CN113469222A - RST-PNN-GA-based high-voltage circuit breaker fault detection method - Google Patents
RST-PNN-GA-based high-voltage circuit breaker fault detection method Download PDFInfo
- Publication number
- CN113469222A CN113469222A CN202110649592.XA CN202110649592A CN113469222A CN 113469222 A CN113469222 A CN 113469222A CN 202110649592 A CN202110649592 A CN 202110649592A CN 113469222 A CN113469222 A CN 113469222A
- Authority
- CN
- China
- Prior art keywords
- circuit breaker
- neural network
- voltage circuit
- pnn
- rst
- 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
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 44
- 238000013528 artificial neural network Methods 0.000 claims abstract description 71
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 41
- 238000000034 method Methods 0.000 claims abstract description 27
- 230000002068 genetic effect Effects 0.000 claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims description 31
- 239000011159 matrix material Substances 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 18
- 238000003745 diagnosis Methods 0.000 claims description 17
- 238000012544 monitoring process Methods 0.000 claims description 17
- 230000009467 reduction Effects 0.000 claims description 16
- 210000002569 neuron Anatomy 0.000 claims description 13
- 238000004891 communication Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 9
- 230000010365 information processing Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 238000013500 data storage Methods 0.000 claims description 5
- 238000005192 partition Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 claims description 4
- 230000003213 activating effect Effects 0.000 claims description 3
- 230000035772 mutation Effects 0.000 claims description 3
- 238000010248 power generation Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 2
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical group [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 8
- 230000007246 mechanism Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 238000010891 electric arc Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000004907 flux Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013529 biological neural network Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000009740 moulding (composite fabrication) Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/327—Testing of circuit interrupters, switches or circuit-breakers
- G01R31/3271—Testing of circuit interrupters, switches or circuit-breakers of high voltage or medium voltage devices
- G01R31/3275—Fault detection or status indication
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a RST-PNN-GA-based high-voltage circuit breaker fault detection method, and belongs to the technical field of high-voltage circuit breaker detection methods. According to the method, a probability neural network is adopted to analyze fault characteristic signals, a rough set is utilized to simplify collected fault characteristic samples, a genetic algorithm is utilized to perform error back-propagation calculation, and a high-voltage circuit breaker fault detection method based on a RST-PNN-GA neural network algorithm is established.
Description
Technical Field
The invention belongs to the technical field of high-voltage circuit breaker detection methods, and particularly relates to a RST-PNN-GA neural network algorithm-based high-voltage circuit breaker fault detection method.
Background
The high-voltage circuit breaker is the most main control and protection device of a power system and is related to the reliability and safety of power transmission, power distribution and power utilization. High voltage circuit breakers can achieve a variety of operations in the event of system faults and non-fault conditions. The breaker can close, bear and break the normal current of the operation loop, and can also close, bear and break the specified overload current within the specified time.
High-voltage circuit breakers generally use an electromagnet as a first control element, and most of the operating mechanisms are direct-current electromagnets. When current passes through the coil, magnetic flux is generated in the magnet, and the movable iron core is influenced by the magnetic force, so that the breaker is opened or closed. The current in the switching-on and switching-off coil can be used as rich information for diagnosing the mechanical fault of the high-voltage circuit breaker.
The existing high-voltage circuit breaker fault maintenance methods are many, and various artificial intelligence algorithms are involved, such as: fuzzy control can clarify fuzzy concepts or natural languages by using an accurate mathematical tool, but certain human factors exist in the determination process of membership functions and fuzzy rules of the fuzzy control; the radial basis function neural network provides a better structural system for the fault diagnosis problem of the circuit breaker, but has the defects that the self reasoning process and reasoning basis cannot be explained, and the neural network cannot work normally when the data is insufficient.
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.
Disclosure of Invention
The invention aims to provide a high-voltage circuit breaker fault detection method based on a RST-PNN-GA neural network algorithm, which is characterized in that a Probability Neural Network (PNN) is adopted to analyze fault characteristic signals, meanwhile, a Rough Set (RST) is utilized to simplify collected fault characteristic samples, a Genetic Algorithm (GA) is utilized to perform error back-propagation calculation, and the high-voltage circuit breaker fault detection method based on the RST-PNN-GA neural network algorithm is established.
In order to achieve the above purpose, the present invention provides the following technical solutions:
the RST-PNN-GA-based high-voltage circuit breaker fault detection method comprises the following steps:
firstly, respectively connecting a magnetic balance type Hall current sensor with a circuit breaker opening and closing coil and a data processing system to construct an opening and closing coil current online monitoring system; then, the on-off coil current data obtained by on-off coil current on-line monitoring system real-time monitoring is utilized, and the on-off coil current data obtained by real-time monitoring is used as an input variable;
secondly, constructing a high-voltage circuit breaker fault type prediction model based on an RST-PNN-GA neural network algorithm, and inputting a part of opening and closing coil current data obtained in the step into the constructed high-voltage circuit breaker fault type prediction model for training;
inputting a part of opening and closing coil current data obtained in the step into the high-voltage circuit breaker fault type prediction model trained in the step II, and processing the input opening and closing coil current data through the high-voltage circuit breaker fault type prediction model to complete fault detection of the high-voltage circuit breaker.
The on-line monitoring system for the current of the split switching coil in the step comprises a single chip microcomputer, wherein the single chip microcomputer is respectively connected with a power module, an information processing unit, a 5G communication module, a Zigbee communication module and a data storage unit; the power supply module is respectively connected with a solar power generation module and a storage battery; the input end of the information processing unit is connected with a magnetic balance type Hall current sensor, and the input end of the magnetic balance type Hall current sensor is connected with a circuit breaker opening and closing coil.
Preferably, the type of the single chip microcomputer is STM32F 407.
Further, the invention also provides a construction method of the high-voltage circuit breaker fault type prediction 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 respectively utilizing a rough set and a genetic algorithm, and establishing a RST-PNN-GA neural network-based high-voltage circuit breaker fault detection model.
Further, the invention also provides a training method of the high-voltage circuit breaker fault type prediction 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 13 input and 5 output models, 3 layers of neurons are preset in the middle layer, and the number of the neurons is 27;
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:
sorting all sample data in matrix XNormalizing to obtain normalized coefficient BTAnd the normalized learning sample C can be expressed as:
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;
3) the method comprises the following steps of exerting the effect of a mode layer, carrying out refinement calculation on 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:
calculating the pattern distance requires using each normalized matrix to be processed and each normalized feature sample matrix to obtain the Euclidean distance;
assume normalized training samples as ciI1, 2.. m; the normalized matrix to be processed is djJ 1,2.. p, then the jth data d to be processedjAnd the ith training sample ciThe euclidean distance of (a) may be expressed as: eijSpecifically, the following is shown:
4) activating gaussian function neurons of the pattern layer:
5) classifying the feature data samples which are not trained yet and solving the probability values of the data:
6) generalizing the probability value formula as follows:
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 body model is completed.
Further, the operation of the genetic algorithm comprises a selection operation, a crossover operation and a mutation operation.
Compared with the prior art, the invention has the beneficial effects that:
(1) the RST-PNN-GA neural network algorithm-based high-voltage circuit breaker fault detection method constructs a high-voltage circuit breaker fault type prediction model, and the model 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 value discrimination, the diagnosis accuracy of the high-voltage circuit breaker fault diagnosis model is improved, and a reliable fault diagnosis result is obtained; the simplest attribute table of the fault feature sample is updated again by using the attribute reduction operation of the rough set, so that the diagnosis process time is reduced, the diagnosis rate is improved, and the diagnosis rate of the high-voltage circuit breaker fault detection model is improved;
(2) constructing an on-line monitoring system for the current of the switching-on and switching-off coil to monitor the current data of the switching-on and switching-off coil in real time, and inputting the data into a high-voltage circuit breaker fault prediction model for analysis to obtain the fault type;
(3) the RST-PNN-GA neural network algorithm-based high-voltage circuit breaker fault detection 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 and updating on errors of the learning task while learning training; the method is applied to fault diagnosis of the high-voltage circuit breaker, and can judge the fault type more accurately and carry out state maintenance.
Drawings
FIG. 1 is a schematic structural diagram of an on-line current monitoring system of a switching-on/switching-off coil adopted in the RST-PNN-GA neural network algorithm-based high-voltage circuit breaker fault detection method of the invention;
FIG. 2 is a structural diagram of a RST-PNN-GA neural network in the RST-PNN-GA neural network algorithm-based high-voltage circuit breaker fault detection method of the invention;
FIG. 3 is a fault type analysis and detection flow chart of the high-voltage circuit breaker fault detection method based on RST-PNN-GA neural network algorithm;
fig. 4 is a characteristic curve of the closing/opening coil current in embodiment 1 of the present invention.
Reference numerals: the device comprises a single chip microcomputer 1, a power supply module 2, an information processing unit 3, a magnetic balance type Hall current sensor 4, a 5-5G communication module, a 6-Zibbee communication module, a 7-solar power generation module, a 8-storage battery, a 9-data storage unit and a 10-breaker opening and closing coil.
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.
Examples
Referring to fig. 1-3, the method for detecting a fault of a high-voltage circuit breaker based on the RST-PNN-GA neural network algorithm provided in this embodiment includes the following steps:
firstly, respectively connecting a magnetic balance type Hall current sensor with a circuit breaker opening and closing coil and a data processing system to construct an opening and closing coil current online monitoring system; and then, the on-off switching coil current data obtained by real-time monitoring of the on-off switching coil current on-line monitoring system is utilized, and the on-off switching coil current data obtained by real-time monitoring is used as an input variable.
The on-line monitoring system for the opening and closing coil current comprises a single chip microcomputer 1, wherein the single chip microcomputer 1 is respectively connected with a power module 2, an information processing unit 3, a 5G communication module 5, a Zigbee communication module and a data storage unit 9, and the model of the single chip microcomputer 1 is STM32F 407; power module 2 is connected with solar energy power module 7 respectively, battery 8, provide the electric energy for whole divide-shut brake coil electric current on-line monitoring system through power module 2 and solar energy power module 7, battery 8 is used for saving unnecessary electric quantity, in order to need from time to time, singlechip 1 passes through 4G communication module, Zigbee communication module externally communicates, singlechip 1 is connected with information processing unit 3, information processing unit 3 is connected with magnetic balance formula hall current sensor 4, magnetic balance formula hall current sensor 4 is connected with circuit breaker divide-shut brake coil 10, mutually support can be to the electric current data who obtains, and handle the data that obtain, and save data information in data storage unit 9.
And secondly, constructing a high-voltage circuit breaker fault type prediction model based on an RST-PNN-GA neural network algorithm, and inputting a part of opening and closing coil current data obtained in the step into the constructed high-voltage circuit breaker fault type prediction model for training.
The method for constructing the high-voltage circuit breaker fault type prediction model comprises the following specific steps;
firstly, establishing a probabilistic neural network main body model, inputting collected original fault feature sample data for training and 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 a rough set and a genetic algorithm respectively, and establishing a high-voltage circuit breaker fault detection model based on the RST-PNN-GA neural network.
The method comprises the following specific steps of training a fault type prediction model of the high-voltage circuit breaker;
firstly, inputting original fault characteristic information parameters collected by a simulation test platform into a probabilistic neural network main body model for training;
secondly, 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;
thirdly, inputting the reduced fault characteristic parameter reduction decision table to a probabilistic neural network main body model;
fourthly, initializing the probabilistic neural network main body model and the genetic algorithm serving as the improvement;
fifthly, utilizing the probabilistic neural network to perform forward feedback calculation, firstly receiving the parameters input in the step three, determining the number of network neurons, and utilizing a basic algorithm of the probabilistic neural network to calculate the weight to obtain a probability density value; the input and output of the probabilistic neural network are 13 input and 5 output models, 3 layers of neurons are preset in the middle layer, and the number of the neurons is 27;
judging the error of the output probability density value and the iteration times thereof, and outputting the result according to the steps if the error meets the requirements;
fifthly, if the errors judged in the step VI do not meet the requirements, error back propagation calculation is carried out by using a genetic algorithm, firstly, fitness is calculated, the population is updated again, the genetic algorithm is operated on the population, new weight values are updated accordingly, the weight values are returned to the probabilistic neural network again, and the step fifthly is carried out again until the errors and the iteration times meet the requirements;
and outputting a diagnosis result.
The training method of the probabilistic neural network body model is implemented according to the following steps:
firstly, normalization processing, namely initializing a network, processing input sample data and training the sample data into matrix samples as follows:
normalizing all sample data in the matrix X to obtain a normalization coefficient BTAnd the normalized learning sample C can be expressed as:
CM×N＝Bm×1[11...1]1×n·Xm×n，
secondly, inputting the sample characteristic data processed in the first step into an input layer;
thirdly, playing the role of a mode layer, carrying out refinement calculation on the mode distance, and then carrying out normalization processing on the sample data with processing characteristics formed by P n-dimensional vectors, wherein the matrix form is as follows:
calculating the pattern distance requires using each normalized matrix to be processed and each normalized feature sample matrix to obtain the Euclidean distance;
assume normalized training samples as ciI1, 2.. m; the normalized matrix to be processed is djJ 1,2.. p, then the jth data d to be processedjAnd the ith training sample ciThe euclidean distance of (a) may be expressed as: eijSpecifically, the following is shown:
activating Gaussian function neurons of the mode layer:
classifying the untrained characteristic data samples and calculating the probability values of the data:
induction probability value formula as follows:
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 body model is completed.
The operation of the genetic algorithm comprises selection operation, crossover operation and mutation operation.
After the input/output design of the probabilistic neural network is determined, the invention carries out normalization processing by taking 10 groups of data as the input vector of the probabilistic neural network to carry out quantization coding on the fault type, and constructs a fault detection model of the high-voltage circuit breaker based on the RST-PNN-GA neural network.
The rough reduction theory comprises:
first, assume that the description of the information table S is:
S＝(U,C,D,V,F)；
in the formula of UFor domain, C is a conditional attribute set, D is a decision attribute set, and 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
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
There is an upper and lower approximation for B, which are described as respectively
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
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)＝∪X∈U/D B(X)；
NEGB(D)＝U-∪X∈U/DB(X)；
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,
the dependency of Q on P can be defined as
k＝θP(Q)＝|POSP(Q)|/|U|；
For any subset of conditional attributes
If:
POSR(D)＝POSC(D)；
POSR(D)≠POSR-{a}(D)，a∈R；
r is a relative reduction of C.
Inputting a part of opening and closing coil current data obtained in the step into the high-voltage circuit breaker fault type prediction model trained in the step II, and processing the input opening and closing coil current data through the high-voltage circuit breaker fault type prediction model to complete fault detection of the high-voltage circuit breaker.
Test examples
On the basis of the embodiment, please refer to fig. 4, the t0 is taken as a zero point of the command time to extract fault characteristic parameters I1, I2, I3, t1, t2, t3, t4, and t5, so as to perform state monitoring on the circuit breaker, and obtain ten sets of fault sample data, where the ten sets of fault sample data include a normal mechanism (a), an excessively low operating voltage (B), a jam (C) at the start stage of a closing iron core, a jam (D) of the operating mechanism, and an excessively large closing iron core idle stroke (E), and the data acquisition conditions are specifically shown in table 1;
TABLE 1 Fault sample data
From the characteristic curve of the closing/opening coil current of fig. 4, it can be seen that:
(1) stage I, t is t 0-t 1; the coil starts to be electrified at the time t0, and the iron core starts to move at the time t 1; t0 is the time of the opening and closing command of the circuit breaker, and is the starting point of the opening and closing action timing of the circuit breaker; t1 is the time when the current and magnetic flux in the coil rise enough to drive the core to move, i.e. the core starts to move; the characteristic of this stage is that the current rises exponentially, the iron core is static; the time of this stage is related to the control supply voltage and the coil resistance;
(2) stage II, t is t 1-t 2; at this stage, the core starts to move and the current drops; t2 is the valley point of the control current, which represents that the iron core has triggered the load of the operating machine to significantly decelerate or stop movement;
(3) stage III, t is t 2-t 3; at this stage, the iron core stops moving, and the current rises exponentially;
(4) stage IV, t is t 3-t 4; the phase is continued from the phase III, and the current reaches an approximate steady state;
(5) stage V, t is t 4-t 5; the current breaking stage, in which the auxiliary switch is broken, an electric arc is generated between the contacts of the auxiliary switch and is elongated, the voltage of the electric arc is rapidly increased, and the current is rapidly reduced until the electric arc is extinguished;
the current waveform is analyzed, the current at t 0-t 1 can reflect the state of the coil (such as whether the resistance is normal or not), and the change of the current at t 1-t 2 represents the change of the mechanical load of the iron core, such as the existence of jamming, tripping and energy release; t2 is the moving moment of the moving contact, and the process of the mechanism driving the moving contact to switch on and off through the transmission system from t2, namely the moving process of the moving contact; t4 is the time when the auxiliary contact of the circuit breaker is cut off; the change of the current at the time from t0 to t4 can reflect the working condition of the transmission system of the mechanical operating mechanism.
The output of the fault type is represented by a binary number, which is specifically shown in table 2:
TABLE 2 Fault type output representation
The high-voltage circuit breaker fault detection method based on the RST-PNN-GA neural network algorithm has the accuracy of 96.6%.
The RST-PNN-GA neural network algorithm-based high-voltage circuit breaker fault detection method analyzes the fault characteristic signal 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 fault type of the circuit breaker, 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 (8)
1. The RST-PNN-GA-based high-voltage circuit breaker fault detection method comprises the steps of constructing an on-line monitoring system for the current of a switching-on and switching-off coil, monitoring the current data of the switching-on and switching-off coil in real time, constructing a fault type detection model of the high-voltage circuit breaker, training, inputting the current data of the switching-on and switching-off coil into the fault type detection model of the high-voltage circuit breaker, and carrying out fault detection, and is characterized in that: the high-voltage circuit breaker fault type detection model is constructed on the basis of a RST-PNN-GA neural network algorithm.
2. The RST-PNN-GA-based high-voltage circuit breaker fault detection method according to claim 1, wherein the high-voltage circuit breaker fault type prediction model is constructed by the following method:
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 respectively utilizing a rough set and a genetic algorithm, and establishing a RST-PNN-GA neural network-based high-voltage circuit breaker fault detection model.
3. The RST-PNN-GA-based high-voltage circuit breaker fault detection method according to claim 2, wherein the training method of the high-voltage circuit breaker fault type prediction model is implemented by 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 13 input and 5 output models, 3 layers of neurons are preset in the middle layer, and the number of the neurons is 27;
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.
4. The RST-PNN-GA-based high-voltage circuit breaker fault detection method of claim 3, wherein 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:
normalizing all sample data in the matrix X to obtain a normalization coefficient BTAnd the normalized learning sample C can be expressed as:
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;
3) the method comprises the following steps of exerting the effect of a mode layer, carrying out refinement calculation on 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:
calculating the pattern distance requires using each normalized matrix to be processed and each normalized feature sample matrix to obtain the Euclidean distance;
assume normalized training samples as ciI is 1,2 … m; the normalized matrix to be processed is djJ 1,2.. p, then the jth data d to be processedjAnd the ith training sample ciThe euclidean distance of (a) may be expressed as: eijSpecifically, the following is shown:
4) activating gaussian function neurons of the pattern layer:
5) classifying the feature data samples which are not trained yet and solving the probability values of the data:
6) generalizing the probability value formula as follows:
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 body model is completed.
5. The RST-PNN-GA-based high-voltage circuit breaker fault detection method according to claim 4, wherein: the operation of the genetic algorithm comprises selection operation, crossover operation and mutation operation.
6. The RST-PNN-GA-based high-voltage circuit breaker fault detection method according to claim 1, wherein: the on-line current monitoring system for the switching-on and switching-off coils comprises a single chip microcomputer, wherein the single chip microcomputer is respectively connected with a power supply module, an information processing unit, a 5G communication module, a Zigbee communication module and a data storage unit; the power supply module is respectively connected with a solar power generation module and a storage battery; the input end of the information processing unit is connected with a magnetic balance type Hall current sensor, and the input end of the magnetic balance type Hall current sensor is connected with a circuit breaker opening and closing coil.
7. The RST-PNN-GA-based high-voltage circuit breaker fault detection method of claim 6, wherein: the model of the single chip microcomputer is STM32F 407.
8. The RST-PNN-GA-based high-voltage circuit breaker fault detection method of claim 3, wherein 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 VaIs the value range of attribute a, f: u (C ^ D) → V is an information decision function;
for equivalence relations, any subset of attributes has
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
There is an upper and lower approximation for B, which are described as respectively
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
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)；
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,
the dependency of Q on P can be defined as
k＝γP(Q)＝|POSP(Q)|/|U|；
For any subset of conditional attributes
If:
POSR(D)＝POSC(D)；
POSR(D)≠POSR-{a}(D)，a∈R；
r is a relative reduction of C.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110649592.XA CN113469222A (en) | 2021-06-10 | 2021-06-10 | RST-PNN-GA-based high-voltage circuit breaker fault detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110649592.XA CN113469222A (en) | 2021-06-10 | 2021-06-10 | RST-PNN-GA-based high-voltage circuit breaker fault detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113469222A true CN113469222A (en) | 2021-10-01 |
Family
ID=77869591
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110649592.XA Pending CN113469222A (en) | 2021-06-10 | 2021-06-10 | RST-PNN-GA-based high-voltage circuit breaker fault detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113469222A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107019501A (en) * | 2017-05-05 | 2017-08-08 | 山东师范大学 | Detection method and system are fallen down based on genetic algorithm and the long-range of probabilistic neural network |
| 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 |
| WO2019216975A1 (en) * | 2018-05-07 | 2019-11-14 | Strong Force Iot Portfolio 2016, Llc | Methods and systems for data collection, learning, and streaming of machine signals for analytics and maintenance using the industrial internet of things |
| 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 |
-
2021
- 2021-06-10 CN CN202110649592.XA patent/CN113469222A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107019501A (en) * | 2017-05-05 | 2017-08-08 | 山东师范大学 | Detection method and system are fallen down based on genetic algorithm and the long-range of probabilistic neural network |
| 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 |
| WO2019216975A1 (en) * | 2018-05-07 | 2019-11-14 | Strong Force Iot Portfolio 2016, Llc | Methods and systems for data collection, learning, and streaming of machine signals for analytics and maintenance using the industrial internet of things |
| 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 |
Non-Patent Citations (3)
| Title |
|---|
| YIMING HUANG 等: "EMD- PNN based welding defects detection using laser-induced plasma electrical signals", 《JOURNAL OF MANUFACTURING PROCESSES》 * |
| 宋爱香 等: "一种基于BP-PNN的图书信息化管理模型", 《中国科技信息》 * |
| 张浩: "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》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109270442B (en) | DBN-GA neural network-based high-voltage circuit breaker fault detection method | |
| CN106291351B (en) | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm | |
| CN111060815B (en) | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method | |
| Roubos et al. | Compact and transparent fuzzy models and classifiers through iterative complexity reduction | |
| CN109086817A (en) | A kind of Fault Diagnosis for HV Circuit Breakers method based on deepness belief network | |
| CN110221200A (en) | A kind of omnipotent breaker attachment method for diagnosing faults based on deep learning | |
| CN109800875A (en) | Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine | |
| CN109298330B (en) | High-voltage circuit breaker fault diagnosis method based on GHPSO-BP | |
| CN106019138A (en) | Online diagnosis method for mechanical fault of high-voltage circuit breaker | |
| CN113283602B (en) | Offline knowledge learning and online dynamic correction-based power grid system fault diagnosis method | |
| CN114661905A (en) | Power grid fault diagnosis method based on BERT | |
| CN109901064B (en) | ICA-LVQ-based high-voltage circuit breaker fault diagnosis method | |
| CN116205265A (en) | Power grid fault diagnosis method and device based on deep neural network | |
| CN112464556A (en) | AC contactor electric service life prediction method based on long-short term memory neural network | |
| CN116050888A (en) | Method applied to intelligent high-voltage switch cabinet sensor health state assessment | |
| CN113326663A (en) | Railway relay running state evaluation method based on extreme learning machine | |
| Chen et al. | Remaining useful life prediction of turbofan engine based on temporal convolutional networks optimized by genetic algorithm | |
| CN113469222A (en) | RST-PNN-GA-based high-voltage circuit breaker fault detection method | |
| CN114839523A (en) | Circuit breaker fault detection system and method based on CNN combined LSTM algorithm | |
| Qian et al. | Fault diagnosis for generator unit based on RBF neural network optimized by GA-PSO | |
| CN115456106A (en) | High-voltage circuit breaker fault diagnosis model optimization method | |
| Gao et al. | Fault detection of electric vehicle charging piles based on extreme learning machine algorithm | |
| CN114625831A (en) | Classification evaluation feedback method for load identification of smart power grid | |
| CN112598186A (en) | Improved LSTM-MLP-based small generator fault prediction method | |
| CN116582450B (en) | Service access sensing method and system for low-voltage distribution network |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211001 |
|
| RJ01 | Rejection of invention patent application after publication |