CN108734202A - A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP - Google Patents
A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP Download PDFInfo
- Publication number
- CN108734202A CN108734202A CN201810390778.6A CN201810390778A CN108734202A CN 108734202 A CN108734202 A CN 108734202A CN 201810390778 A CN201810390778 A CN 201810390778A CN 108734202 A CN108734202 A CN 108734202A
- Authority
- CN
- China
- Prior art keywords
- neural network
- algorithm
- particle
- breeding
- sample
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The Fault Diagnosis for HV Circuit Breakers method based on improved BP that the invention discloses a kind of, specially, training sample and test sample are divided into sample of the high-voltage circuitbreaker with class label acquired, then the BP neural network model based on Breeding Algorithm and particle cluster algorithm is established, and after being trained using training sample, decoding generates new connection weight and threshold value;It is controlled using iteration controller, so that two kinds of algorithms is carried out information exchange every number generation, the content of information exchange is the relevant information of optimal particle seed, and obtains optimal global parameter;Obtained globally optimal solution is replaced to all weights and threshold parameter of original BP neural network after decoding, establishes the high-voltage circuitbreaker fault model after optimization, failure modes are carried out to test sample, and export result.The method of the present invention carrys out the connection weight and threshold value of Optimized BP Neural Network with BA and PSO algorithms instead of the network learning procedure of error-duration model, effectively improves fault diagnosis precision.
Description
Technical field
The invention belongs to Fault Diagnosis for HV Circuit Breakers method and technology field, it is related to a kind of based on improved BP
Fault Diagnosis for HV Circuit Breakers method.
Background technology
High-voltage circuitbreaker plays dual parts of control and protection in distribution network system, and the quality of operation conditions is directly
Decide the operation of entire electric system.Therefore, significant to high-voltage circuitbreaker progress fault diagnosis.It has been proposed at present
A variety of diagnostic methods are directed to various intelligent algorithms, such as:Fuzzy control and radial base neural net etc..It is wherein fuzzy
Control can be with accurate mathematical tool by fuzzy concept or natural language sharpening, but its membership function and fuzzy rule are really
Determining process, there are certain human factors;Radial base neural net provides a kind of relatively good structural system, but there is nothings
The shortcomings that neural network can not work normally when the reasoning process of method interpretation oneself and reasoning foundation and insufficient data.
BP neural network in artificial neural network is simple in structure, and plasticity is strong, be in neural network it is the most commonly used, most have
Effect, a kind of most active model.But there is also local minimum is easily trapped into, the shortcomings of network convergence rate is slower.
Invention content
The Fault Diagnosis for HV Circuit Breakers method based on improved BP that the object of the present invention is to provide a kind of, with BA
The connection weight and threshold value for carrying out Optimized BP Neural Network instead of the network learning procedure of error-duration model with PSO algorithms effectively improve event
Hinder diagnostic accuracy.
The technical solution adopted in the present invention is a kind of Fault Diagnosis for HV Circuit Breakers side based on improved BP
Method, which is characterized in that be specifically implemented according to the following steps:
Step 1, every to sample of the high-voltage circuitbreaker with class label acquired a kind of by 3:1 ratio is divided into training sample
And test sample;
Step 2, the BP neural network model based on Breeding Algorithm and particle cluster algorithm is established.And it is carried out using training sample
After training, decoding generates new connection weight and threshold value;
Step 3, it is controlled using iteration controller, two kinds of algorithms is made to carry out information exchange, information exchange every number generation
Content be the relevant information of optimal particle seed, and obtain optimal global parameter;
Step 4, by the globally optimal solution that step 3 obtains replaced after decoding original BP neural network all weights and
Threshold parameter establishes the high-voltage circuitbreaker fault model after optimization, carries out failure modes to test sample, and export result.
The features of the present invention also characterized in that
The step 1 is specially:Sample set the S={ (x of class label are carried to collected high-voltage circuitbreaker1,y1),
(x2,y2),...,(xn,yn) every a kind of by 3:1 ratio is divided into training sample and test sample;Wherein xiRepresent i-th of sample category
Property,The data of current waveform extraction when sample is switching on and off on coil, wherein including I1,I2,I3,t1,t2,
t3,t4,t5Eight attribute are as mode input, yiThe fault category of i-th of sample is represented, the fault category is respectively:Point
Closing operation brownout, iron core operation have bite, operating mechanism to have bite, auxiliary switch action poor contact, mechanism normal,
This 5 kinds of fault categories correspond to faulty tag 1,2,3,4,5 respectively.
The step 2 is specifically implemented according to the following steps:
Step 2.1, PSO-BP models are established,
Step 2.2, BA-BP models are established.
The step 2.1 is specifically implemented according to the following steps,
Step 2.1.1, initiation parameter determine BP neural network topological structure.Select three layers of BP neural network, input layer
For 8 nodes, output layer is 5 nodes, and rule of thumb formula primarily determines that hidden layer is 12 nodes,
Step 2.1.2, with the parameter of particle cluster algorithm optimization neural network,
Step 2.1.3, the training sample that will be obtained in step 1 input the defeated of BP neural network after normalized
Enter layer, carries out Training, calculated according to the calculation formula (1.1) of the forwards algorithms of neural network and mean square error index
Each particle fitness function value, wherein fitness function value indicate the error threshold of neural network, and error is smaller to show particle
Performance in the search is better, and using the desired positions of each particle as its history optimum position, starts iteration;
In formula:N is number of training;yi dIt is the idea output of i-th group of sample;yiIt is the reality output of i-th group of sample
Value, E is fitness value.
Step 2.1.4 is updated the speed of particle and position using the formula of PSO algorithms, and considers updated
Whether speed and position are in limited range;
Step 2.1.5 checks whether to meet termination condition.If current location or maximum iteration reach scheduled mistake
When difference requires, algorithmic statement, the result of last time iteration is the weights and threshold value of required global optimum;Otherwise step is returned
Rapid 2.1.3, algorithm continue iteration.
The core iterative formula of PSO algorithms is as follows in the step 2.1.4:
Wherein, vi k+1Indicate i-th of particle+1 generation of kth flying speed;xi k+1Indicate i-th of particle k+1 for when
Position;pi kIndicate i-th of particle to kth on behalf of only found optimal location;pg kIndicate up to the present current population is looked for
The optimal location arrived;pi k-xi kIndicate individual cognition;pg k-xi kIndicate social recognition;ω is inertia coeffeicent, believes oneself
Degree;Accelerated factor c1And c2It is respectively regulated to the maximum step-length of the directions pbest and ghest flight, suitable c1And c2It can add
It rapid convergence and is not easy to be absorbed in local optimum, usually takes c1=c2=2.r1And r2Indicate the random number in (0,1).
The step 2.2 is specifically implemented according to the following steps:
Step 2.2.1, initiation parameter determine BP neural network topological structure.Neural network structure and PSO-BP herein
Model structure is consistent, and it is 8 nodes equally to select three-layer neural network, input layer, and output layer is 5 nodes, rule of thumb public
Formula primarily determines that hidden layer is 12 nodes.The population scale of Breeding Algorithm takes 30, learning coefficient 0.15, maximum iteration
It is 2000;
Step 2.2.2, to all Optimal Parameters weights and threshold value, in its restriction range, using binary coding method
It is random to generate initial population.In the BP Learning Algorithms based on Breeding Algorithm, every one-dimensional representation BP god in Seeding vector
Value through weights in network of network or threshold value;
Step 2.2.3 inputs the input layer of BP neural network by the training sample obtained in step 1 after normalization, leads to
It crosses BP neural network forwards algorithms and calculates each seed fitness function value.Then, optimal population is searched for, following mean square deviation is made
Index is minimum:
In formula:N is number of training;yi dIt is the idea output of i-th group of sample;yiIt is the reality output of i-th group of sample
Value, EgFor the adaptive optimal control value in m population.
Step 2.2.4 selects the maximum individual of adaptive value to carry out rotaring gene breeding as parent.In reproductive process, at random
Displacement, the digit for generating variation, to generate new gene, and are replaced.Generate next-generation population.For example, to seed with
The change point that machine generates is 3, and the variation digit randomly generated is 2, then the new gene randomly generated is 23;
Step 2.2.5, checks whether and meets iterated conditional, such as reaches preset maximum iteration or adaptive value consecutive numbers
Dai Wei changes, then terminates to calculate, and exports result.Otherwise return to step 2.2.3 continues.
The step 3 is specially:
Step 3.1, parameter initialization:
If G:Total iterations;g1:The iteration controller of particle cluster algorithm;g2:The iteration controller of Breeding Algorithm;Δ
g1:The algebraically interval of the information exchange of particle cluster algorithm, the i.e. inverse of the frequency of interaction of particle cluster algorithm;Δg2:Breeding Algorithm
The algebraically interval of information exchange, the i.e. inverse of the frequency of interaction of Breeding Algorithm;W:Times that weight coefficient, i.e. frequency of interaction reduce
Number;
Step 3.2, it scans for, distributes, individual evaluation, information exchange picks out optimum individual;
If:When overall iteration controller g is less than or equal to G,
If:g1It cannot be by Δ g1Divide exactly
Particle cluster algorithm is once searched for;
g1+=1;
If:g2It cannot be by Δ g2Divide exactly
Breeding group algorithm is once searched for;
g2+=1
Otherwise
If:Particle cluster algorithm result is better than Breeding Algorithm
Using the optimal solution of particle cluster algorithm as the seed of Breeding Algorithm;
Δg1=(int) (Δ g1/W)
Otherwise, if:Breeding Algorithm result is better than particle cluster algorithm
Optimal particle in particle cluster algorithm is replaced with to the optimal seed of Breeding Algorithm,
And change the optimal particle information of particle cluster algorithm;
Δg2=(int) (Δ g2/W)
If:The optimal value variation that two kinds of algorithms of continuous 5 iteration generate is little
Δg1=(int) (Δ g1/W);Δg2=(int) (Δ g2/W);
Step 3.3, output makes two kinds of particles carry out information exchange as a result, by iteration controller, and one kind is standard PSO
Particle, one kind are breeding particle, standard PSO particles conventionally more new route, but all extreme values are in two kinds of particles
Most the superior, breeding particle generate new particle according to all optimal particles according to the breeding operation of Breeding Algorithm, and constantly cycle changes
Generation, until meeting stop condition.And export globally optimal solution.
The step 4 is specifically implemented according to the following steps,
Step 4.1, initiation parameter determines the BP neural network topological structure for last diagnostic.Neural network herein
Structure and the PSO-BP in step 2 and step 3 are consistent with BA-BP model structures, equally select the three-layer neural network, input layer to be
8 nodes, output layer are 5 nodes, and hidden layer is 12 nodes, after study to certain number, if specification error is not achieved
Then increase and decrease the number of hidden layer node on the basis of initial value;
Step 4.2, the optimal solution obtained after particle cluster algorithm and Breeding Algorithm mixed iteration in step 2 is decoded
It is substituted into afterwards in the BP neural network obtained in step 1), as the weights and threshold parameter of BP neural network, obtains being based on BP
The Fault Diagnosis for HV Circuit Breakers model of neural network;
Step 4.3, failure modes are carried out to the test sample in step 1, i.e., by the test sample normalizing of gained in step 1
Change between [0,1], and is input to the BP god after parameter optimization that is obtained in step 4.2 by network fault diagnosis model
In.Failure modes are obtained as a result, simultaneously statistical model accuracy rate of diagnosis.
The invention has the advantages that
1. Breeding Algorithm is combined with particle cluster algorithm, effectively directs and how to find the BP for adapting to specific sample data god
Through network parameter (weights and threshold value), has the advantages that fast convergence rate and be not easy to be absorbed in local optimum.
2. optimizing promotion to BP neural network using particle cluster algorithm herein, declines instead of traditional gradient and calculate
Method.BP neural network after optimization has many advantages, such as that the training time is short, computational accuracy is high.
3. substituting the parameter that the gradient descent method in BP algorithm trains neural network with Breeding Algorithm, gradient decline is avoided
To be found a function in method can process micro-, to function derivation;Also avoid selection, intersection, the variation etc. in common genetic algorithm
Evolutional operation shortens the training time of neural network.
4. being controlled using iteration controller, two kinds of algorithms alternately, every number generation carry out information exchange, this algorithm
Gather two kinds of algorithms advantage, complementarity and mutual benefit.The accuracy rate of failure modes can effectively be promoted.
Description of the drawings
The present invention is based on the flow charts of the Fault Diagnosis for HV Circuit Breakers method of improved BP by Fig. 1;
The present invention is based on the neural network models of the Fault Diagnosis for HV Circuit Breakers method of improved BP by Fig. 2;
The present invention is based on seminal propagations in the Fault Diagnosis for HV Circuit Breakers method and step 3 of improved BP to grasp by Fig. 3
Make process;
When Fig. 4 (a) the present invention is based on the Fault Diagnosis for HV Circuit Breakers method of improved BP using being tested
Coil current waveform schematic diagram when separating brake;
When Fig. 4 (b) the present invention is based on the Fault Diagnosis for HV Circuit Breakers method of improved BP using being tested
Coil current waveform schematic diagram when combined floodgate.
Specific implementation mode
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP, as shown in Figure 1, specifically according to following
Step is implemented:
Step 1, every to sample of the high-voltage circuitbreaker with class label acquired a kind of by 3:1 ratio is divided into training sample
And test sample, it is specifically implemented according to the following steps:
Sample set the S={ (x of class label are carried to collected high-voltage circuitbreaker1,y1),(x2,y2),...,(xn,yn)}
3 are pressed per a kind of:1 ratio is divided into training sample and test sample.Wherein xiI-th of sample attribute is represented,Sample is
The data of current waveform extraction when switching on and off on coil, wherein including I1,I2,I3,t1,t2,t3,t4,t58 attribute conducts
Mode input, as shown in figure 4, yiThe fault category of i-th of sample is represented, the fault category is respectively:Breaking-closing operating
Brownout, iron core operation have bite, operating mechanism to have bite, auxiliary switch action poor contact, mechanism normal, this 5 kinds events
Barrier classification corresponds to faulty tag 1,2,3,4,5 respectively;
Step 2:The BP neural network model based on Breeding Algorithm and particle cluster algorithm is established, as shown in Fig. 2, and utilizing instruction
After white silk sample is trained, decoding generates new connection weight and threshold value;The specific steps are:
Step 2.1, PSO-BP models are established, specific establishment step is as follows:
Step 2.1.1, initiation parameter determine BP neural network topological structure.Select three layers of BP neural network, input layer
For 8 nodes, output layer is 5 nodes, and rule of thumb formula primarily determines that hidden layer is 12 nodes.
Step 2.1.2, with the parameter of particle cluster algorithm optimization neural network,
By the position x of particlei=(xi1,xi2,…xid,…xiD), speed vi=(vi1,vi2,…vid,…viD) carry out initially
Change, the random number between [- 1,1] is initialized to per one-dimensional parameter;Determine the minimum value and maximum of particle number, inertia weight
Value, accelerator coefficient, iterations etc..Wherein:xiFor the position of current particle, what its every one-dimensional parameter list showed is BP nerve nets
Weights inside network model and threshold value;viFor the speed of current particle;I is current particle label;D is the label of digit, and D is total
Dimension;The population of initial setting population is 30, accelerated factor c1=c2=2, maximum iteration 2000.
Step 2.1.3, the training sample that will be obtained in step 1 input the defeated of BP neural network after normalized
Enter layer, carries out Training, calculated according to the calculation formula (1.1) of the forwards algorithms of neural network and mean square error index
Each particle fitness function value, wherein fitness function value indicate the error threshold of neural network, and error is smaller to show particle
Performance in the search is better, and using the desired positions of each particle as its history optimum position, starts iteration;
In formula:N is number of training;yi dIt is the idea output of i-th group of sample;yiIt is the reality output of i-th group of sample
Value, E is fitness value.
Step 2.1.4 is updated the speed of particle and position using the formula (1.2) of PSO algorithms, and considers to update
Whether rear speed and position are in limited range;
The core iterative formula of PSO algorithms is as follows:
Wherein, vi k+1Indicate i-th of particle+1 generation of kth flying speed;xi k+1Indicate i-th of particle k+1 for when
Position;pi kIndicate i-th of particle to kth on behalf of only found optimal location;pg kIndicate up to the present current population is looked for
The optimal location arrived;pi k-xi kIndicate individual cognition;pg k-xi kIndicate social recognition;ω is inertia coeffeicent, believes oneself
Degree;Accelerated factor c1And c2It is respectively regulated to the maximum step-length of the directions pbest and ghest flight, suitable c1And c2It can add
It rapid convergence and is not easy to be absorbed in local optimum, usually takes c1=c2=2.r1And r2Indicate the random number in (0,1).
Step 2.1.5 checks whether to meet termination condition.If current location or maximum iteration reach scheduled mistake
When difference requires, algorithmic statement, the result of last time iteration is the weights and threshold value of required global optimum;Otherwise step is returned
Rapid 2.1.3, algorithm continue iteration.
Step 2.2, BA-BP models are established, specific establishment step is as follows:
Step 2.2.1, initiation parameter determine BP neural network topological structure.Neural network structure and PSO-BP herein
Model structure is consistent, and it is 8 nodes equally to select three-layer neural network, input layer, and output layer is 5 nodes, rule of thumb public
Formula primarily determines that hidden layer is 12 nodes.The population scale of Breeding Algorithm takes 30, learning coefficient 0.15, maximum iteration
It is 2000.
Step 2.2.2, to all Optimal Parameters weights and threshold value, in its restriction range, using binary coding method
It is random to generate initial population.In the BP Learning Algorithms based on Breeding Algorithm, every one-dimensional representation BP god in Seeding vector
Value through weights in network of network or threshold value;
Step 2.2.3 inputs the input layer of BP neural network by the training sample obtained in step 1 after normalization, leads to
It crosses BP neural network forwards algorithms and calculates each seed fitness function value.Then, optimal population is searched for, following mean square deviation is made
Index (adaptive value) is minimum:
In formula:N is number of training;yi dIt is the idea output of i-th group of sample;yiIt is the reality output of i-th group of sample
Value, EgFor the adaptive optimal control value in m population.
Step 2.2.4 selects the maximum individual of adaptive value to carry out rotaring gene breeding as parent.In reproductive process, at random
Displacement, the digit for generating variation, to generate new gene, and are replaced.Generate next-generation population.For example, to seed with
The change point that machine generates is 3, and the variation digit randomly generated is 2, then the new gene randomly generated is 23, reproductive process and result
As shown in Figure 3.
Step 2.2.5, checks whether and meets iterated conditional, such as reaches preset maximum iteration or adaptive value consecutive numbers
Dai Wei changes, then terminates to calculate, and exports result.Otherwise return to step 2.2.3 continues.
Step 3:It is controlled using iteration controller, two kinds of algorithms is made to carry out information exchange, information exchange every number generation
Content be the relevant information of optimal particle seed, and obtain optimal global parameter.
Step 3.1, parameter initialization:If G:Total iterations;g1:The iteration controller of particle cluster algorithm;g2:Breeding
The iteration controller of algorithm;Δg1:Algebraically interval (the i.e. frequency of interaction of particle cluster algorithm of the information exchange of particle cluster algorithm
It is reciprocal);Δg2:The algebraically interval (i.e. the inverse of the frequency of interaction of Breeding Algorithm) of the information exchange of Breeding Algorithm;W:Weight system
Number (multiple that i.e. frequency of interaction reduces);
Step 3.2, it scanning for, distributes, individual evaluation, information exchange picks out optimum individual,
If:When overall iteration controller g is less than or equal to G,
If:g1It cannot be by Δ g1Divide exactly
Particle cluster algorithm is once searched for;
g1+=1;
If:g2It cannot be by Δ g2Divide exactly
Breeding group algorithm is once searched for;
g2+=1
Otherwise
If:Particle cluster algorithm result is better than Breeding Algorithm
Using the optimal solution of particle cluster algorithm as the seed of Breeding Algorithm;
Δg1=(int) (Δ g1/W)
Otherwise, if:Breeding Algorithm result is better than particle cluster algorithm
Optimal particle in particle cluster algorithm is replaced with to the optimal seed of Breeding Algorithm, and changes particle cluster algorithm most
Excellent particle information;
Δg2=(int) (Δ g2/W)
If:The optimal value variation that two kinds of algorithms of continuous 5 iteration generate is little
Δg1=(int) (Δ g1/W);Δg2=(int) (Δ g2/W);
Step 3.3, result is exported.By iteration controller, two kinds of particles is made to carry out information exchange, one kind is standard PSO
Particle, one kind are breeding particle, standard PSO particles conventionally more new route, but all extreme values are in two kinds of particles
Most the superior, breeding particle generate new particle according to all optimal particles according to the breeding operation of Breeding Algorithm, and constantly cycle changes
Generation, until meeting stop condition.And export globally optimal solution.
Step 4:By the globally optimal solution that step 3 obtains replaced after decoding original BP neural network all weights and
Threshold parameter establishes the high-voltage circuitbreaker fault model after optimization, carries out failure modes to test sample, and export result.Tool
Steps are as follows for body:
Step 4.1, initiation parameter determines the BP neural network topological structure for last diagnostic.Neural network herein
Structure and the PSO-BP in step 2 and step 3 are consistent with BA-BP model structures, equally select the three-layer neural network, input layer to be
8 nodes, output layer are 5 nodes, and hidden layer is 12 nodes, after study to certain number, if specification error is not achieved
Then increase and decrease the number of hidden layer node on the basis of initial value.
Step 4.2, the optimal solution obtained after particle cluster algorithm and Breeding Algorithm mixed iteration in step 2 is decoded
It is substituted into afterwards in the BP neural network obtained in step 4.1, as the weights and threshold parameter of BP neural network, obtains being based on BP
The Fault Diagnosis for HV Circuit Breakers model of neural network.
Step 4.3, failure modes are carried out to the test sample in step 1, i.e., by the test sample normalizing of gained in step 1
Change between [0,1], and is input to the BP god after parameter optimization that is obtained in step 4.2 by network fault diagnosis model
In.Failure modes are obtained as a result, simultaneously statistical model accuracy rate of diagnosis.
Do each 400 times of switching on and off experiment altogether herein, it is random to select according to the data sample that simulation test data obtain
Each 300 test datas are tested with switching on and off to be trained hybrid algorithm model, and each 100 are tested with remaining switching on and off
Secondary test data is verified.Show that the hybrid algorithm can overcome precocious phenomenon, local search ability by test result
Also it makes moderate progress, it is easier to find globally optimal solution, effectively improve the accuracy rate of failure modes.
Particle cluster algorithm (particle swarm optimization, PSO) has the characteristics that fast convergence rate, and educates
Kind algorithm (breeding algorithm, BA) is not easy to be absorbed in local optimum.After the two combines, which can be by Breeding Algorithm
Variation mechanism is introduced into particle cluster algorithm, when the iteration point of hybrid algorithm is absorbed in locally optimal solution, passes through Breeding Algorithm
Variation mechanism (jump search capability), the optimal solution that particle cluster algorithm is searched is bred into row variation, to expand search
Range may finally make search particle jump out current locally optimal solution, and find globally optimal solution, that is, find connection weight and threshold
Value.
The Fault Diagnosis for HV Circuit Breakers method based on improved BP of the present invention, for particle swarm optimization algorithm
In practical applications in place of existing problem and shortage, it is put forward for the first time particle cluster algorithm and Breeding Algorithm combination, structure
Double population search mechanisms are not only utilized the rapid evolution ability of PSO algorithms, but also are increased using the breeding operation in Breeding Algorithm model
Population diversity, while the iteration controller module proposed can merge the respective advantage of two kinds of algorithms well, and wherein
Be added iteration frequency concept, solve the problems, such as iteration inefficiency, it is proposed that breeding particle swarm optimization algorithm to
Optimize the Fault Diagnosis for HV Circuit Breakers model based on BP neural network.
Claims (8)
1. a kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP, which is characterized in that specifically according to following
Step is implemented:
Step 1, every to sample of the high-voltage circuitbreaker with class label acquired a kind of by 3:1 ratio is divided into training sample and survey
Sample sheet,
Step 2, the BP neural network model based on Breeding Algorithm and particle cluster algorithm is established;And it is trained using training sample
Afterwards, decoding generates new connection weight and threshold value;
Step 3, controlled using iteration controller, make two kinds of algorithms every number generation carry out information exchanges, information exchange it is interior
Appearance is the relevant information of optimal particle seed, and obtains optimal global parameter;
Step 4, the globally optimal solution that step 3 obtains is replaced to all weights and threshold value of original BP neural network after decoding
Parameter establishes the high-voltage circuitbreaker fault model after optimization, carries out failure modes to test sample, and export result.
2. the Fault Diagnosis for HV Circuit Breakers method according to claim 1 based on improved BP, feature exist
In the step 1 is specially:Sample set the S={ (x of class label are carried to collected high-voltage circuitbreaker1,y1),(x2,
y2),...,(xn,yn) every a kind of by 3:1 ratio is divided into training sample and test sample;Wherein xiI-th of sample attribute is represented,The data of current waveform extraction when sample is switching on and off on coil, wherein including I1,I2,I3,t1,t2,t3,
t4,t5Eight attribute are as mode input, yiThe fault category of i-th of sample is represented, the fault category is respectively:Division
Lock operation brownout, iron core operation have bite, operating mechanism to have bite, auxiliary switch action poor contact, mechanism normal, this
5 kinds of fault categories correspond to faulty tag 1,2,3,4,5 respectively.
3. the Fault Diagnosis for HV Circuit Breakers method according to claim 1 based on improved BP, feature exist
In the step 2 is specifically implemented according to the following steps:
Step 2.1, PSO-BP models are established,
Step 2.2, BA-BP models are established.
4. the Fault Diagnosis for HV Circuit Breakers method according to claim 3 based on improved BP, feature exist
In, the step 2.1 is specifically implemented according to the following steps,
Step 2.1.1, initiation parameter determine BP neural network topological structure;Select three layers of BP neural network, input layer 8
A node, output layer are 5 nodes, and rule of thumb formula primarily determines that hidden layer is 12 nodes,
Step 2.1.2, with the parameter of particle cluster algorithm optimization neural network,
Step 2.1.3, the training sample that will be obtained in step 1 input the input layer of BP neural network after normalized,
Training is carried out, is calculated according to the calculation formula (1.1) of the forwards algorithms of neural network and mean square error index each
Particle fitness function value, wherein fitness function value indicate the error threshold of neural network, and error is smaller to show that particle is being searched
Performance in rope is better, and using the desired positions of each particle as its history optimum position, starts iteration;
In formula:N is number of training;yi dIt is the idea output of i-th group of sample;yiIt is the real output value of i-th group of sample, E
For fitness value;
Step 2.1.4 is updated the speed of particle and position using the formula of PSO algorithms, and considers updated speed
With position whether in limited range;
Step 2.1.5 checks whether to meet termination condition;It is wanted if current location or maximum iteration reach scheduled error
When asking, algorithmic statement, the result of last time iteration is the weights and threshold value of required global optimum;Otherwise return to step
2.1.3, algorithm continues iteration.
5. the Fault Diagnosis for HV Circuit Breakers method according to claim 4 based on improved BP, feature exist
In the core iterative formula of PSO algorithms is as follows in the step 2.1.4:
Wherein, vi k+1Indicate i-th of particle+1 generation of kth flying speed;xi k+1Indicate i-th of particle k+1 for when position
It sets;pi kIndicate i-th of particle to kth on behalf of only found optimal location;pg kIndicate up to the present current population is found
Optimal location;pi k-xi kIndicate individual cognition;pg k-xi kIndicate social recognition;ω is inertia coeffeicent, believes the journey of oneself
Degree;Accelerated factor c1And c2It is respectively regulated to the maximum step-length of the directions pbest and ghest flight, suitable c1And c2It can accelerate
Local optimum is restrained and be not easy to be absorbed in, c is usually taken1=c2=2;r1And r2Indicate the random number in (0,1).
6. the Fault Diagnosis for HV Circuit Breakers method according to claim 3 based on improved BP, feature exist
In the step 2.2 is specifically implemented according to the following steps:
Step 2.2.1, initiation parameter determine BP neural network topological structure;Neural network structure and PSO-BP models herein
Structure is consistent, and it is 8 nodes equally to select three-layer neural network, input layer, and output layer is 5 nodes, rule of thumb at the beginning of formula
Step determines that hidden layer is 12 nodes;The population scale of Breeding Algorithm takes 30, learning coefficient 0.15, and maximum iteration is
2000;
Step 2.2.2, it is random using binary coding method in its restriction range to all Optimal Parameters weights and threshold value
Generate initial population;In the BP Learning Algorithms based on Breeding Algorithm, every one-dimensional representation BP nerve nets in Seeding vector
The value of weights or threshold value in network diagram network;
Step 2.2.3 inputs the input layer of BP neural network, passes through BP by the training sample obtained in step 1 after normalization
Neural network forwards algorithms calculate each seed fitness function value;Then, optimal population is searched for, following mean square deviation index is made
It is minimum:
In formula:N is number of training;yi dIt is the idea output of i-th group of sample;yiIt is the real output value of i-th group of sample, Eg
For the adaptive optimal control value in m population;
Step 2.2.4 selects the maximum individual of adaptive value to carry out rotaring gene breeding as parent;In reproductive process, randomly generate
The displacement of variation, digit to generate new gene, and are replaced;Generate next-generation population;For example, being produced at random to seed
Raw change point is 3, and the variation digit randomly generated is 2, then the new gene randomly generated is 23;
Step 2.2.5, checks whether and meets iterated conditional, such as reaches preset maximum iteration or adaptive value consecutive numbers generation not
It changes, then terminates to calculate, export result;Otherwise return to step 2.2.3 continues.
7. the Fault Diagnosis for HV Circuit Breakers method according to claim 1 based on improved BP, feature exist
In the step 3 is specially:
Step 3.1, parameter initialization:
If G:Total iterations;g1:The iteration controller of particle cluster algorithm;g2:The iteration controller of Breeding Algorithm;Δg1:Grain
The algebraically interval of the information exchange of swarm optimization, the i.e. inverse of the frequency of interaction of particle cluster algorithm;Δg2:The information of Breeding Algorithm
Interactive algebraically interval, the i.e. inverse of the frequency of interaction of Breeding Algorithm;W:The multiple that weight coefficient, i.e. frequency of interaction reduce;
Step 3.2, it scans for, distributes, individual evaluation, information exchange picks out optimum individual;
If:When overall iteration controller g is less than or equal to G,
If:g1It cannot be by Δ g1Divide exactly
Particle cluster algorithm is once searched for;
g1+=1;
If:g2It cannot be by Δ g2Divide exactly
Breeding group algorithm is once searched for;
g2+=1
Otherwise
If:Particle cluster algorithm result is better than Breeding Algorithm
Using the optimal solution of particle cluster algorithm as the seed of Breeding Algorithm;
Δg1=(int) (Δ g1/W)
Otherwise, if:Breeding Algorithm result is better than particle cluster algorithm
Optimal particle in particle cluster algorithm is replaced with to the optimal seed of Breeding Algorithm, and changes the optimal grain of particle cluster algorithm
Sub-information;
Δg2=(int) (Δ g2/W)
If:The optimal value variation that two kinds of algorithms of continuous 5 iteration generate is little
Δg1=(int) (Δ g1/W);Δg2=(int) (Δ g2/W);
Step 3.3, output makes two kinds of particles carry out information exchange as a result, by iteration controller, and one kind is standard PSO particles,
One kind is breeding particle, standard PSO particles conventionally more new route, but all extreme values are optimal in two kinds of particles
Person, breeding particle operate according to the breeding of Breeding Algorithm according to all optimal particles and generate new particle, continuous loop iteration, directly
To meeting stop condition;And export globally optimal solution.
8. the Fault Diagnosis for HV Circuit Breakers method according to claim 1 based on improved BP, feature exist
In, the step 4 is specifically implemented according to the following steps,
Step 4.1, initiation parameter determines the BP neural network topological structure for last diagnostic;Neural network structure herein
Consistent with BA-BP model structures with the PSO-BP in step 2 and step 3, it is 8 equally to select three-layer neural network, input layer
Node, output layer are 5 nodes, and hidden layer is 12 nodes, study to after certain number, if specification error is not achieved
Increase and decrease the number of hidden layer node on the basis of initial value;
Step 4.2, by the optimal solution obtained after particle cluster algorithm and Breeding Algorithm mixed iteration in step 2 it is decoded after replace
It changes in the BP neural network obtained in step 4.1, as the weights and threshold parameter of BP neural network, obtains based on BP nerves
The Fault Diagnosis for HV Circuit Breakers model of network;
Step 4.3, failure modes are carried out to the test sample in step 1, i.e., normalized to the test sample of gained in step 1
Between [0,1], and being input to the BP god after parameter optimization obtained in step 4.2 will be in network fault diagnosis model;?
Be out of order classification results, and statistical model accuracy rate of diagnosis.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810390778.6A CN108734202A (en) | 2018-04-27 | 2018-04-27 | A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810390778.6A CN108734202A (en) | 2018-04-27 | 2018-04-27 | A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108734202A true CN108734202A (en) | 2018-11-02 |
Family
ID=63939891
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810390778.6A Pending CN108734202A (en) | 2018-04-27 | 2018-04-27 | A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108734202A (en) |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109298330A (en) * | 2018-11-26 | 2019-02-01 | 西安工程大学 | Fault Diagnosis for HV Circuit Breakers method based on GHPSO-BP |
CN109492816A (en) * | 2018-11-16 | 2019-03-19 | 辽宁工程技术大学 | A kind of coal and gas prominent dynamic prediction method based on hybrid intelligent |
CN109558677A (en) * | 2018-11-29 | 2019-04-02 | 东北大学 | A kind of hot rolling strip crown prediction technique based on data-driven |
CN109856969A (en) * | 2018-11-06 | 2019-06-07 | 皖西学院 | A kind of failure prediction method and forecasting system based on BP neural network model |
CN109948194A (en) * | 2019-02-27 | 2019-06-28 | 北京航空航天大学 | A kind of high-voltage circuitbreaker mechanical defect integrated study diagnostic method |
CN110210632A (en) * | 2019-05-31 | 2019-09-06 | 国网河北省电力有限公司沧州供电分公司 | Failure emergency processing method, device and terminal based on ubiquitous electric power Internet of Things |
CN110232372A (en) * | 2019-06-26 | 2019-09-13 | 电子科技大学成都学院 | Gait recognition method based on particle group optimizing BP neural network |
CN110263907A (en) * | 2019-07-01 | 2019-09-20 | 上海海事大学 | Based on the ship short trouble diagnostic method for improving GA-PSO-BP |
CN110648315A (en) * | 2019-09-06 | 2020-01-03 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker state evaluation method based on big data technology |
CN110703077A (en) * | 2019-09-25 | 2020-01-17 | 西安工程大学 | HPSO-TSVM-based high-voltage circuit breaker fault diagnosis method |
CN111007799A (en) * | 2019-12-18 | 2020-04-14 | 宁波财经学院 | Numerical control equipment remote diagnosis system based on neural network |
CN111060304A (en) * | 2020-02-03 | 2020-04-24 | 国网江苏省电力有限公司宿迁供电分公司 | Circuit breaker fault diagnosis method based on neural network |
CN111123087A (en) * | 2019-12-24 | 2020-05-08 | 深圳供电局有限公司 | Power distribution network switching equipment fault detection method and device and computer equipment |
CN111507422A (en) * | 2020-04-24 | 2020-08-07 | 西安工程大学 | CQFPA-WNN-based transformer fault diagnosis method |
CN112131096A (en) * | 2020-05-07 | 2020-12-25 | 北京国网信通埃森哲信息技术有限公司 | Automatic generation method and device for ICT system fault analysis and auxiliary study and judgment test cases |
CN112504682A (en) * | 2020-12-21 | 2021-03-16 | 中国人民解放军63963部队 | Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm |
CN112631261A (en) * | 2021-02-08 | 2021-04-09 | 南京农业大学 | High-power tractor remote fault diagnosis system and method |
CN112710401A (en) * | 2021-03-29 | 2021-04-27 | 国网江西省电力有限公司供电服务管理中心 | Method for detecting temperature of electric energy meter terminal |
CN112800976A (en) * | 2021-01-29 | 2021-05-14 | 南京理工大学 | Fault diagnosis method based on vibration signal of high-voltage circuit breaker |
CN112819646A (en) * | 2021-01-18 | 2021-05-18 | 南京理工大学 | Fault diagnosis system for flexible production line of customized wood furniture |
CN112816211A (en) * | 2020-12-31 | 2021-05-18 | 无锡东研信科科技研发有限公司 | Belt conveyor fault diagnosis method based on PSO-BP algorithm |
CN113139606A (en) * | 2021-04-27 | 2021-07-20 | 上海联创设计集团股份有限公司 | Heating and ventilation fan fault detection method and system based on artificial intelligence algorithm |
CN113157732A (en) * | 2021-02-23 | 2021-07-23 | 北京科技大学 | Underground scraper fault diagnosis method based on PSO-BP neural network |
CN113486601A (en) * | 2021-09-07 | 2021-10-08 | 国网江西省电力有限公司电力科学研究院 | Feeder voltage calculation method and device based on CPSO-BP optimization model |
CN113554116A (en) * | 2021-08-16 | 2021-10-26 | 重庆大学 | Buckwheat disease identification method based on convolutional neural network |
CN115268418A (en) * | 2022-10-01 | 2022-11-01 | 深圳市世坤科技实业有限公司 | Electrical control equipment fault alarm system and method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105138717A (en) * | 2015-07-09 | 2015-12-09 | 上海电力学院 | Transformer state evaluation method by optimizing neural network with dynamic mutation particle swarm |
CN106291351A (en) * | 2016-09-20 | 2017-01-04 | 西安工程大学 | Primary cut-out fault detection method based on convolutional neural networks algorithm |
KR101697022B1 (en) * | 2015-10-16 | 2017-01-17 | 한국남부발전 주식회사 | Diagnostic analysis device and control method of the intelligent performance high-voltage circuit breaker |
CN106707153A (en) * | 2016-12-27 | 2017-05-24 | 北京合纵科技股份有限公司 | FOA-RBF based high-voltage circuit breaker fault diagnosis method |
-
2018
- 2018-04-27 CN CN201810390778.6A patent/CN108734202A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105138717A (en) * | 2015-07-09 | 2015-12-09 | 上海电力学院 | Transformer state evaluation method by optimizing neural network with dynamic mutation particle swarm |
KR101697022B1 (en) * | 2015-10-16 | 2017-01-17 | 한국남부발전 주식회사 | Diagnostic analysis device and control method of the intelligent performance high-voltage circuit breaker |
CN106291351A (en) * | 2016-09-20 | 2017-01-04 | 西安工程大学 | Primary cut-out fault detection method based on convolutional neural networks algorithm |
CN106707153A (en) * | 2016-12-27 | 2017-05-24 | 北京合纵科技股份有限公司 | FOA-RBF based high-voltage circuit breaker fault diagnosis method |
Non-Patent Citations (3)
Title |
---|
岳小斌等: "基于粒子群优化神经网络的高压断路器故障诊断", 《电力学报》 * |
张楠等: "一种基于粒子群算法和育种算法的混合算法", 《西北大学学报(自然科学版)》 * |
王羽等: "基于育种算法的BP网络学习研究", 《水力发电》 * |
Cited By (38)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109856969A (en) * | 2018-11-06 | 2019-06-07 | 皖西学院 | A kind of failure prediction method and forecasting system based on BP neural network model |
CN109856969B (en) * | 2018-11-06 | 2023-10-03 | 皖西学院 | Fault prediction method and prediction system based on BP neural network model |
CN109492816A (en) * | 2018-11-16 | 2019-03-19 | 辽宁工程技术大学 | A kind of coal and gas prominent dynamic prediction method based on hybrid intelligent |
CN109492816B (en) * | 2018-11-16 | 2021-11-09 | 辽宁工程技术大学 | Coal and gas outburst dynamic prediction method based on hybrid intelligence |
CN109298330A (en) * | 2018-11-26 | 2019-02-01 | 西安工程大学 | Fault Diagnosis for HV Circuit Breakers method based on GHPSO-BP |
CN109558677A (en) * | 2018-11-29 | 2019-04-02 | 东北大学 | A kind of hot rolling strip crown prediction technique based on data-driven |
CN109948194B (en) * | 2019-02-27 | 2020-07-03 | 北京航空航天大学 | High-voltage circuit breaker mechanical defect integrated learning diagnosis method |
CN109948194A (en) * | 2019-02-27 | 2019-06-28 | 北京航空航天大学 | A kind of high-voltage circuitbreaker mechanical defect integrated study diagnostic method |
CN110210632A (en) * | 2019-05-31 | 2019-09-06 | 国网河北省电力有限公司沧州供电分公司 | Failure emergency processing method, device and terminal based on ubiquitous electric power Internet of Things |
CN110232372A (en) * | 2019-06-26 | 2019-09-13 | 电子科技大学成都学院 | Gait recognition method based on particle group optimizing BP neural network |
CN110263907A (en) * | 2019-07-01 | 2019-09-20 | 上海海事大学 | Based on the ship short trouble diagnostic method for improving GA-PSO-BP |
CN110263907B (en) * | 2019-07-01 | 2020-12-29 | 上海海事大学 | Ship short-circuit fault diagnosis method based on improved GA-PSO-BP |
CN110648315A (en) * | 2019-09-06 | 2020-01-03 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker state evaluation method based on big data technology |
CN110648315B (en) * | 2019-09-06 | 2022-04-26 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker state evaluation method based on big data technology |
CN110703077A (en) * | 2019-09-25 | 2020-01-17 | 西安工程大学 | HPSO-TSVM-based high-voltage circuit breaker fault diagnosis method |
CN110703077B (en) * | 2019-09-25 | 2021-10-15 | 西安工程大学 | HPSO-TSVM-based high-voltage circuit breaker fault diagnosis method |
CN111007799A (en) * | 2019-12-18 | 2020-04-14 | 宁波财经学院 | Numerical control equipment remote diagnosis system based on neural network |
CN111007799B (en) * | 2019-12-18 | 2021-04-27 | 宁波财经学院 | Numerical control equipment remote diagnosis system based on neural network |
CN111123087A (en) * | 2019-12-24 | 2020-05-08 | 深圳供电局有限公司 | Power distribution network switching equipment fault detection method and device and computer equipment |
CN111060304A (en) * | 2020-02-03 | 2020-04-24 | 国网江苏省电力有限公司宿迁供电分公司 | Circuit breaker fault diagnosis method based on neural network |
CN111507422A (en) * | 2020-04-24 | 2020-08-07 | 西安工程大学 | CQFPA-WNN-based transformer fault diagnosis method |
CN111507422B (en) * | 2020-04-24 | 2023-04-07 | 西安工程大学 | CQFPA-WNN-based transformer fault diagnosis method |
CN112131096B (en) * | 2020-05-07 | 2024-05-24 | 北京国网信通埃森哲信息技术有限公司 | ICT system fault analysis and auxiliary research and judgment test case automatic generation method and device |
CN112131096A (en) * | 2020-05-07 | 2020-12-25 | 北京国网信通埃森哲信息技术有限公司 | Automatic generation method and device for ICT system fault analysis and auxiliary study and judgment test cases |
CN112504682A (en) * | 2020-12-21 | 2021-03-16 | 中国人民解放军63963部队 | Chassis engine fault diagnosis method and system based on particle swarm optimization algorithm |
CN112816211A (en) * | 2020-12-31 | 2021-05-18 | 无锡东研信科科技研发有限公司 | Belt conveyor fault diagnosis method based on PSO-BP algorithm |
CN112819646A (en) * | 2021-01-18 | 2021-05-18 | 南京理工大学 | Fault diagnosis system for flexible production line of customized wood furniture |
CN112800976A (en) * | 2021-01-29 | 2021-05-14 | 南京理工大学 | Fault diagnosis method based on vibration signal of high-voltage circuit breaker |
CN112631261A (en) * | 2021-02-08 | 2021-04-09 | 南京农业大学 | High-power tractor remote fault diagnosis system and method |
CN113157732A (en) * | 2021-02-23 | 2021-07-23 | 北京科技大学 | Underground scraper fault diagnosis method based on PSO-BP neural network |
CN113157732B (en) * | 2021-02-23 | 2023-07-25 | 北京科技大学 | Underground scraper fault diagnosis method based on PSO-BP neural network |
CN112710401B (en) * | 2021-03-29 | 2021-08-24 | 国网江西省电力有限公司供电服务管理中心 | Method for detecting temperature of electric energy meter terminal |
CN112710401A (en) * | 2021-03-29 | 2021-04-27 | 国网江西省电力有限公司供电服务管理中心 | Method for detecting temperature of electric energy meter terminal |
CN113139606A (en) * | 2021-04-27 | 2021-07-20 | 上海联创设计集团股份有限公司 | Heating and ventilation fan fault detection method and system based on artificial intelligence algorithm |
CN113554116A (en) * | 2021-08-16 | 2021-10-26 | 重庆大学 | Buckwheat disease identification method based on convolutional neural network |
CN113486601A (en) * | 2021-09-07 | 2021-10-08 | 国网江西省电力有限公司电力科学研究院 | Feeder voltage calculation method and device based on CPSO-BP optimization model |
CN113486601B (en) * | 2021-09-07 | 2021-12-31 | 国网江西省电力有限公司电力科学研究院 | Feeder voltage calculation method and device based on CPSO-BP optimization model |
CN115268418A (en) * | 2022-10-01 | 2022-11-01 | 深圳市世坤科技实业有限公司 | Electrical control equipment fault alarm system and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108734202A (en) | A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP | |
Esmin et al. | Hybrid evolutionary algorithm based on PSO and GA mutation | |
CN107977743B (en) | Multi-unmanned aerial vehicle cooperative task allocation method and device | |
Juang | A hybrid of genetic algorithm and particle swarm optimization for recurrent network design | |
Niknam et al. | Reserve constrained dynamic environmental/economic dispatch: A new multiobjective self-adaptive learning bat algorithm | |
CN108573326B (en) | Video server site selection method based on genetic ant colony hybrid algorithm | |
CN107886201B (en) | Multi-objective optimization method and device for multi-unmanned aerial vehicle task allocation | |
CN109542103A (en) | A kind of robot welding paths planning method based on fireworks particle swarm algorithm | |
Panda et al. | Evolutionary techniques for model order reduction of large scale linear systems | |
CN108133258A (en) | A kind of mixing global optimization method | |
CN111325356A (en) | Neural network search distributed training system and training method based on evolutionary computation | |
CN104679947B (en) | A kind of electronic product cable-assembly optimizes the automatic generation method of structure | |
CN108400940B (en) | A kind of multicast virtual network function dispositions method based on Estimation of Distribution Algorithm | |
CN106777449A (en) | Distribution Network Reconfiguration based on binary particle swarm algorithm | |
CN110069805A (en) | Spiral inductance optimization method based on HPSO and GA-BP algorithm | |
CN108805346A (en) | A kind of hot continuous rolling force forecasting method based on more hidden layer extreme learning machines | |
CN108832615A (en) | A kind of reconstruction method of power distribution network and system based on improvement binary particle swarm algorithm | |
CN106611378A (en) | Priority encoding-based hybrid genetic algorithm for solving job-shop scheduling problem | |
CN108075974A (en) | A kind of flow transmission control method, device and SDN architecture systems | |
CN111414681A (en) | In-building evacuation simulation method and system based on shared deep reinforcement learning | |
CN111767977A (en) | Group particle gradient descent algorithm based on improved genetic algorithm | |
CN109784497A (en) | Based on the method for calculating the AI auto-building model that figure is evolved | |
CN108985440A (en) | Two type intuitionistic fuzzy random vector function Connection Neural Network design method of section | |
CN116169776A (en) | Cloud edge cooperative artificial intelligent regulation and control method, system, medium and equipment for electric power system | |
Buditjahjanto et al. | An intelligent decision support based on a subtractive clustering and fuzzy inference system for multiobjective optimization problem in serious game |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181102 |