CN110705657A - Mode identification fault diagnosis method of proton exchange membrane fuel cell system - Google Patents

Mode identification fault diagnosis method of proton exchange membrane fuel cell system Download PDF

Info

Publication number
CN110705657A
CN110705657A CN201911146929.4A CN201911146929A CN110705657A CN 110705657 A CN110705657 A CN 110705657A CN 201911146929 A CN201911146929 A CN 201911146929A CN 110705657 A CN110705657 A CN 110705657A
Authority
CN
China
Prior art keywords
fuel cell
honey source
cell system
exchange membrane
proton exchange
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.)
Granted
Application number
CN201911146929.4A
Other languages
Chinese (zh)
Other versions
CN110705657B (en
Inventor
韩峰
田颖
邹强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Jiaotong University
Original Assignee
Beijing Jiaotong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Jiaotong University filed Critical Beijing Jiaotong University
Priority to CN201911146929.4A priority Critical patent/CN110705657B/en
Publication of CN110705657A publication Critical patent/CN110705657A/en
Application granted granted Critical
Publication of CN110705657B publication Critical patent/CN110705657B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • 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
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/30—Hydrogen technology
    • Y02E60/50—Fuel cells

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Economics (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Fuel Cell (AREA)

Abstract

The invention discloses a mode recognition fault diagnosis method of a proton exchange membrane fuel cell system, which comprises the steps of collecting diagnosis variables such as compressor motor voltage, compressor motor current, compressor rotating speed, fuel cell voltage, fuel cell current, hydrogen stacking pressure, air stacking pressure, compressor outlet pressure and the like in a normal state and a fault state, carrying out data normalization processing aiming at the obtained diagnosis variables in the normal state and the fault state, establishing an initialization sample set, carrying out fault diagnosis on the proton exchange membrane fuel cell system by adopting a PFCM-OABC-SVM combined mode recognition algorithm, and determining the state of the proton exchange membrane fuel cell system, carrying out fault diagnosis on the proton exchange membrane fuel cell system by adopting the PFCM-OABC-combined mode recognition algorithm, realizing data filtration, the optimal target parameters are obtained, faults can be accurately identified, and normal operation of the system can be effectively guaranteed.

Description

Mode identification fault diagnosis method of proton exchange membrane fuel cell system
Technical Field
The invention relates to the technical field of proton exchange membrane fuel cell systems, in particular to a mode identification fault diagnosis method of a proton exchange membrane fuel cell system.
Background
With the increasing severity of the problems of environmental pollution, global warming and the like, human beings pay more attention to energy and environmental problems, therefore, hydrogen energy is rapidly developed as a green energy, hydrogen energy is one of important secondary energy, a product generated by chemical reaction of hydrogen is water, which belongs to clean energy, a fuel cell is a chemical device for directly converting chemical energy into electric energy, and the electric energy is released by causing the hydrogen fuel and an oxidant to generate electrochemical reaction under the action of a catalyst, a Proton Exchange Membrane Fuel Cell (PEMFC) is a high-efficiency, clean and environment-friendly power generation device, is an ideal power source of electric automobiles, can also be used as a military power source or a portable power source of a dispersed power station, a submarine, a spacecraft and the like, has a very wide application prospect, and can reduce the dependence on the traditional fossil fuel by efficiently utilizing the hydrogen energy, the proton exchange membrane fuel cell system is a multi-input multi-output nonlinear strong coupling system, and has more related components, various problems are difficult to avoid in the working process, and the faults are difficult to accurately identify under the noise state, so that the fault diagnosis is carried out on the proton exchange membrane fuel cell system, and the normal operation of the system can be effectively ensured.
Disclosure of Invention
The invention aims to provide a mode recognition fault diagnosis method for a proton exchange membrane fuel cell system, which is used for carrying out fault diagnosis on the proton exchange membrane fuel cell system by adopting a PFCM-OABC-SVM combined mode recognition algorithm, realizing data filtration, obtaining optimal target parameters, accurately recognizing faults, and effectively ensuring the normal operation of the system so as to solve the problems in the background art.
In order to achieve the purpose, the invention provides the following technical scheme:
a mode identification fault diagnosis method of a proton exchange membrane fuel cell system comprises the following steps:
step S1, aiming at a known proton exchange membrane fuel cell system model, selecting compressor motor voltage, compressor motor current, compressor rotating speed, fuel cell voltage, fuel cell current, hydrogen stack inlet pressure, air stack inlet pressure, compressor outlet pressure and the like as diagnosis variables;
step S2, carrying out simulation experiment on the proton exchange membrane fuel cell system in a normal state to obtain 5000 groups of data, wherein each group of data comprises 8 groups of diagnosis variables;
k by electromechanical constants of compressorvIncrement Δ kvTo simulate a compressor failure;
constant k through cathode output apertureca,outIncrement of Δ kca,outTo simulate a supply manifold failure;
Wca,out＝(kca,out+Δkca,out)(pca-prm) (2)
carrying out a simulation experiment on the proton exchange membrane fuel cell system in a fault state to obtain 5000 groups of data, wherein each group of data comprises 8 groups of diagnosis variables;
step S3, carrying out data normalization processing aiming at the acquired diagnosis variables in the normal state and the fault state, and establishing an initialization sample set;
step S4, processing the initialized sample set by adopting a probability fuzzy C-means clustering algorithm (PFCM), eliminating sample points with membership and typicality lower than 90%, realizing data filtering, establishing the sample set, and dividing the data of the sample set into a training set and a testing set according to the ratio of 2: 1;
step S5, optimizing a penalty factor c and a kernel function parameter g of a Support Vector Machine (SVM) classifier by adopting an optimized artificial bee colony algorithm (OABC) to obtain an optimal target parameter;
and step S6, performing mode recognition fault diagnosis on the proton exchange membrane fuel cell system by adopting the optimized SVM classifier, and determining the state of the proton exchange membrane fuel cell system.
Further, in the steps S1 and S2, the diagnostic variables of the pem fuel cell system under normal state and fault state are used as raw data.
Further, in step S4, the training set and the testing set are subsets of the sample set, and are respectively used for training and testing the SVM classifier model.
Further, in the step S4, the objective function of the PFCM algorithm is as follows:
Figure BDA0002282451910000031
Figure BDA0002282451910000032
in the PFCM algorithm, the parameter a represents the influence of the membership value, the b represents the influence of the possibility value, and if the algorithm has better anti-noise capability, the value of the b can be increased, otherwise, the value of the b is reduced.
Wherein the penalty coefficient etaiThe values of (A) are as follows:
Figure BDA0002282451910000033
usually, K is 1, and the optimal solution of formula (3) is obtained, the following formula is obtained:
Figure BDA0002282451910000034
Figure BDA0002282451910000035
Figure BDA0002282451910000041
further, the core steps of the PFCM algorithm are as follows:
step 1: initializing a weighting index m to 2, setting a stop threshold to be epsilon, setting a maximum stop iteration number L, initializing an iteration number L to 0, initializing a clustering center V (0), initializing a membership matrix U (0), and initializing a typicality matrix T (0);
step 2: calculating a penalty coefficient according to a formula (5);
and step 3: updating the membership matrix according to a formula (6);
Figure BDA0002282451910000042
and 4, step 4: updating the typicality matrix according to formula (7);
Figure BDA0002282451910000043
and 5: updating the clustering center matrix according to the formula (8);
Figure BDA0002282451910000044
step 6: judgment | | | V(l+1)-VlIf | < ε, if true or the number of iterations L<Stopping iteration and outputting a clustering center, a membership matrix and a typicality matrix; otherwise, let l be l +1, go to step 2.
Further, the step of optimizing the SVM by the OABC algorithm is as follows:
step (1): the initialization of parameters in the OABC algorithm mainly comprises the following steps: the bee colony scale is 20, the number N of honey sources is 10, namely the number of collected bees; the maximum cycle number Limit of the honey source is 100; the maximum iteration number maxIter is 10; the search range for the penalty factor c is [0.01,100]The search range of the kernel function parameter g is [0.01,100 ]]Initializing each honey source to xij，i＝1,2,…,10，j＝1,2；
Step (2): the fitness function in the OABC algorithm is determined, the SVM parameter is optimized to improve the accuracy of system fault classification, the optimization problem solving process can be regarded as a process of searching a honey source by bees, the fitness function is selected as a formula (12), and the objective function value is the classification accuracy:
in the formula: fitnessiIs the fitness value of the ith set of parameters, fiThe objective function value of the ith honey source is obtained;
and (3): obtaining a formula (14) by adopting a Levy formula (13), searching the neighborhood of the current honey source by the honey bee according to the formula (14), calculating the fitness of the new honey source according to a formula (12), replacing the position of the original honey source with the position of the new honey source if the fitness of the new honey source is better than the fitness of the original honey source, otherwise, keeping the original honey source unchanged;
Figure BDA0002282451910000052
x'ij＝xij+a(xij-xbest)L(α) (14)
in the formula, alpha is a characteristic index and takes the value of 1.5; gamma function and satisfies
Figure BDA0002282451910000053
a is a stepping length, x 'satisfying normal distribution'ijIs the position of the new honey source.
And (4): after the honey bee is subjected to global search, selecting a honey source according to a formula (15) by the following bee, then performing neighborhood search by using a formula (14) to obtain a new honey source, and if the fitness value of the new honey source is better than that of the original honey source, replacing the position of the original honey source with the position of the new honey source, otherwise, keeping the original honey source unchanged;
Figure BDA0002282451910000061
in the formula, PiProbability of being selected for the ith honey source; fitnessiThe fitness value of the ith honey source is obtained; n is the total number of the honey sources;
and (5): judging whether the cycle number of a certain honey source is greater than Limit, if so, generating a new honey source according to a formula (16);
Figure BDA0002282451910000062
in the formula, xijIs the value of the jth dimension of the ith honey source, j belongs to {1,2 };
and (6): recording the current optimal honey source, judging whether a loop termination condition (iteration maxIter times) is met, if so, turning to the step (7), otherwise, turning to the step (3);
and (7): and constructing an optimal SVM classifier according to the obtained global optimal honey source, namely optimal parameters c and g, and verifying the trained SVM classifier model through a test set sample.
Further, in the step (5), the position of the generated new honey source is random, and in the step (7), a fault diagnosis result is determined according to the verification result of the SVM classifier model.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention adopts PFCM-OABC-SVM combined mode recognition algorithm to diagnose the fault of the proton exchange membrane fuel cell system, realizes the filtration of data, obtains the optimal target parameter, can accurately recognize the fault and can effectively ensure the normal operation of the system;
2. the invention adopts the optimized SVM classifier to carry out the mode recognition fault diagnosis of the proton exchange membrane fuel cell system, determines the state of the proton exchange membrane fuel cell system, can accurately recognize the fault in a noise state, and improves the accuracy of system fault classification.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention without limiting the invention.
FIG. 1 is a fault diagnosis flow chart of a proton exchange membrane fuel cell system based on PFCM-OABC-SVM of the invention;
fig. 2 is a flow chart of the PFCM algorithm of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, and it should be understood that the preferred embodiments described herein are merely for purposes of illustration and explanation and are not intended to limit the present invention.
The first embodiment is as follows:
as shown in fig. 1-2, the present invention provides a technical solution: a mode identification fault diagnosis method of a proton exchange membrane fuel cell system comprises the following steps:
step S1, aiming at a known proton exchange membrane fuel cell system model, selecting compressor motor voltage, compressor motor current, compressor rotating speed, fuel cell voltage, fuel cell current, hydrogen stack inlet pressure, air stack inlet pressure, compressor outlet pressure and the like as diagnosis variables;
step S2, carrying out simulation experiment on the proton exchange membrane fuel cell system in a normal state to obtain 5000 groups of data, wherein each group of data comprises 8 groups of diagnosis variables;
k by electromechanical constants of compressorvIncrement Δ kvTo simulate a compressor failure;
Figure BDA0002282451910000081
constant k through cathode output apertureca,outIncrement of Δ kca,outTo simulate a supply manifold failure;
Wca,out＝(kca,out+Δkca,out)(pca-prm) (18)
carrying out a simulation experiment on the proton exchange membrane fuel cell system in a fault state to obtain 5000 groups of data, wherein each group of data comprises 8 groups of diagnosis variables;
step S3, carrying out data normalization processing aiming at the acquired diagnosis variables in the normal state and the fault state, and establishing an initialization sample set;
step S4, processing the initialized sample set by adopting a probability fuzzy C-means clustering algorithm (PFCM), eliminating sample points with membership and typicality lower than 90%, realizing data filtering, establishing the sample set, and dividing the data of the sample set into a training set and a testing set according to the ratio of 2: 1;
step S5, optimizing a penalty factor c and a kernel function parameter g of a Support Vector Machine (SVM) classifier by adopting an optimized artificial bee colony algorithm (OABC) to obtain an optimal target parameter;
and step S6, performing mode recognition fault diagnosis on the proton exchange membrane fuel cell system by adopting the optimized SVM classifier, and determining the state of the proton exchange membrane fuel cell system.
In the present embodiment, in the steps S1 and S2, the diagnostic variables of the pem fuel cell system under normal state and fault state are used as raw data.
In this embodiment, in step S4, the training set and the test set are subsets of a sample set, and are respectively used for training and testing an SVM classifier model.
Example two:
in the first embodiment, the objective function of the PFCM algorithm is as follows:
Figure BDA0002282451910000091
in the PFCM algorithm, the parameter a represents the influence of the membership value, the b represents the influence of the possibility value, and if the algorithm has better anti-noise capability, the value of the b can be increased, otherwise, the value of the b is reduced.
Wherein the penalty coefficient etaiThe values of (A) are as follows:
Figure BDA0002282451910000092
usually, K is 1, and the optimal solution of equation (19) is found, the following equation can be obtained:
Figure BDA0002282451910000093
Figure BDA0002282451910000094
Figure BDA0002282451910000095
as shown in fig. 2, in the first embodiment, the core steps of the PFCM algorithm are as follows:
step 1: initializing a weighting index m to 2, setting a stop threshold to be epsilon, setting a maximum stop iteration number L, initializing an iteration number L to 0, initializing a clustering center V (0), initializing a membership matrix U (0), and initializing a typicality matrix T (0);
step 2: calculating a penalty coefficient according to a formula (21);
and step 3: updating the membership matrix according to a formula (22);
and 4, step 4: updating the typicality matrix according to formula (23);
Figure BDA0002282451910000101
and 5: updating the cluster center matrix according to formula (24);
Figure BDA0002282451910000102
step 6: judgment | | | V(l+1)-VlIf | < ε, if true or the number of iterations L<Stopping iteration and outputting a clustering center, a membership matrix and a typicality matrix; otherwise, let l be l +1, go to step 2.
Example three:
as shown in fig. 1, in the first embodiment, the step of optimizing SVM by the OABC algorithm is as follows:
step (1): the initialization of parameters in the OABC algorithm mainly comprises the following steps: the bee colony scale is 20, the number N of honey sources is 10, namely the number of collected bees; the maximum cycle number Limit of the honey source is 100; the maximum iteration number maxIter is 10; punishment causeSearch range for sub-c is [0.01,100 ]]The search range of the kernel function parameter g is [0.01,100 ]]Initializing each honey source to xij，i＝1,2,…,10，j＝1,2；
Step (2): the fitness function in the OABC algorithm is determined, the SVM parameter is optimized to improve the accuracy of system fault classification, the optimization problem solving process can be regarded as a process of searching a honey source by bees, the fitness function is selected as a formula (28), and the objective function value is the classification accuracy:
Figure BDA0002282451910000111
in the formula: fitnessiIs the fitness value of the ith set of parameters, fiThe objective function value of the ith honey source is obtained;
and (3): obtaining a formula (30) by adopting a Levy formula (29), searching the neighborhood of the current honey source by the honey bee according to the formula (30), calculating the fitness of the new honey source according to a formula (28), and replacing the position of the original honey source with the position of the new honey source if the fitness of the new honey source is better than the fitness of the original honey source, otherwise, keeping the original honey source unchanged;
Figure BDA0002282451910000112
x'ij＝xij+a(xij-xbest)L(α) (30)
in the formula, alpha is a characteristic index and takes the value of 1.5; gamma function and satisfies
Figure BDA0002282451910000113
a is a stepping length, x 'satisfying normal distribution'ijIs the position of the new honey source.
And (4): after the honey bee is subjected to global search, selecting a honey source according to a formula (31) by the following bee, then performing neighborhood search by using a formula (30) to obtain a new honey source, and if the fitness value of the new honey source is better than that of the original honey source, replacing the position of the original honey source with the position of the new honey source, otherwise, keeping the original honey source unchanged;
Figure BDA0002282451910000114
in the formula, PiProbability of being selected for the ith honey source; fitnessiThe fitness value of the ith honey source is obtained; n is the total number of the honey sources;
and (5): judging whether the cycle number of a certain honey source is greater than Limit, if so, generating a new honey source according to a formula (32);
Figure BDA0002282451910000121
in the formula, xijIs the value of the jth dimension of the ith honey source, j belongs to {1,2 };
and (6): recording the current optimal honey source, judging whether a loop termination condition (iteration maxIter times) is met, if so, turning to the step (7), otherwise, turning to the step (3);
and (7): and constructing an optimal SVM classifier according to the obtained global optimal honey source, namely optimal parameters c and g, and verifying the trained SVM classifier model through a test set sample.
In this embodiment, in the step (5), the positions of the generated new honey sources are random, and in the step (7), a fault diagnosis result is determined according to a verification result of the SVM classifier model.
The invention is improved as follows: the invention relates to a mode identification fault diagnosis method of a proton exchange membrane fuel cell system, which comprises the steps of collecting diagnosis variables such as compressor motor voltage, compressor motor current, compressor rotating speed, fuel cell voltage, fuel cell current, hydrogen stack inlet pressure, air stack inlet pressure, compressor outlet pressure and the like in a normal state and a fault state, carrying out data normalization processing aiming at the obtained diagnosis variables in the normal state and the fault state, establishing an initialization sample set, then fault diagnosis of the proton exchange membrane fuel cell system is carried out by adopting a PFCM-OABC-SVM combined mode recognition algorithm, the state of the proton exchange membrane fuel cell system is determined, the filtration of data is realized, the optimal target parameter is obtained, and under the noise state, faults are accurately identified, the accuracy of system fault classification is improved, and the normal operation of the system can be effectively guaranteed.
Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (6)

1. A mode identification fault diagnosis method of a proton exchange membrane fuel cell system is characterized by comprising the following steps:
step S1, aiming at a known proton exchange membrane fuel cell system model, selecting compressor motor voltage, compressor motor current, compressor rotating speed, fuel cell voltage, fuel cell current, hydrogen stack inlet pressure, air stack inlet pressure, compressor outlet pressure and the like as diagnosis variables;
step S2, carrying out simulation experiment on the proton exchange membrane fuel cell system in a normal state to obtain 5000 groups of data, wherein each group of data comprises 8 groups of diagnosis variables;
k by electromechanical constants of compressorvIncrement Δ kvTo simulate a compressor failure;
Figure FDA0002282451900000011
constant k through cathode output apertureca,outIncrement of Δ kca,outTo simulate a supply manifold failure;
Wca,out＝(kca,out+Δkca,out)(pca-prm) (2)
carrying out a simulation experiment on the proton exchange membrane fuel cell system in a fault state to obtain 5000 groups of data, wherein each group of data comprises 8 groups of diagnosis variables;
step S3, carrying out data normalization processing aiming at the acquired diagnosis variables in the normal state and the fault state, and establishing an initialization sample set;
step S4, processing the initialized sample set by adopting a probability fuzzy C-means clustering algorithm (PFCM), eliminating sample points with membership and typicality lower than 90%, realizing data filtering, establishing the sample set, and dividing the data of the sample set into a training set and a testing set according to the ratio of 2: 1;
step S5, optimizing a penalty factor c and a kernel function parameter g of a Support Vector Machine (SVM) classifier by adopting an optimized artificial bee colony algorithm (OABC) to obtain an optimal target parameter;
and step S6, performing mode recognition fault diagnosis on the proton exchange membrane fuel cell system by adopting the optimized SVM classifier, and determining the state of the proton exchange membrane fuel cell system.
2. The method of claim 1, wherein the diagnostic variables of the PEM fuel cell system in normal state and fault state are used as raw data in the steps S1 and S2.
3. The method of claim 1, wherein in step S4, the training set and the testing set are subsets of a sample set, and are respectively used for training and testing an SVM classifier model.
4. The method of claim 1, wherein in step S5, the step of the OABC algorithm optimizing SVM is as follows:
step (1): the initialization of parameters in the OABC algorithm mainly comprises the following steps: the bee colony scale is 20, the number N of honey sources is 10, namely the number of collected bees; maximum cycle number L of honey sourceThe init is 100; the maximum iteration number maxIter is 10; the search range for the penalty factor c is [0.01,100]The search range of the kernel function parameter g is [0.01,100 ]]Initializing each honey source to xij，i＝1,2,…,10，j＝1,2；
Step (2): the fitness function in the OABC algorithm is determined, the SVM parameter is optimized to improve the accuracy of system fault classification, the optimization problem solving process can be regarded as a process of searching a honey source by bees, the fitness function is selected as a formula (3), and the objective function value is the classification accuracy:
Figure FDA0002282451900000021
in the formula: fitnessiIs the fitness value of the ith set of parameters, fiThe objective function value of the ith honey source is obtained;
and (3): obtaining a formula (5) by adopting a Levy formula (4), searching the neighborhood of the current honey source by the honey bee according to the formula (5), and calculating the fitness of the new honey source according to the formula (3), wherein if the fitness value of the new honey source is better than that of the original honey source, the position of the new honey source is used for replacing the position of the original honey source, otherwise, the original honey source is kept unchanged;
Figure FDA0002282451900000031
x′ij＝xij+a(xij-xbest)L(α) (5)
in the formula, alpha is a characteristic index and takes the value of 1.5; gamma function and satisfies
Figure FDA0002282451900000032
a is a stepping length, x 'satisfying normal distribution'ijThe position of the new honey source;
and (4): after the honey bee is subjected to global search, selecting a honey source according to a formula (6) by the following bee, then performing neighborhood search by using a formula (5) to obtain a new honey source, and if the fitness value of the new honey source is better than that of the original honey source, replacing the position of the original honey source with the position of the new honey source, otherwise, keeping the original honey source unchanged;
Figure FDA0002282451900000033
in the formula, PiProbability of being selected for the ith honey source; fitnessiThe fitness value of the ith honey source is obtained; n is the total number of the honey sources;
and (5): judging whether the cycle number of a certain honey source is greater than Limit, if so, generating a new honey source according to a formula (7);
Figure FDA0002282451900000034
in the formula, xijIs the value of the jth dimension of the ith honey source, j belongs to {1,2 };
and (6): recording the current optimal honey source, judging whether a loop termination condition (iteration maxIter times) is met, if so, turning to the step (7), otherwise, turning to the step (3);
and (7): and constructing an optimal SVM classifier according to the obtained global optimal honey source, namely optimal parameters c and g, and verifying the trained SVM classifier model through a test set sample.
5. The pattern recognition fault diagnosis method of a proton exchange membrane fuel cell system as claimed in claim 4, wherein in the step (5), the position of the generated new honey source is random.
6. The pattern recognition fault diagnosis method of a proton exchange membrane fuel cell system as claimed in claim 4, wherein in the step (7), the fault diagnosis result is determined according to the verification result of the SVM classifier model.
CN201911146929.4A 2019-11-21 2019-11-21 Mode identification fault diagnosis method of proton exchange membrane fuel cell system Active CN110705657B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911146929.4A CN110705657B (en) 2019-11-21 2019-11-21 Mode identification fault diagnosis method of proton exchange membrane fuel cell system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911146929.4A CN110705657B (en) 2019-11-21 2019-11-21 Mode identification fault diagnosis method of proton exchange membrane fuel cell system

Publications (2)

Publication Number Publication Date
CN110705657A true CN110705657A (en) 2020-01-17
CN110705657B CN110705657B (en) 2022-08-30

Family

ID=69206479

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911146929.4A Active CN110705657B (en) 2019-11-21 2019-11-21 Mode identification fault diagnosis method of proton exchange membrane fuel cell system

Country Status (1)

Country Link
CN (1) CN110705657B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111090050A (en) * 2020-01-21 2020-05-01 合肥工业大学 Lithium battery fault diagnosis method based on support vector machine and K mean value
CN112782589A (en) * 2021-01-26 2021-05-11 武汉理工大学 Vehicle-mounted fuel cell remote fault classification diagnosis method and device and storage medium
CN113036189A (en) * 2021-03-01 2021-06-25 中国科学技术大学 Method and device for detecting running state of proton exchange membrane fuel cell
CN113467423A (en) * 2021-07-01 2021-10-01 中山大学 PEMFC fault diagnosis method and system based on cloud platform
CN114492087A (en) * 2022-04-02 2022-05-13 国网浙江省电力有限公司电力科学研究院 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station
CN117117258A (en) * 2023-10-24 2023-11-24 新研氢能源科技有限公司 Fault monitoring method and device for hydrogen fuel cell system

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103700871A (en) * 2013-12-07 2014-04-02 西南交通大学 Control method for optimal efficiency of proton exchange membrane fuel cell system of locomotive
CN105930864A (en) * 2016-04-15 2016-09-07 杭州电子科技大学 EEG (electroencephalogram) signal feature classification method based on ABC-SVM
CN107644127A (en) * 2017-09-08 2018-01-30 西安工程大学 Diagnosis Method of Transformer Faults based on IMABC Support Vector Machines Optimizeds
US20180184915A1 (en) * 2017-01-04 2018-07-05 Awaire Inc. Bio-Sensing Glasses to Detect Cardiovascular Disease
CN108539802A (en) * 2018-04-13 2018-09-14 曲阜师范大学 A kind of optimization method of alternating current-direct current series-parallel connection bulk power grid
CN108596118A (en) * 2018-04-28 2018-09-28 北京师范大学 A kind of Remote Image Classification and system based on artificial bee colony algorithm
CN110298399A (en) * 2019-06-27 2019-10-01 东北大学 The oil well fault diagnostic method merged based on Freeman chain code and moment characteristics

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103700871A (en) * 2013-12-07 2014-04-02 西南交通大学 Control method for optimal efficiency of proton exchange membrane fuel cell system of locomotive
CN105930864A (en) * 2016-04-15 2016-09-07 杭州电子科技大学 EEG (electroencephalogram) signal feature classification method based on ABC-SVM
US20180184915A1 (en) * 2017-01-04 2018-07-05 Awaire Inc. Bio-Sensing Glasses to Detect Cardiovascular Disease
CN107644127A (en) * 2017-09-08 2018-01-30 西安工程大学 Diagnosis Method of Transformer Faults based on IMABC Support Vector Machines Optimizeds
CN108539802A (en) * 2018-04-13 2018-09-14 曲阜师范大学 A kind of optimization method of alternating current-direct current series-parallel connection bulk power grid
CN108596118A (en) * 2018-04-28 2018-09-28 北京师范大学 A kind of Remote Image Classification and system based on artificial bee colony algorithm
CN110298399A (en) * 2019-06-27 2019-10-01 东北大学 The oil well fault diagnostic method merged based on Freeman chain code and moment characteristics

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
刘嘉蔚等: "基于多分类相关向量机和模糊C均值聚类的有轨电车用燃料电池系统故障诊断方法", 《中国电机工程学报》 *
吴印华等: "基于改进蜂群算法优化支持向量机的故障诊断", 《机械强度》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111090050A (en) * 2020-01-21 2020-05-01 合肥工业大学 Lithium battery fault diagnosis method based on support vector machine and K mean value
CN112782589A (en) * 2021-01-26 2021-05-11 武汉理工大学 Vehicle-mounted fuel cell remote fault classification diagnosis method and device and storage medium
CN113036189A (en) * 2021-03-01 2021-06-25 中国科学技术大学 Method and device for detecting running state of proton exchange membrane fuel cell
CN113036189B (en) * 2021-03-01 2022-09-30 中国科学技术大学 Method and device for detecting running state of proton exchange membrane fuel cell
CN113467423A (en) * 2021-07-01 2021-10-01 中山大学 PEMFC fault diagnosis method and system based on cloud platform
CN114492087A (en) * 2022-04-02 2022-05-13 国网浙江省电力有限公司电力科学研究院 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station
CN114492087B (en) * 2022-04-02 2022-07-19 国网浙江省电力有限公司电力科学研究院 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station
CN117117258A (en) * 2023-10-24 2023-11-24 新研氢能源科技有限公司 Fault monitoring method and device for hydrogen fuel cell system
CN117117258B (en) * 2023-10-24 2024-01-09 新研氢能源科技有限公司 Fault monitoring method and device for hydrogen fuel cell system

Also Published As

Publication number Publication date
CN110705657B (en) 2022-08-30

Similar Documents

Publication Publication Date Title
CN110705657B (en) Mode identification fault diagnosis method of proton exchange membrane fuel cell system
CN110752410B (en) Method for rapidly sorting and recombining retired lithium battery
CN113484787B (en) Electrochemical impedance spectrum-based rapid sorting and recombining method for retired lithium ion battery
CN104573879A (en) Photovoltaic power station output predicting method based on optimal similar day set
Liu et al. Sequence fault diagnosis for PEMFC water management subsystem using deep learning with t-SNE
CN114899457B (en) Fault detection method for proton exchange membrane fuel cell system
Zheng et al. Performance prediction of fuel cells using long short‐term memory recurrent neural network
Varanasi et al. K-means clustering based photo voltaic power forecasting using artificial neural network, particle swarm optimization and support vector regression
CN115598536A (en) PEMFC fault diagnosis method based on fuzzy C-means clustering and probabilistic neural network
CN109858564A (en) Modified Adaboost-SVM model generating method suitable for wind electric converter fault diagnosis
Abd Elaziz et al. Optimal parameters extracting of fuel cell based on Gorilla Troops Optimizer
CN112069741A (en) Fuel cell stack transient aging prediction method, device, equipment and storage medium
CN116484299A (en) Charging pile fault diagnosis method based on integration of gradient lifting tree and multi-layer perceptron
Quan et al. An enhanced fault diagnosis method for fuel cell system using a kernel extreme learning machine optimized with improved sparrow search algorithm
Huang et al. Fuel cell engine fault diagnosis expert system based on decision tree
CN116738204A (en) Power load data anomaly detection method based on generation of countermeasure transformers
Sun et al. Short-term photovoltaic power prediction modeling based on AdaBoost algorithm and Elman
CN111062509A (en) Wind power medium-short term prediction method and system based on improved neural network
CN115310373A (en) Method for predicting residual life of hydrogen production electrolytic cell
Xu et al. One-day ahead forecast of PV output based on deep belief network and weather classification
CN113780537A (en) Fault diagnosis method and device for proton exchange membrane fuel cell power generation system
CN114841266A (en) Voltage sag identification method based on triple prototype network under small sample
CN115021269A (en) Two-stage optimal power flow solving method based on data driving
CN109242008A (en) A kind of combined failure recognition methods under the conditions of sample class is incomplete
CN114234392A (en) Air conditioner load fine prediction method based on improved PSO-LSTM

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
GR01 Patent grant
GR01 Patent grant