judge the early warning more accurately and carry out state maintenance.
Drawings
FIG. 1 is a schematic structural diagram of an online monitoring system for temperature rise of an electric power device, which is adopted in the RST-PNN-GA-based electric power device temperature rise detection and early warning method of the present invention;
FIG. 2 is a flow chart of a RST-PNN-GA-based power equipment temperature rise detection early warning method of the invention;
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1
Referring to fig. 1-2, a method for detecting and warning temperature rise of an electric power device based on RST-PNN-GA includes the following steps:
the method comprises the steps of constructing an electric power equipment temperature rise online monitoring system, then monitoring the temperature of electric power equipment in real time by using the electric power equipment temperature rise online monitoring system to obtain electric power equipment temperature data, and taking the electric power equipment temperature data obtained through real-time monitoring as an input variable;
secondly, constructing a power equipment temperature rise detection early warning model based on an RST-PNN-GA neural network algorithm, and inputting a part of power equipment temperature data obtained in the step into the constructed power equipment temperature rise detection early warning model for training;
inputting a part of temperature data of the electric power equipment obtained in the step into the trained temperature rise detection and early warning model of the electric power equipment, processing the input temperature data of the electric power equipment by the temperature rise detection and early warning model of the electric power equipment, and completing temperature rise detection and early warning of the electric power equipment.
The power equipment temperature rise on-line monitoring system comprises a plurality of monitoring slave machines, a plurality of groups of monitoring host machines and an upper computer monitoring center, wherein the monitoring slave machines are used for collecting the temperature of the power equipment temperature measurement nodes and uploading collected temperature data to the monitoring host machines; each group of monitoring host corresponds to a plurality of monitoring slave machines, and the monitoring host is used for summarizing, displaying and storing data of the corresponding monitoring slave machines and uploading the data to an upper computer monitoring center in a GPRS or RS485 mode; the upper computer monitoring center collects data collected by each monitoring host, realizes storage and display of the data and issuing of a threshold value, realizes fault early warning while analyzing the data, and can perform fault early warning according to temperature data; the monitoring slave machine comprises a temperature and humidity sensor and a temperature collector, the temperature and humidity sensor monitors the temperature and the humidity of the working environment of the power equipment and the temperature of a monitoring node in real time, and the temperature collector receives temperature data from the temperature and humidity sensor and uploads the temperature data to the monitoring master machine in a wireless communication mode; the monitoring host comprises a wireless module, a clock, an alarm indicator light, a touch screen and a single chip microcomputer, wherein the single chip microcomputer is electrically connected with the wireless module, the clock, the alarm indicator light and the touch screen respectively.
In this embodiment, the wireless module is a 433MHz wireless communication module, the type adopted by the temperature and humidity sensor is SHT715, and the touch screen is an industrial-grade resistive touch screen.
In this embodiment, the monitoring slave unit can monitor the battery voltage in addition to measuring the temperature of the temperature measurement node of the power equipment and performing simple fault diagnosis, so that an alarm can be given when the electric quantity of the battery is about to be exhausted, and a user can be reminded to replace the battery. The requirement on low power consumption is strict in view of the special working environment of the monitoring slave. In the design process, the power consumption is strictly controlled by the model selection of the single chip microcomputer and the circuit design of each module. The CPU of the monitoring slave machine adopts an STM8L151K4T6 single chip microcomputer with low power consumption design, and the communication between the monitoring slave machine and the monitoring host machine adopts 433MHz wireless communication technology.
Considering the reason of power consumption, the monitoring slave sends data to the host at regular time, and the 433MHz wireless module of the monitoring host needs to be in a receiving mode all the time. And monitoring the stability of the received information by implanting a UCOS-III operating system in the host program, and scheduling the operation of other tasks. The environment temperature is monitored in real time through the SHT715 temperature and humidity sensor, and the SHT715 temperature and humidity sensor provides reference for temperature rise calculation while recording the working environment of the power equipment. And by comparing with the ambient temperature, the temperature rise has more reference value. The monitoring host is also provided with an industrial-grade resistive touch screen, the data such as environment temperature and humidity, temperature of each monitoring node, temperature rise, battery voltage and the like are displayed in real time, fault alarm information is inquired, an alarm threshold value can be keyed in through the resistive touch screen, and the monitoring host uploads the data to the host monitoring center for further processing in a GPRS or RS485 mode at regular time.
The upper computer monitoring center adopts VC + + to realize the storage and display of data and the issuing of threshold value, realizes the fault early warning while carrying out data analysis, and can carry out the fault early warning according to the temperature data. The upper computer monitoring center and the monitoring host can set different corresponding relations according to different actual conditions. In the embodiment, one upper computer monitoring center corresponds to one monitoring host, one monitoring host corresponds to 27 temperature measuring nodes, the monitoring host and the monitoring slave adopt 433MHz wireless communication technology, and the monitoring host and the upper computer monitoring center adopt GPRS communication mode.
The construction method of the power equipment temperature rise detection early warning model comprises the following steps:
the method comprises the steps of establishing a probabilistic neural network main model, inputting collected original fault characteristic sample data to perform training diagnosis, and forming a detection model based on the probabilistic neural network;
and secondly, optimizing the established detection model based on the probabilistic neural network by using the rough set and the genetic algorithm respectively, and establishing a RST-PNN-GA neural network-based power equipment temperature rise detection early warning model.
The training method of the temperature rise detection early warning model of the power equipment is implemented according to the following specific steps:
inputting original fault characteristic information parameters acquired through a simulation test platform into a probabilistic neural network main body model for training;
performing attribute reduction on the original fault characteristic information parameters by using a rough set reduction theory, firstly determining condition attributes and decision attributes, forming a decision on fault type information, and finally forming a reduction attribute decision;
inputting the reduced fault characteristic parameter reduction decision table into the probabilistic neural network main body model;
fourthly, network initialization is carried out on a probabilistic neural network main body model and an improved genetic algorithm;
fifthly, performing forward feedback calculation by using a probabilistic neural network, firstly receiving the parameters input in the step three, determining the number of network neurons, and calculating the weight by using a basic algorithm of the probabilistic neural network to obtain a probability density value; the input and output of the probabilistic neural network are a 5-input 2-output model, 3 layers of neurons are preset in the middle layer, and the number of the neurons is 18;
sixthly, judging the output probability density value error and the iteration times thereof, and outputting a result according to the requirement;
if the errors judged in the step sixteenth are not in accordance with the requirements, performing error back propagation calculation by using a genetic algorithm, firstly calculating fitness, updating the population again, performing the operation of the genetic algorithm on the population again, updating new weight values, and recalculating the weight values again to the probabilistic neural network until the errors and the iteration times meet the requirements;
and outputting the diagnosis result.
The training method of the probabilistic neural network body model is implemented according to the following steps:
1) normalization processing, namely initializing a network, processing input sample data and training the sample data into matrix samples as follows:
Figure BDA0003110137880000121
normalizing all sample data in the matrix X to obtain a normalization coefficient BTAnd the normalized learning sample C can be expressed as:
Figure BDA0003110137880000122
CM×N＝Bm×1[11...1]1×n·Xm×n
2) inputting the sample characteristic data processed in the first step into an input layer;
Figure BDA0003110137880000131
3) exerting the left and right of a mode layer, carrying out refinement calculation on the mode distance, and then carrying out normalization processing on sample data with processing characteristics formed by P n-dimensional vectors, wherein the matrix form is as follows:
Figure BDA0003110137880000132
the calculation of the pattern distance requires that each normalized matrix to be processed and each normalized feature sample matrix are used to calculate the euclidean distance.
Assume normalized training samples as ciI is 1,2 … m; the normalized matrix to be processed is djJ ═ 1,2.. p. The jth data to be processed (d)j) And the ith training sample (c)i) The euclidean distance of (a) may be expressed as: eij. The details are as follows:
Figure BDA0003110137880000133
4) activating gaussian function neurons of the pattern layer:
Figure BDA0003110137880000141
5) classifying the feature data samples which are not trained yet and solving the probability values of the data:
Figure BDA0003110137880000142
6) generalizing the probability value formula as follows:
Figure BDA0003110137880000143
in order to obtain the probability value of the finally output unprocessed feature signal data sample, learning and training are carried out one by one according to the aforementioned calculation steps, and meanwhile, the threshold value of the probabilistic neural network is set to search for the optimal probability value in a specified range, so that the output type corresponding to the obtained optimal probability value is the type of the feature signal data sample which is not processed, namely, the training and diagnosis process of the whole probabilistic neural network model is completed.
The operation of the genetic algorithm comprises selection operation, crossover operation and mutation operation.
The rough reduction theory comprises:
first, assume that the description of the information table S is:
S＝(U,C,D,V,F)；
wherein, U is discourse domain, C is condition attribute set, D is decision attribute set, V is Va∈C∪DVaIs the value range of the attribute, wherein is the attribute aVaU (coud) → V is an information decision function;
for equivalence relations, any subset of attributes has
Figure BDA0003110137880000144
Figure BDA0003110137880000151
A partition U/B may be formed for the discourse using equivalence relations. Wherein each small region divided is of an equivalent type and can be described as [ x ]]B＝{y∈U|(x，y)∈IND（B)}；
For each subset in any domain of discourse
Figure BDA0003110137880000152
There is an upper and lower approximation for B, which are described as respectively
Figure BDA0003110137880000153
Figure BDA0003110137880000154
In the formula, the B-based partition domain must be partitioned into the X class set of objects, which is called the upper approximation. The lower approximation refers to the set of objects that are likely to be classified into X classes in the B-based partitioned region.
Wherein collection of attributes
Figure BDA0003110137880000155
Further, the definition of the positive, negative regions (outer regions) and the boundary region with respect to the decision set D is as follows:
POSB(D)＝∪XeU/D B(X)；
NEGB(D)＝U-∪XeU/DB(X)；
Figure BDA0003110137880000156
in the formula, all the regions that are necessarily divided into one divided region based on D in the divided region of B are called positive regions, and mainly reflect the classification capability of the attribute B relative to D.
In the case of P, the number of P,
Figure BDA0003110137880000157
the dependency of Q on P can be defined as
k＝γP(Q)＝|POSP(Q)|/|U|；
For any subset of conditional attributes
Figure BDA0003110137880000158
If:
POSR(D)＝POSC(D)；
POSR(D)≠POSR-{a}(D)，a∈R；
r is a relative reduction of C.
The accuracy of the RST-PNN-GA neural network algorithm-based power equipment temperature rise detection early warning method is 97%.
The RST-PNN-GA neural network algorithm-based power equipment temperature rise detection and early warning method analyzes the characteristic signals by adopting a method of fusing and improving a plurality of neural networks, overcomes the defects of artificial neural network detection, can more accurately and effectively judge the prediction and early warning of the power equipment, and further can efficiently overhaul.
The above description is of the preferred embodiment of the present invention and is not intended to limit the invention, and those skilled in the art may make modifications and improvements within the spirit and principle of the present invention.

Claims (9)

1. A RST-PNN-GA-based power equipment temperature rise detection and early warning method is characterized by comprising the following steps:
the method comprises the steps of constructing an electric power equipment temperature rise online monitoring system, then monitoring the temperature of electric power equipment in real time by using the electric power equipment temperature rise online monitoring system to obtain electric power equipment temperature data, and taking the electric power equipment temperature data obtained through real-time monitoring as an input variable;
secondly, constructing a power equipment temperature rise detection early warning model based on an RST-PNN-GA neural network algorithm, and inputting a part of power equipment temperature data obtained in the step into the constructed power equipment temperature rise detection early warning model for training;
inputting a part of temperature data of the electric power equipment obtained in the step into the trained temperature rise detection and early warning model of the electric power equipment, processing the input temperature data of the electric power equipment by the temperature rise detection and early warning model of the electric power equipment, and completing temperature rise detection and early warning of the electric power equipment.
2. The RST-PNN-GA-based power equipment temperature rise detection and early warning method according to claim 1, wherein the RST-PNN-GA-based power equipment temperature rise detection and early warning method comprises the following steps: the power equipment temperature rise on-line monitoring system comprises a plurality of monitoring slave machines, a plurality of groups of monitoring host machines and an upper computer monitoring center, wherein the monitoring slave machines are used for collecting the temperature of the power equipment temperature measurement nodes and uploading collected temperature data to the monitoring host machines; each group of monitoring host corresponds to a plurality of monitoring slave machines, and the monitoring host is used for summarizing, displaying and storing data of the corresponding monitoring slave machines and uploading the data to an upper computer monitoring center in a GPRS or RS485 mode; the upper computer monitoring center collects data collected by the monitoring hosts, realizes storage and display of the data and issuing of a threshold value, realizes fault early warning while carrying out data analysis, and can carry out fault early warning according to temperature data.
3. The RST-PNN-GA-based power equipment temperature rise detection and early warning method according to claim 2, wherein: the monitoring slave machine comprises a temperature and humidity sensor and a temperature collector, the temperature and humidity sensor monitors the temperature and the humidity of the working environment of the power equipment and the temperature of a monitoring node in real time, and the temperature collector receives temperature data from the temperature and humidity sensor and uploads the temperature data to the monitoring master machine in a wireless communication mode; the monitoring host comprises a wireless module, a clock, an alarm indicator light, a touch screen and a single chip microcomputer, wherein the single chip microcomputer is electrically connected with the wireless module, the clock, the alarm indicator light and the touch screen respectively.
4. The RST-PNN-GA-based power equipment temperature rise detection and early warning method according to claim 3, wherein: the wireless module is a 433MHz wireless communication module, the type that temperature and humidity sensor adopted is SHT715, the singlechip model is STM8L151K4T6, the touch-sensitive screen adopts industrial resistance-type touch-sensitive screen.
5. The RST-PNN-GA-based power equipment temperature rise detection and early warning method according to claim 1, wherein the method for constructing the power equipment temperature rise detection and early warning model comprises the following steps:
the method comprises the steps of establishing a probabilistic neural network main model, inputting collected original fault characteristic sample data to perform training diagnosis, and forming a detection model based on the probabilistic neural network;
and secondly, optimizing the established detection model based on the probabilistic neural network by using the rough set and the genetic algorithm respectively, and establishing a RST-PNN-GA neural network-based power equipment temperature rise detection early warning model.
6. The RST-PNN-GA-based power equipment temperature rise detection and early warning method is characterized in that the power equipment temperature rise detection and early warning model training method is implemented according to the following specific steps:
inputting original fault characteristic information parameters acquired through a simulation test platform into a probabilistic neural network main body model for training;
performing attribute reduction on the original fault characteristic information parameters by using a rough set reduction theory, firstly determining condition attributes and decision attributes, forming a decision on fault type information, and finally forming a reduction attribute decision;
inputting the reduced fault characteristic parameter reduction decision table into the probabilistic neural network main body model;
fourthly, network initialization is carried out on a probabilistic neural network main body model and an improved genetic algorithm;
fifthly, performing forward feedback calculation by using a probabilistic neural network, firstly receiving the parameters input in the step three, determining the number of network neurons, and calculating the weight by using a basic algorithm of the probabilistic neural network to obtain a probability density value; the input and output of the probabilistic neural network are a 5-input 2-output model, 3 layers of neurons are preset in the middle layer, and the number of the neurons is 18;
sixthly, judging the output probability density value error and the iteration times thereof, and outputting a result according to the requirement;
if the errors judged in the step sixteenth are not in accordance with the requirements, performing error back propagation calculation by using a genetic algorithm, firstly calculating fitness, updating the population again, performing the operation of the genetic algorithm on the population again, updating new weight values, and recalculating the weight values again to the probabilistic neural network until the errors and the iteration times meet the requirements;
and outputting the diagnosis result.
7. The RST-PNN-GA-based power equipment temperature rise detection and early warning method according to claim 6, wherein the training method of the probabilistic neural network body model is implemented specifically according to the following steps:
1) normalization processing, namely initializing a network, processing input sample data and training the sample data into matrix samples as follows:
Figure FDA0003110137870000031
normalizing all sample data in the matrix X to obtain a normalization coefficient BTAnd the normalized learning sample C can be expressed as:
Figure FDA0003110137870000032
CM×N＝Bm×1[11...1]1×n·Xm×n
2) inputting the sample characteristic data processed in the first step into an input layer;
Figure FDA0003110137870000041
3) exerting the left and right of a mode layer, carrying out refinement calculation on the mode distance, and then carrying out normalization processing on sample data with processing characteristics formed by P n-dimensional vectors, wherein the matrix form is as follows:
Figure FDA0003110137870000042
the calculation of the pattern distance requires that each normalized matrix to be processed and each normalized feature sample matrix are used to calculate the euclidean distance.
Assume normalized training samples as ciI is 1,2 … m; the normalized matrix to be processed is djJ ═ 1,2.. p. The jth data to be processed (d)j) And the ith training sample (c)i) The euclidean distance of (a) may be expressed as: eij. The details are as follows:
Figure FDA0003110137870000043
4) activating gaussian function neurons of the pattern layer:
Figure FDA0003110137870000051
5) classifying the feature data samples which are not trained yet and solving the probability values of the data:
Figure FDA0003110137870000052
6) generalizing the probability value formula as follows:
Figure FDA0003110137870000053
in order to obtain the probability value of the finally output unprocessed feature signal data sample, learning and training are carried out one by one according to the aforementioned calculation steps, and meanwhile, the threshold value of the probabilistic neural network is set to search for the optimal probability value in a specified range, so that the output type corresponding to the obtained optimal probability value is the type of the feature signal data sample which is not processed, namely, the training and diagnosis process of the whole probabilistic neural network model is completed.
8. The RST-PNN-GA-based power equipment temperature rise detection and early warning method as claimed in claim 7, wherein the operation of the genetic algorithm comprises selection operation, crossover operation and mutation operation.
9. The RST-PNN-GA-based power equipment temperature rise detection and early warning method according to claim 8, wherein the rough intensive theory comprises:
first, assume that the description of the information table S is:
S＝(U,C,D,V,F)；
wherein, U is discourse domain, C is condition attribute set, D is decision attribute set, V is Va∈C∪DVaV being the attributeaValue range, where is the value range of attribute a, f: u (C U)D) → V is the information decision function;
for equivalence relations, any subset of attributes has
Figure FDA0003110137870000066
Figure FDA0003110137870000061
A partition U/B may be formed for the discourse using equivalence relations. Wherein each small region divided is of an equivalent type and can be described as [ x ]]B＝{y∈U|(x，y)∈IND(B)}；
For each subset in any domain of discourse
Figure FDA0003110137870000067
There is an upper and lower approximation for B, which are described as respectively
Figure FDA0003110137870000062
Figure FDA0003110137870000063
In the formula, the B-based partition domain must be partitioned into the X class set of objects, which is called the upper approximation. The lower approximation refers to the set of objects that are likely to be classified into X classes in the B-based partitioned region.
Wherein collection of attributes
Figure FDA0003110137870000068
Further, the definition of the positive, negative regions (outer regions) and the boundary region with respect to the decision set D is as follows:
Figure FDA0003110137870000064
NEGB(D)＝U-∪XeU/DB(X)；
Figure FDA0003110137870000065
in the formula, all the regions that are necessarily divided into one divided region based on D in the divided region of B are called positive regions, and mainly reflect the classification capability of the attribute B relative to D.
In the case of P, the number of P,
Figure FDA0003110137870000069
the dependency of Q on P can be defined as
k＝γP(Q)＝|POSP(Q)|/|U|；
For any subset of conditional attributes
Figure FDA00031101378700000610
If:
POSR(D)＝POSC(D)；
POSR(D)≠POSR-{a}(D)，a∈R；
r is a relative reduction of C.
CN202110648478.5A 2021-06-10 2021-06-10 RST-PNN-GA-based power equipment temperature rise detection and early warning method Pending CN113358157A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110648478.5A CN113358157A (en) 2021-06-10 2021-06-10 RST-PNN-GA-based power equipment temperature rise detection and early warning method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110648478.5A CN113358157A (en) 2021-06-10 2021-06-10 RST-PNN-GA-based power equipment temperature rise detection and early warning method

Publications (1)

Publication Number Publication Date
CN113358157A true CN113358157A (en) 2021-09-07

Family

ID=77533666

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110648478.5A Pending CN113358157A (en) 2021-06-10 2021-06-10 RST-PNN-GA-based power equipment temperature rise detection and early warning method

Country Status (1)

Country Link
CN (1) CN113358157A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113867460A (en) * 2021-11-10 2021-12-31 成都捷豹恒昌电气有限公司 Temperature and humidity monitoring and treatment method and system for transformer substation
CN117539717A (en) * 2023-11-23 2024-02-09 国网黑龙江省电力有限公司牡丹江供电公司 Device working condition intelligent supervision system and method based on data analysis

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107450016A (en) * 2017-07-24 2017-12-08 西安工程大学 Fault Diagnosis for HV Circuit Breakers method based on RST CNN
US20200063665A1 (en) * 2018-01-25 2020-02-27 Dalian University Of Technology Aero-engine full flight envelope model adaptive modification method based on deep learning algorithm
CN111060815A (en) * 2019-12-17 2020-04-24 西安工程大学 GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method
CN111273125A (en) * 2020-01-07 2020-06-12 西安工程大学 RST-CNN-based power cable channel fault diagnosis method
CN112529275A (en) * 2020-12-02 2021-03-19 中国矿业大学 Wind power climbing event prediction method based on feature extraction and deep learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107450016A (en) * 2017-07-24 2017-12-08 西安工程大学 Fault Diagnosis for HV Circuit Breakers method based on RST CNN
US20200063665A1 (en) * 2018-01-25 2020-02-27 Dalian University Of Technology Aero-engine full flight envelope model adaptive modification method based on deep learning algorithm
CN111060815A (en) * 2019-12-17 2020-04-24 西安工程大学 GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method
CN111273125A (en) * 2020-01-07 2020-06-12 西安工程大学 RST-CNN-based power cable channel fault diagnosis method
CN112529275A (en) * 2020-12-02 2021-03-19 中国矿业大学 Wind power climbing event prediction method based on feature extraction and deep learning

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
张浩等: "10kV断路器智能化关键技术研究", 中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑, no. 6, pages 26 - 29 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113867460A (en) * 2021-11-10 2021-12-31 成都捷豹恒昌电气有限公司 Temperature and humidity monitoring and treatment method and system for transformer substation
CN117539717A (en) * 2023-11-23 2024-02-09 国网黑龙江省电力有限公司牡丹江供电公司 Device working condition intelligent supervision system and method based on data analysis
CN117539717B (en) * 2023-11-23 2024-05-14 国网黑龙江省电力有限公司牡丹江供电公司 Device working condition intelligent supervision system and method based on data analysis

Similar Documents

Publication Publication Date Title
CN110990461A (en) Big data analysis model algorithm model selection method and device, electronic equipment and medium
CN113762329A (en) Method and system for constructing state prediction model of large rolling mill
CN108665112A (en) Photovoltaic fault detection method based on Modified particle swarm optimization Elman networks
CN113358157A (en) RST-PNN-GA-based power equipment temperature rise detection and early warning method
WO2013081717A2 (en) System and method employing a hierarchical load feature database to identify electric load types of different electric loads
CN111273125A (en) RST-CNN-based power cable channel fault diagnosis method
Liao et al. Remaining useful life with self-attention assisted physics-informed neural network
CN112906764B (en) Communication safety equipment intelligent diagnosis method and system based on improved BP neural network
CN116050888A (en) Method applied to intelligent high-voltage switch cabinet sensor health state assessment
CN111563827A (en) Load decomposition method based on electrical appliance physical characteristics and residential electricity consumption behaviors
Li et al. An optimal stacking ensemble for remaining useful life estimation of systems under multi-operating conditions
Yao et al. Fault identification of lithium-ion battery pack for electric vehicle based on ga optimized ELM neural network
CN117951629B (en) Method and system for detecting abnormal behavior of charge and discharge temperature of battery pack
Dhibi et al. Interval-valued reduced ensemble learning based fault detection and diagnosis techniques for uncertain grid-connected PV systems
Guan et al. Application of a novel PNN evaluation algorithm to a greenhouse monitoring system
Xiong et al. Critical Review on Improved Electrochemical Impedance Spectroscopy-Cuckoo Search-Elman Neural Network Modeling Methods for Whole-Life-Cycle Health State Estimation of Lithium-Ion Battery Energy Storage Systems
Niu et al. An online fault diagnosis method for lithium-ion batteries based on signal decomposition and dimensionless indicators selection
Bai et al. Towards trustworthy remaining useful life prediction through multi-source information fusion and a novel LSTM-DAU model
CN113033898A (en) Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network
CN113884807A (en) Power distribution network fault prediction method based on random forest and multi-layer architecture clustering
CN116259161B (en) Power failure early warning system
Cao et al. A novel fault diagnosis strategy for heterogeneous wireless sensor networks
Zhu et al. State of Health Estimation of Lithium‐Ion Battery Using Time Convolution Memory Neural Network
CN110837932A (en) Thermal power prediction method of solar heat collection system based on DBN-GA model
Yan et al. OLTC Fault Diagnosis Method Based on Time Domain Analysis and Kernel Extreme Learning Machine

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination