CN114492087A - Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station - Google Patents

Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station Download PDF

Info

Publication number
CN114492087A
CN114492087A CN202210342798.2A CN202210342798A CN114492087A CN 114492087 A CN114492087 A CN 114492087A CN 202210342798 A CN202210342798 A CN 202210342798A CN 114492087 A CN114492087 A CN 114492087A
Authority
CN
China
Prior art keywords
fuel cell
output
model
fuzzy
parameters
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
CN202210342798.2A
Other languages
Chinese (zh)
Other versions
CN114492087B (en
Inventor
赵波
章雷其
张雪松
王激华
马丽军
叶夏明
谢长君
张领先
朱文超
刘敏
林达
唐雅洁
汪湘晋
李志浩
倪筹帷
徐珂
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.)
Electric Power Research Institute of State Grid Zhejiang Electric Power Co Ltd
Original Assignee
Electric Power Research Institute of State Grid Zhejiang Electric Power Co Ltd
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 Electric Power Research Institute of State Grid Zhejiang Electric Power Co Ltd filed Critical Electric Power Research Institute of State Grid Zhejiang Electric Power Co Ltd
Priority to CN202210342798.2A priority Critical patent/CN114492087B/en
Publication of CN114492087A publication Critical patent/CN114492087A/en
Application granted granted Critical
Publication of CN114492087B publication Critical patent/CN114492087B/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
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Automation & Control Theory (AREA)
  • Water Supply & Treatment (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Mathematical Optimization (AREA)
  • Fuzzy Systems (AREA)
  • Public Health (AREA)
  • Pure & Applied Mathematics (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 fault diagnosis method and device for a proton exchange membrane fuel cell of a hydrogen energy storage power station. The method of the invention comprises the following steps: 1) periodically collecting voltage and current data of all proton exchange membrane fuel cell stacks in an operating state and uploading the data to a cloud end for storage; 2) in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing; 3) in the same sampling period, the identified fuel cell output characteristic model parameters are used as input quantity of the T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the T-S fuzzy model; 4) analyzing and comparing the variation trend of the output quantity of the T-S fuzzy model, and recording the characteristics when the proton exchange membrane fuel cell has faults. The invention can effectively improve the accuracy of the fault diagnosis of the fuel cell and realize the accurate positioning of the fault point of the fuel cell.

Description

Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station
Technical Field
The invention belongs to the field of hydrogen energy and fuel cells, relates to fault diagnosis of fuel cells, and particularly relates to a method and a device for fault diagnosis of proton exchange membrane fuel cells of a hydrogen energy storage power station based on cloud computing.
Background
Hydrogen energy storage power stations are an important way to achieve low-cost, large-scale, long-term energy storage, and thus have been increasingly gaining attention. In a hydrogen energy storage power station, an electricity-to-hydrogen device, a hydrogen storage device, and a hydrogen-to-electricity device are generally included, and a fuel cell is a core device that converts hydrogen energy into electric energy.
The proton exchange membrane fuel cell is the most widely applied fuel cell technology at present, has the advantages of high working efficiency, no pollution, low running noise and the like, and has wide application in the fields of aerospace and electric automobiles. Compared with the traditional application, the proton exchange membrane fuel cell applied to the hydrogen energy storage power station has two prominent differences: firstly, the capacity is large, and the proton exchange membrane battery for energy storage needs to reach MW level capacity; secondly, the number of the galvanic piles is large, the capacity of a single galvanic pile is difficult to meet the requirement of a power station, and a certain number of galvanic piles are required to be cascaded.
The two outstanding characteristics bring great challenges to the operation and maintenance of the fuel cell, and if the fault galvanic pile cannot be identified quickly and replaced, the shutdown of the energy storage power station or even certain potential safety hazards are caused. Therefore, research and development of related strategies are urgently needed to realize rapid diagnosis of faults of proton exchange membrane fuel cells in hydrogen energy storage power stations.
Disclosure of Invention
In order to improve the fault diagnosis efficiency of single cells and electric piles of a proton exchange membrane fuel cell in a hydrogen energy storage power station and realize multi-terminal intelligent control of the energy storage power station, the invention provides a method and a device for diagnosing the fault of the proton exchange membrane fuel cell in the hydrogen energy storage power station based on cloud computing, so that the accuracy of fault diagnosis of the fuel cell is effectively improved, the fault point of the fuel cell is accurately positioned, the unified management and data sharing of a large number of power stations are realized through the cloud computing (a cloud platform), and the on-site computing burden is reduced.
Therefore, the invention adopts a technical scheme as follows: the fault diagnosis method for the proton exchange membrane fuel cell of the hydrogen energy storage power station comprises the following steps:
1) periodically collecting voltage and current data of all proton exchange membrane fuel cell stacks in an operating state and uploading the data to a cloud end for storage;
2) in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing;
3) in the same sampling period, the identified fuel cell output characteristic model parameters are used as the input quantity of the one-dimensional T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the one-dimensional T-S fuzzy model;
4) analyzing and comparing the variation trend of the output quantity of the one-dimensional T-S fuzzy model, recording the characteristics when the proton exchange membrane fuel cell has faults, and using the characteristics as a historical knowledge base and a reference basis for subsequent fault diagnosis of the proton exchange membrane fuel cell.
As a supplement to the above solution, in the step 2), a fuel cell output characteristic model is created as follows, the fuel cell output characteristic model including a voltage model, an anode gas supply model, and a cathode gas supply model, and the fuel cell output characteristic model is created based on the following assumption conditions:
21) all the gas entering the cathode and the anode is fully humidified;
22) all gases belong to ideal gases and accord with the relevant reaction equation of the ideal gases;
23) the internal temperature of the fuel cell changes slowly, and the internal temperature is kept consistent;
24) neglecting the influence of water on a fuel cell power generation system in a gaseous state;
25) the basic principles of pressure balance and material conservation are followed.
As a supplement to the above technical solution, in step 2), the specific steps of using the chaotic particle swarm algorithm at the cloud end to perform parameter identification on all the electric piles in parallel are as follows:
31) the purpose of algorithm optimization is to make the objective function smaller, and the objective optimization function is specifically defined as follows:
Figure 262539DEST_PATH_IMAGE002
，
wherein the content of the first and second substances,RMSEis the root mean square error of the received signal,Nis the number of cells in the stack,V cellis the output voltage of the stack of cells,V i is the output voltage of the fuel cell output characteristic model;
32) defining the size of a population, iteration times and input parameter dimensions, wherein the size of the population is positively correlated with the number of galvanic piles;
33) giving upper and lower limits of parameters to be identified based on historical data experience;
34) creating a chaotic sequence based on the SkewTent mapping, fusing the chaotic sequence with upper and lower parameter limits to generate an initial population, and increasing the optimization probability of a global optimal solution;
35) creating a chaotic disturbance sequence based on Logistic mapping, and fusing the chaotic disturbance sequence with the initial position of each particle individual to obtain a disturbance increment;
36) calculating the fitness value of each particle, and updating the historical individual optimal position and the group optimal position of each particle;
37) updating the speed and position of the particle, calculating the positionXCorresponding fitness value is recorded asfFor updated positions of particlesX′Applying disturbance increment to calculate positionX′Corresponding fitness value is recorded asf′，ComparisonfAndf′the size of (a) is (b),
if it isf<f′The final position of the particle is not changed; if it isf>f′The final position of the particle is updated toX′；
38) Carrying out-of-range processing on the speed and the position, setting parameters exceeding the upper limit as boundary upper limits, and setting parameters exceeding the lower limit as boundary lower limits;
39) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting parameters and a final fitness value; and if not, continuing to repeatedly execute the steps 36) to 38) until the iteration termination condition is met.
As a supplement to the above technical solution, in the step 3), the identified parameters of the fuel cell output characteristic model are normalized and used as input quantities of the one-dimensional T-S fuzzy model, the number of the fuzzy reference subsets is L, the membership function parameters are identified by using an optimization algorithm and input into the fuzzy model, and the fuzzy reference subsets belonging to the input of each one-dimensional T-S fuzzy model are inputkMembership function ofu Bn,k Comprises the following steps:
Figure 289269DEST_PATH_IMAGE003
，
wherein the content of the first and second substances,Brepresenting the identified fuel cell output characteristic model parameters,B n is a specific value corresponding to the fuel cell output characteristic model parameter,nis the number of identified parameters;m Bn,k ands Bn,k is a membership function parameter to be determined;
according to the inference rule of the T-S fuzzy model, when L fuzzy reference subsets are input, L fuzzy reference subsets are output, and the output is as follows:
Figure 769929DEST_PATH_IMAGE004
；
wherein the content of the first and second substances,u output,m a function of membership representing the output,mis a fuzzy reference subset output by the one-dimensional T-S fuzzy model;
the method adopts an optimization algorithm to identify membership function parameters, and comprises the following steps:
41) optimizing the objective function of the algorithm as the output of the fuzzy modeloutputThe purpose of algorithm optimization is to find the optimal parameters to maximize the objective function;
42) the input parameters are normalized, the parameter range is limited to (0,1), and two significant digits are reserved for each parameter;
43) identifying parameters of the membership function by adopting an optimization algorithm;
44) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting optimization parameters and corresponding objective function values; if not, continue to 43) until the iteration termination requirement is met.
As a supplement to the above technical solution, in the step 4), a variation trend of the output quantity of the fuzzy model is analyzed and compared, and the variation trend of the output quantity of the fuzzy model is expressed as:
Figure 62370DEST_PATH_IMAGE005
，
wherein the content of the first and second substances,output(t) Is at a certain momenttThe output of the fuzzy model is then used,Tis the period of the sampling, and,output(t+T) Is the output of the fuzzy model spaced one sampling period apart,Doutputis the amount of change in output;
when the fuel cell fails, the change trend of the output quantity of the fuzzy model is abnormal, the abnormal conditions respectively correspond to different fault states of the fuel cell, the abnormal states are recorded at the cloud end to serve as a historical knowledge base, and in the following working period, when the output quantity of the fuzzy model is abnormal, the historical knowledge base is inquired for matching, so that accurate fault diagnosis of the fuel cell is carried out, and accurate fault positioning of the fuel cell is carried out; if the fault is reversible, automatically issuing an instruction to a fuel cell stack controller, and adjusting operation parameters to enable the fuel cell to recover the normal working state; and if the fault is irreversible, issuing a shutdown instruction in time to avoid further damage to the fuel cell. The more times of fault diagnosis, the more data accumulated in the historical knowledge base, and the higher the diagnosis accuracy.
The other technical scheme adopted by the invention is as follows: proton exchange membrane fuel cell fault diagnosis device of hydrogen energy storage power station, it includes:
voltage and current data acquisition unit: periodically collecting voltage and current data of all proton exchange membrane fuel cell stacks in an operating state and uploading the data to a cloud end for storage;
proton exchange membrane fuel cell model parameter identification unit: in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing;
a one-dimensional T-S fuzzy model construction unit: in the same sampling period, the identified fuel cell output characteristic model parameters are used as the input quantity of the one-dimensional T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the one-dimensional T-S fuzzy model;
fuel cell failure diagnosis unit: analyzing and comparing the variation trend of the output quantity of the one-dimensional T-S fuzzy model, recording the characteristics when the proton exchange membrane fuel cell has faults, and using the characteristics as a historical knowledge base and a reference basis for subsequent fault diagnosis of the proton exchange membrane fuel cell.
The invention has the beneficial effects that: the whole fuel cell stack can be accurately diagnosed and positioned at a fault point by only detecting the working voltage and the current of the proton exchange membrane fuel cell without additional devices. When the fault is reversible, the cloud automatically issues an instruction to the fuel cell stack controller, and the operation parameters are adjusted to enable the fuel cell to recover the normal working state. And if the fault is irreversible, issuing a shutdown instruction in time to avoid further damage to the fuel cell. The more times of fault diagnosis, the more data accumulated in the historical knowledge base, and the higher the diagnosis accuracy. Unified management and data sharing of a large number of power stations can be achieved through cloud computing (a cloud platform), and local computing burden is reduced.
Drawings
FIG. 1 is a schematic diagram of a hydrogen energy storage power plant according to an embodiment of the present invention;
FIG. 2 is a flow chart of a diagnostic method of the present invention;
FIG. 3 is a flow chart of parallel identification of model parameters according to the present invention;
FIG. 4 is a diagram illustrating the result of parameter identification of voltage and current data according to the present invention;
FIG. 5 is a diagram illustrating the result of parameter identification of voltage and current data according to the present invention;
FIG. 6 is a flow chart of the ant colony algorithm identifying membership function parameters of the present invention;
FIG. 7 is a flow chart of the fault diagnosis of the present invention;
fig. 8 is a graph showing output variation under a flooding fault in a certain time period of a certain cell stack in embodiment 1 of the present invention;
fig. 9 is a graph showing the variation of output under dry membrane failure for a certain period of time in a certain cell stack according to example 1 of the present invention;
fig. 10 is a graph showing output variation under a flooding fault in a certain time period of a certain cell stack in embodiment 2 of the present invention;
fig. 11 is a graph showing the output variation under dry membrane failure for a certain period of time in a certain cell stack according to example 2 of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of examples of the present invention, and not all examples. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
Example 1
The embodiment provides a fault diagnosis method for a proton exchange membrane fuel cell of a hydrogen energy storage power station based on cloud computing.
The hydrogen energy storage power station structure is shown in fig. 1 and comprises a cloud platform, a hydrogen energy storage device, a PEMFC power generation system, a control and protection system (control system for short), an inverter and the like. The output power of the hydrogen energy storage power station is larger, and the hydrogen energy storage power station is also divided into a plurality of cell stacks for convenient management, the cell stacks form a PEMFC power generation system through series-parallel coupling, and the PEMFC power generation system is connected with an inverter to realize direct-alternating current conversion and grid-connected power control. The control and protection system realizes interaction and protection of a hydrogen energy storage power station, a proton exchange membrane fuel cell and a power grid, and is also a place for realizing the fault diagnosis method, the flow of the fault diagnosis method is shown in figure 2, and the steps are as follows:
1) periodically collecting voltage and current data of all proton exchange membrane fuel cell stacks in an operating state and uploading the data to a cloud end for storage;
2) in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing;
3) in the same sampling period, the identified fuel cell output characteristic model parameters are used as the input quantity of the one-dimensional T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the one-dimensional T-S fuzzy model;
4) analyzing and comparing the variation trend of the output quantity of the one-dimensional T-S fuzzy model, recording the characteristics when the proton exchange membrane fuel cell has faults, and using the characteristics as a historical knowledge base and a reference basis for subsequent fault diagnosis of the proton exchange membrane fuel cell.
Firstly, a fuel cell output characteristic model is established at the cloud based on the relevant characteristics and electrochemical reaction of the proton exchange membrane fuel cell, and the fuel cell output characteristic model comprises a single cell voltage model, an anode gas supply model and a cathode gas supply model.
The fuel cell output characteristic model is established based on the following assumption conditions:
21) all the gas entering the cathode and the anode is fully humidified;
22) all gases belong to ideal gases and accord with the relevant reaction equation of the ideal gases;
23) the internal temperature of the fuel cell changes slowly (fluctuates in the range of 3 ℃), and the internal temperature is kept consistent;
24) neglecting the influence of water on a fuel cell power generation system in a gaseous state;
25) the basic principles of pressure balance and material conservation are followed.
The single cell voltage model includes a nernst voltage model to represent the theoretical maximum output voltage of the fuel cell, a concentration overvoltage model to describe the change in reactant concentration due to mass transfer in the electrochemical reaction, an ohmic polarization overvoltage model to describe the effect of obstruction encountered by protons in the form of hydronium ions across the proton exchange membrane and electrons through the external circuit, an activated polarization overvoltage model caused by electrochemical delay of the electrodes.
The single battery voltage model is specifically:
Figure 919468DEST_PATH_IMAGE006
，
wherein the content of the first and second substances,V cellis the output voltage of the fuel cell and,E nernstis the opening of a fuel cellThe voltage of the circuit is measured by a voltage meter,V actis the activation of the polarization overvoltage,V ohmis an ohmic polarization overvoltage which is a voltage of,V conis a concentration polarization overvoltage.
The Nernst voltage model specifically comprises:
Figure 551438DEST_PATH_IMAGE007
，
wherein, ΔGIs the free energy of Gibbs and the free energy of Gibbs,Fin order to be the faraday constant,Ris a universal gas constant，∆SIn order to be subject to a change in entropy,T ref for the purpose of the reference temperature, the temperature,P 2HandP O2the pressure of hydrogen in the anode and the pressure of oxygen in the cathode respectively,Tindicating the stack temperature.
The concentration overvoltage model specifically comprises the following steps:
Figure 202999DEST_PATH_IMAGE008
，
wherein the content of the first and second substances,i maxin order to achieve the maximum current density,iin order to achieve an actual current density of the stack,Bthe magnitude of the equation coefficient is related to the property of the fuel cell itself, and is a parameter to be identified in the output characteristic model of the fuel cell.
The ohmic polarization overvoltage model specifically comprises the following steps:
Figure 248315DEST_PATH_IMAGE009
，
Figure 643524DEST_PATH_IMAGE010
，
wherein the content of the first and second substances,Irepresents the output current of the electric pile,r m is the electrical resistivity of the proton exchange membrane,l m is the thickness of the proton exchange membrane,Ais the effective activation area of the proton exchange membrane,R C in order to be an electron flow resistance,lis an aqueous proton exchange membraneThe quantity is a parameter to be identified in the fuel cell output characteristic model.
The activated polarization overvoltage model specifically comprises the following steps:
Figure 379268DEST_PATH_IMAGE011
，
wherein the content of the first and second substances,
Figure 467310DEST_PATH_IMAGE012
、
Figure 734343DEST_PATH_IMAGE013
、
Figure 933243DEST_PATH_IMAGE014
、
Figure 274226DEST_PATH_IMAGE015
is the corresponding coefficient in the formula, is the parameter that needs to be identified in the fuel cell output characteristic model;Tis the temperature at which the stack is operated,Ito output the current for the electric pile,C O2the dissolved concentration of oxygen is given by henry's law:
Figure 533169DEST_PATH_IMAGE016
，
wherein the content of the first and second substances,P O2is the oxygen partial pressure;
Figure 287498DEST_PATH_IMAGE017
，
the anode gas supply model specifically comprises the following steps:
Figure 214391DEST_PATH_IMAGE018
，
wherein the content of the first and second substances,m ,anH2andm w,an the mass of hydrogen and water vapor in the anode flow field respectively;W an,inH2,andW v,an,in the flow rates of hydrogen and water vapor flowing into the anode flow field are respectively;W an,outH2,andW v,an,out respectively the flow of hydrogen and water vapor flowing out of the anode flow field;W reactedH2,hydrogen flow participating in the reaction in the electrochemical reaction process;W l,an,out is the flow of liquid water discharged from the anode flow field.
The cathode gas supply model is specifically as follows:
Figure 737776DEST_PATH_IMAGE019
，
wherein the content of the first and second substances,m O2、m N2andm v,ca respectively the mass of oxygen, nitrogen and water vapor in the cathode flow field;W ca,inO2,、W ,ca,inN2、W v,ca,in the mass of oxygen, nitrogen and water vapor flowing into the cathode flow field respectively;W ca,outO2,、W ,ca,outN2、W v,ca,out the mass of oxygen, nitrogen and water vapor flowing out of the cathode flow field respectively;W reactedO2,、W v,gen respectively representing the flow rate of the oxygen reacted and the mass of the generated water vapor in the electrochemical reaction process;W v,mem is the flow of water vapor that migrates through the proton exchange membrane to the cathode flow field.
The hydrogen energy storage power station control system collects voltage and current data of all fuel cell stacks in the running state through communication equipment, uploads the data to the cloud for storage, and the voltage of a single cell of the Kth fuel cell stack is recorded asU K1,U K2,…U nKWhereinnIs the number of cells in the kth stack. The parameter identification of the proton exchange membrane fuel cell output characteristic model is carried out in parallel based on the current and voltage data, the identified fuel cell output characteristic model parameters are stored in the cloud, and the flow chart of the identification is shown in fig. 3.
The method comprises the following specific steps of using a chaotic particle swarm algorithm at the cloud end to perform parameter identification on all electric piles in parallel:
31) the purpose of algorithm optimization is to make the objective function smaller, and the objective optimization function is specifically defined as follows:
Figure 622052DEST_PATH_IMAGE020
，
wherein the content of the first and second substances,RMSEis the root mean square error of the received signal,Nis the number of cells in the stack,V cellis the output voltage of the stack of cells,V i is the output voltage of the fuel cell output characteristic model;
32) defining the size of a population, iteration times and input parameter dimensions, wherein the size of the population is positively correlated with the number of galvanic piles;
33) giving upper and lower limits of parameters to be identified based on historical data experience;
34) creating a chaotic sequence based on the SkewTent mapping, fusing the chaotic sequence with upper and lower parameter limits to generate an initial population, and increasing the optimization probability of a global optimal solution;
35) creating a chaotic disturbance sequence based on Logistic mapping, and fusing the chaotic disturbance sequence with the initial position of each particle individual to obtain a disturbance increment;
36) calculating the fitness value of each particle, and updating the historical individual optimal position and the group optimal position of each particle;
37) updating the speed and position of the particle, calculating the positionXCorresponding fitness value is recorded asfFor updated positions of particlesX′Applying disturbance increment to calculate positionX′Corresponding fitness value is recorded asf′，Comparison offAndf′the size of (a) is smaller than (b),
if it isf<f′The final position of the particle is not changed; if it isf>f′The final position of the particle is updated toX′；
38) Carrying out-of-range processing on the speed and the position, setting parameters exceeding the upper limit as boundary upper limits, and setting parameters exceeding the lower limit as boundary lower limits;
39) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting parameters and a final fitness value; and if not, continuing to repeatedly execute the steps 36) to 38) until the iteration termination condition is met.
Taking a certain stack as an example, the parameters are identified, and the identified parameters are input into the output characteristic model of the fuel cell again, so that the comparison result between the output voltage and the stack voltage is shown in fig. 4 and 5.
In fig. 4, when the input data Pa/Pc is 3/5bar or 1/1bar, the identified parameters are input to the fuel cell output characteristic model to obtain the output voltage (model voltage for short) of the fuel cell output characteristic model, and the output voltage is compared with the stack voltage; in fig. 5, when the input data Pa/Pc is 2.5/3bar and 1.5/1.5bar, the identified parameters are input to the fuel cell output characteristic model to obtain the output voltage (model voltage for short) of the fuel cell output characteristic model, and the output voltage is compared with the stack voltage.
Then normalizing the identified fuel cell output characteristic model parameters to be used as the input quantity of a one-dimensional T-S fuzzy model, wherein the reference fuzzy subsets are 7, and the membership function is as follows:
Figure 265523DEST_PATH_IMAGE021
，
wherein the content of the first and second substances,Brepresenting the identified fuel cell output characteristic model parameters,B n is a specific value corresponding to the fuel cell output characteristic model parameter,nis the number of identified parameters;m Bn,k ands Bn,k is a parameter to be determined by the membership function.
According to the inference rule of the T-S fuzzy model, when 7 fuzzy reference subsets are input, 7 fuzzy reference subsets are output, and the output is as follows:
Figure 787640DEST_PATH_IMAGE022
，
wherein the content of the first and second substances,u output,m a function of membership representing the output,mis a fuzzy reference subset of the output of the one-dimensional T-S fuzzy model.
And identifying undetermined parameters of the membership function by using an ant colony algorithm. The ant colony algorithm target optimization function is the sum of the fuzzy model outputs of the sample points, the ant colony algorithm optimization process is the process of solving the fuzzy model output sum to the maximum, and the flow chart is shown in fig. 6.
The method adopts an optimization algorithm to identify membership function parameters, and comprises the following steps:
41) optimizing the objective function of the algorithm as the output of the fuzzy modeloutputThe purpose of algorithm optimization is to find the optimal parameters to maximize the objective function;
42) the input parameters are normalized, the parameter range is limited to (0,1), and two significant digits are reserved for each parameter;
43) identifying parameters of the membership function by adopting an optimization algorithm;
44) judging whether the iteration times reach the maximum iteration times or whether the error meets the requirement, if so, terminating the algorithm and outputting an optimization parameter and a corresponding objective function value; if not, continue to 43) until the iteration termination requirement is met.
Then analyzing and comparing the variation trend of the output quantity of the fuzzy model, wherein the variation trend of the output quantity of the fuzzy model is represented as:
Figure 516562DEST_PATH_IMAGE023
，
wherein, the first and the second end of the pipe are connected with each other,output(t) Is the output of the fuzzy model at a certain time,Tis the period of the sampling, and,output(t+T) Is the output of the fuzzy model spaced one sampling period apart,Doutputis the amount of change in output. And repeating the steps at the cloud periodically, calculating the output quantity of the fuzzy model, and tracking the change trend of the fuzzy model. The amount of change in the output of the fuzzy model is constantly changing periodically while the fuel cell is operating normally. When the fuel cell fails, the trend of the change in the output quantity of the fuzzy model is also abnormal,the abnormal conditions respectively correspond to different fault states of the fuel cell, the abnormal states are recorded at the cloud end to serve as a historical knowledge base, and in the later working period, when the output quantity of the fuzzy model is abnormal in change, the historical knowledge base is inquired for matching, so that accurate fault diagnosis of the fuel cell is carried out, and accurate fault location of the fuel cell is carried out. If the fault is reversible, the cloud automatically issues an instruction to the fuel cell stack controller, and the operation parameters are adjusted to enable the fuel cell to recover the normal working state. And if the fault is irreversible, issuing a shutdown instruction in time to avoid further damage to the fuel cell. The diagnostic flow is shown in fig. 7. The more times of fault diagnosis, the more data accumulated in the historical knowledge base, and the higher the diagnosis accuracy.
The invention takes the data of 20 voltage and current sample points in a certain sampling period of a certain electric pile as an example, the population size of the chaotic particle swarm algorithm is defined to be 50, and the iteration termination condition is that the iteration times meet 300 or the identification error is less than 10-2The parameter dimension is set to 7 according to the fuel cell output characteristic model. Under a given parameter range, the parameter values identified by the chaotic particle swarm algorithm are shown in the following table:
Figure 533059DEST_PATH_IMAGE024
the input fuzzy reference subset of the fuzzy model is set to be 7, and Z is zero; "VS" is very small; "S" is small; "M" is medium; "B" is large; "VB" is very large; "EB" is very large.
The output fuzzy subsets of the fuzzy model are also set to be 7 in the same way as the input fuzzy subsets. The membership function is defined as a single-valued function, and the order is as follows:
Z= 0；VS= 0.1；S= 0.2；M=0.3；B=0.4；VB=0.5；EB=0.1。
assuming that n sampling cycles are passed, namely n groups of different parameter values (each group is 7-dimensional) are obtained through chaotic particle algorithm identification, normalization processing is carried out on data, and then an ant colony algorithm is adopted to obtain the sum optimal value of fuzzy models of the n groups of parameters, so that the optimal parameter of a membership function is identified; and then calculating the numerical value of the output of the fuzzy model of each sampling period according to the optimal parameters of the membership function.
To illustrate a specific fault determination method, fig. 8 shows a variation curve of the fuzzy model output under the flooding fault in a certain time period of a certain cell stack, and fig. 9 shows a variation curve of the fuzzy model output under the dry membrane fault in a certain time period of a certain cell stack.
As can be seen from fig. 8 and 9, when output falls below 0.8, it means that the stack has failed, but this does not allow a specific fault type to be identified. But according toDoutputThe following correspondence may be established in correspondence with the fault type:Doutput<0.1, a flooding fault occurs;Doutput>=0.1, a film dry failure occurred. And recording the variation trends of output quantities corresponding to two fault states of flooding and membrane trunk as a historical knowledge base, and inquiring the historical knowledge base for matching when the output quantity of the fuzzy model is abnormally changed in the later working period so as to diagnose whether the fuel cell stack has the flooding and membrane trunk faults. For other fault types, a historical knowledge base is also established for judgment.
Because the fault diagnosis is synchronously carried out on the galvanic pile of the power station, the health state of each galvanic pile can be accurately judged according to the output quantity of the fuzzy model calculated by the voltage and current data of the galvanic pile.
Example 2
The embodiment provides a fault diagnosis device for a proton exchange membrane fuel cell of a hydrogen energy storage power station based on cloud computing, which comprises a voltage and current data acquisition unit, a proton exchange membrane fuel cell model parameter identification unit, a one-dimensional T-S fuzzy model unit and a fuel cell fault diagnosis unit.
Voltage and current data acquisition unit: and periodically acquiring voltage and current data of all the proton exchange membrane fuel cell stacks in the running state and uploading the data to a cloud for storage.
Proton exchange membrane fuel cell model parameter identification unit: and in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing.
A one-dimensional T-S fuzzy model construction unit: and in the same sampling period, the identified fuel cell output characteristic model parameters are used as the input quantity of the one-dimensional T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the one-dimensional T-S fuzzy model.
Fuel cell failure diagnosis unit: analyzing and comparing the variation trend of the output quantity of the one-dimensional T-S fuzzy model, recording the characteristics when the proton exchange membrane fuel cell has faults, and using the characteristics as a historical knowledge base and a reference basis for subsequent fault diagnosis of the proton exchange membrane fuel cell.
Specifically, in the proton exchange membrane fuel cell model parameter identification unit, a fuel cell output characteristic model is established as follows, and the fuel cell output characteristic model comprises a single cell voltage model, an anode gas supply model and a cathode gas supply model. The fuel cell output characteristic model is established based on the following assumption conditions:
21) all the gas entering the cathode and the anode is fully humidified;
22) all gases belong to ideal gases and accord with the relevant reaction equation of the ideal gases;
23) the internal temperature of the fuel cell changes slowly (fluctuates in the range of 3 ℃), and the internal temperature is kept consistent;
24) neglecting the influence of water on a fuel cell power generation system in a gaseous state;
25) the basic principles of pressure balance and material conservation are followed.
The single battery voltage model specifically comprises:
Figure 31037DEST_PATH_IMAGE006
，
wherein the content of the first and second substances,V cellis the output voltage of the fuel cell and,E nernstis the open circuit voltage of the fuel cell,V actis the activation of the polarization overvoltage,V ohmis an ohmic polarization overvoltage which is a voltage of,V conis a concentration polarization overvoltage.
The anode gas supply model specifically comprises the following steps:
Figure 537104DEST_PATH_IMAGE025
，
wherein the content of the first and second substances,m H2andm w,an the mass of hydrogen and water vapor in the anode flow field respectively;W an,inH2,andW v,an,in the flow rates of hydrogen and water vapor flowing into the anode flow field are respectively;W an,outH2,andW v,an,out respectively the flow rates of hydrogen and water vapor flowing out of the anode flow field;W reactedH2,hydrogen flow participating in the reaction in the electrochemical reaction process;W l,an,out is the flow of liquid water discharged from the anode flow field.
The cathode gas supply model is specifically as follows:
Figure 753322DEST_PATH_IMAGE019
，
wherein the content of the first and second substances,m O2、m N2andm v,ca respectively the mass of oxygen, nitrogen and water vapor in the cathode flow field;W ca,inO2,、W ,ca,inN2、W v,ca,in the mass of oxygen, nitrogen and water vapor flowing into the cathode flow field respectively;W ca,outO2,、W ,ca,outN2、W v,ca,out the mass of oxygen, nitrogen and water vapor flowing out of the cathode flow field respectively;W reactedO2,、W v,gen respectively representing the flow rate of the oxygen reacted and the mass of the generated water vapor in the electrochemical reaction process;W v,mem is the flow of water vapor that migrates through the proton exchange membrane to the cathode flow field.
The proton exchange membrane fuel cell model parameter identification unit identifies fuel cell voltage model parameters based on an improved chaotic particle swarm optimization algorithm in parallel at the cloud, and comprises the following steps:
31) the purpose of algorithm optimization is to make the objective function smaller, and the objective optimization function is specifically defined as follows:
Figure 117810DEST_PATH_IMAGE002
，
wherein the content of the first and second substances,RMSEis the root mean square error of the received signal,Nis the number of cells in the stack,V cellis the output voltage of the electric pile,V i is the output voltage of the fuel cell output characteristic model;
32) defining the size of a population, iteration times and input parameter dimensions, wherein the size of the population is positively correlated with the number of galvanic piles;
33) giving upper and lower limits of parameters to be identified based on historical data experience;
34) creating a chaotic sequence based on the SkewTent mapping, fusing the chaotic sequence with upper and lower parameter limits to generate an initial population, and increasing the optimization probability of a global optimal solution;
35) creating a chaotic disturbance sequence based on Logistic mapping, and fusing the chaotic disturbance sequence with the initial position of each particle individual to obtain a disturbance increment;
36) calculating the fitness value of each particle, and updating the historical individual optimal position and the group optimal position of each particle;
37) updating the speed and position of the particle, calculating the positionXCorresponding fitness value is recorded asfFor updated positions of particlesX′Applying disturbance increment to calculate positionX′Corresponding fitness value is recorded asf′，ComparisonfAndf′the size of (a) is (b),
if it isf<f′The final position of the particle is not changed; if it isf>f′The final position of the particle is updated toX′；
38) Carrying out-of-range processing on the speed and the position, setting parameters exceeding the upper limit as boundary upper limits, and setting parameters exceeding the lower limit as boundary lower limits;
39) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting parameters and a final fitness value; and if not, continuing to repeatedly execute the steps 36) to 38) until the iteration termination condition is met.
In the one-dimensional T-S fuzzy model construction unit, the identified parameters of the fuel cell output characteristic model are normalized and used as the input quantity of the one-dimensional T-S fuzzy model, the number of fuzzy reference subsets is L, the membership function parameters are identified by adopting an optimization algorithm and input into the fuzzy model, and the fuzzy reference subsets which belong to the input of each one-dimensional T-S fuzzy modelkMembership function ofu Bn,k Comprises the following steps:
Figure 493428DEST_PATH_IMAGE026
，
wherein the content of the first and second substances,Brepresenting the identified fuel cell output characteristic model parameters,B n is a specific value corresponding to the fuel cell output characteristic model parameter,nis the number of identified parameters;m Bn,k ands Bn,k is a membership function parameter to be determined;
according to the inference rule of the T-S fuzzy model, when L fuzzy reference subsets are input, L fuzzy reference subsets are output, and the output is as follows:
Figure 914045DEST_PATH_IMAGE027
；
wherein the content of the first and second substances,u output,m a function of membership representing the output,mis a fuzzy reference subset of the output of the one-dimensional T-S fuzzy model.
The method adopts an optimization algorithm to identify membership function parameters, and comprises the following steps:
41) optimizing the objective function of the algorithm as the output of the fuzzy modeloutputThe purpose of algorithm optimization is to find the optimal parameters to maximize the objective function;
42) the input parameters are normalized, the parameter range is limited to (0,1), and two significant digits are reserved for each parameter;
43) identifying parameters of the membership function by adopting an optimization algorithm;
44) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting optimization parameters and corresponding objective function values; if not, continue to 43) until the iteration termination requirement is met.
In the fuel cell fault diagnosis unit, the variation trend of the output quantity of the fuzzy model is analyzed and compared, and the variation trend of the output quantity of the fuzzy model is expressed as follows:
Figure 386615DEST_PATH_IMAGE023
，
wherein the content of the first and second substances,output(t) Is the output of the fuzzy model at a certain time,Tis the period of the sampling, and,output(t+T) Is the output of the fuzzy model spaced one sampling period apart,Doutputis the amount of change in output.
The fuel cell fault diagnosis unit is characterized in that when the fuel cell fails, the output quantity variation trend of the fuzzy model is abnormal, the abnormal conditions correspond to different fault states of the fuel cell respectively, the abnormal states are recorded at the cloud end to serve as a historical knowledge base, and in the following working period, when the output quantity variation of the fuzzy model is abnormal, the historical knowledge base is inquired for matching, so that accurate fault diagnosis of the fuel cell is performed, and accurate fault positioning of the fuel cell is performed. If the fault is reversible, an instruction is automatically issued to the fuel cell stack controller, and the operation parameters are adjusted to enable the fuel cell to recover the normal working state. And if the fault is irreversible, issuing a shutdown instruction in time to avoid further damage to the fuel cell. The more times of fault diagnosis, the more data accumulated in the historical knowledge base, and the higher the diagnosis accuracy.
The invention defines chaotic particle swarm optimization by taking the data of 20 voltage and current sample points in a certain sampling period of a certain electric pile as an exampleThe population size is 50, the iteration termination condition is that the iteration number satisfies 300 or the identification error is less than 10-2The parameter dimension is set to 7 according to the output characteristic model. Under a given parameter range, the parameter values identified by the chaotic particle swarm algorithm are shown in the following table:
Figure 421436DEST_PATH_IMAGE028
the input fuzzy reference subset of the fuzzy model is set to be 7, and Z is zero; "VS" is very small; "S" is small; "M" is medium; "B" is large; "VB" is very large; "EB" is very large.
The output fuzzy subsets of the fuzzy model are also set to be 7 in the same way as the input fuzzy subsets. The membership function is defined as a single-valued function, and the order is as follows:
Z=0；VS=0.1；S=0.2；M=0.3；B=0.4；VB=0.5；EB=0.1。
assuming that n sampling cycles are passed, namely n groups of different parameter values (each group is 7-dimensional) are obtained through chaotic particle algorithm identification, normalization processing is carried out on data, and then an ant colony algorithm is adopted to obtain the sum optimal value of fuzzy models of the n groups of parameters, so that the optimal parameter of the membership function is identified. And then calculating the numerical value of the output of the fuzzy model of each sampling period according to the optimal parameters of the membership function.
To illustrate specific fault determination, fig. 10 shows the variation curve of the fuzzy model output under the flooding fault in a certain time period of a certain cell stack, and fig. 11 shows the variation curve of the fuzzy model output under the dry membrane fault in a certain time period of a certain cell stack.
As can be seen from fig. 10 and 11, when output falls below 0.8, this means that the stack has failed, but this does not yet allow a specific fault type to be identified. But according toDoutputThe following correspondence may be established in correspondence with the fault type:Doutput<0.1, a flooding fault occurs;Doutput>=0.1, a film dry failure occurred. Record flooding and Membrane drynessThe output quantity variation trends corresponding to the two fault states are used as a historical knowledge base, and in the following working period, when the output quantity variation of the fuzzy model is abnormal, the historical knowledge base is inquired for matching, so that whether the fuel cell stack has the flooding or the dry membrane fault can be diagnosed. For other fault types, a historical knowledge base is also established for judgment.
Because the electric piles of the power station are subjected to fault diagnosis synchronously, the health state of each electric pile can be accurately judged according to the output quantity of the fuzzy model obtained by calculating the voltage and current data of the electric piles.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (10)

1. The fault diagnosis method for the proton exchange membrane fuel cell of the hydrogen energy storage power station is characterized by comprising the following steps of:
1) periodically collecting voltage and current data of all proton exchange membrane fuel cell stacks in an operating state and uploading the data to a cloud end for storage;
2) in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing;
3) in the same sampling period, the identified fuel cell output characteristic model parameters are used as the input quantity of the one-dimensional T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the one-dimensional T-S fuzzy model;
4) analyzing and comparing the variation trend of the output quantity of the one-dimensional T-S fuzzy model, recording the characteristics when the proton exchange membrane fuel cell has faults, and using the characteristics as a historical knowledge base and a reference basis for subsequent fault diagnosis of the proton exchange membrane fuel cell.
2. The fault diagnosis method for the proton exchange membrane fuel cell of the hydrogen energy storage power station as claimed in claim 1, wherein in the step 2), the fuel cell output characteristic model is established according to the following method, the fuel cell output characteristic model comprises a single cell voltage model, an anode gas supply model and a cathode gas supply model, and the fuel cell output characteristic model is established based on the following assumed conditions:
21) all the gas entering the cathode and the anode is fully humidified;
22) all gases belong to ideal gases and accord with the relevant reaction equation of the ideal gases;
23) the internal temperature of the fuel cell changes slowly, and the internal temperature is kept consistent;
24) neglecting the influence of water on a fuel cell power generation system in a gaseous state;
25) the basic principles of pressure balance and material conservation are followed.
3. The fault diagnosis method for the proton exchange membrane fuel cell of the hydrogen energy storage power station as claimed in claim 1, wherein in the step 2), the specific steps of using the chaotic particle swarm algorithm at the cloud end to perform parameter identification on all the electric piles in parallel are as follows:
31) the purpose of algorithm optimization is to make the objective function smaller, and the objective optimization function is specifically defined as follows:
Figure 105739DEST_PATH_IMAGE002
，
wherein the content of the first and second substances,RMSEis the root mean square error of the received signal,Nis the number of cells in the stack,V cellis the output voltage of the stack of cells,V i is the output voltage of the fuel cell output characteristic model;
32) defining the size of a population, iteration times and input parameter dimensions, wherein the size of the population is positively correlated with the number of galvanic piles;
33) giving upper and lower limits of parameters to be identified based on historical data experience;
34) creating a chaotic sequence based on the SkewTent mapping, fusing the chaotic sequence with upper and lower parameter limits to generate an initial population, and increasing the optimization probability of a global optimal solution;
35) creating a chaotic disturbance sequence based on Logistic mapping, and fusing the chaotic disturbance sequence with the initial position of each particle individual to obtain a disturbance increment;
36) calculating the fitness value of each particle, and updating the historical individual optimal position and the group optimal position of each particle;
37) updating the speed and position of the particle, calculating the positionXCorresponding fitness value is recorded asfFor updated positions of particlesX′Applying disturbance increment to calculate positionX′Corresponding fitness value is recorded asf′，ComparisonfAndf′the size of (a) is (b),
if it isf<f′The final position of the particle is not changed; if it isf>f′The final position of the particle is updated toX′；
38) Carrying out-of-range processing on the speed and the position, setting parameters exceeding the upper limit as boundary upper limits, and setting parameters exceeding the lower limit as boundary lower limits;
39) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting parameters and a final fitness value; and if not, continuing to repeatedly execute the steps 36) to 38) until the iteration termination condition is met.
4. The method for diagnosing the faults of the proton exchange membrane fuel cell of the hydrogen energy storage power station as claimed in claim 1, wherein in the step 3), the identified parameters of the fuel cell output characteristic model are normalized and then used as input quantities of a one-dimensional T-S fuzzy model, the number of fuzzy reference subsets is L, the membership function parameters are identified by adopting an optimization algorithm and input into the fuzzy model, and the fuzzy reference subsets input by the one-dimensional T-S fuzzy model belong tokMembership function ofu Bn,k Comprises the following steps:
Figure 251856DEST_PATH_IMAGE003
，
wherein the content of the first and second substances,Brepresenting identified firesThe parameters of the output characteristic model of the fuel cell,B n is a specific value corresponding to the fuel cell output characteristic model parameter,nis the number of identified parameters;m Bn,k ands Bn,k is a membership function parameter to be determined;
according to the inference rule of the T-S fuzzy model, when L fuzzy reference subsets are input, L fuzzy reference subsets are output, and the output is as follows:
Figure 347988DEST_PATH_IMAGE004
；
wherein the content of the first and second substances,u output,m a function of membership representing the output,mis a fuzzy reference subset output by the one-dimensional T-S fuzzy model;
the method adopts an optimization algorithm to identify membership function parameters, and comprises the following steps:
41) optimizing the objective function of the algorithm as the output of the fuzzy modeloutputThe purpose of algorithm optimization is to find the optimal parameters to maximize the objective function;
42) the input parameters are normalized, the parameter range is limited to (0,1), and two significant digits are reserved for each parameter;
43) identifying parameters of the membership function by adopting an optimization algorithm;
44) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting optimization parameters and corresponding objective function values; if not, continue to 43) until the iteration termination requirement is met.
5. The method for diagnosing the faults of the proton exchange membrane fuel cell of the hydrogen energy storage power station as claimed in claim 4, wherein in the step 4), the variation trend of the output quantity of the fuzzy model is analyzed and compared, and the variation trend of the output quantity of the fuzzy model is expressed as:
Figure 59592DEST_PATH_IMAGE005
，
wherein the content of the first and second substances,output(t) Is at a certain momenttThe output of the fuzzy model is then used,Tis the period of the sampling, and,output(t+T) Is the output of the fuzzy model spaced one sampling period apart,Doutputis the amount of change in output;
when the fuel cell fails, the change trend of the output quantity of the fuzzy model is abnormal, the abnormal conditions respectively correspond to different fault states of the fuel cell, the abnormal states are recorded at the cloud end to serve as a historical knowledge base, and in the following working period, when the output quantity of the fuzzy model is abnormal, the historical knowledge base is inquired for matching, so that accurate fault diagnosis of the fuel cell is carried out, and accurate fault positioning of the fuel cell is carried out; if the fault is reversible, automatically issuing an instruction to a fuel cell stack controller, and adjusting operation parameters to enable the fuel cell to recover the normal working state; and if the fault is irreversible, issuing a shutdown instruction in time to avoid further damage to the fuel cell.
6. Proton exchange membrane fuel cell fault diagnosis device of hydrogen energy storage power station, its characterized in that includes:
voltage and current data acquisition unit: periodically collecting voltage and current data of all proton exchange membrane fuel cell stacks in an operating state and uploading the data to a cloud end for storage;
proton exchange membrane fuel cell model parameter identification unit: in the same sampling period, based on voltage and current data, performing fuel cell output characteristic model parameter identification on all the electric piles in parallel by using a chaotic particle swarm algorithm at the cloud end, and storing;
a one-dimensional T-S fuzzy model construction unit: in the same sampling period, the identified fuel cell output characteristic model parameters are used as the input quantity of the one-dimensional T-S fuzzy model, and membership function parameters are identified by adopting an optimization algorithm and input into the one-dimensional T-S fuzzy model;
fuel cell failure diagnosis unit: analyzing and comparing the variation trend of the output quantity of the one-dimensional T-S fuzzy model, recording the characteristics when the proton exchange membrane fuel cell has faults, and using the characteristics as a historical knowledge base and a reference basis for subsequent fault diagnosis of the proton exchange membrane fuel cell.
7. The proton exchange membrane fuel cell fault diagnosis device of the hydrogen energy storage power station as claimed in claim 6, wherein the proton exchange membrane fuel cell model parameter identification unit is configured to establish a fuel cell output characteristic model, the fuel cell output characteristic model comprises a voltage model, an anode gas supply model and a cathode gas supply model, and the fuel cell output characteristic model is established based on the following assumed conditions:
21) all the gas entering the cathode and the anode is fully humidified;
22) all gases belong to ideal gases and accord with the relevant reaction equation of the ideal gases;
23) the internal temperature of the fuel cell changes slowly, and the internal temperature is kept consistent;
24) neglecting the influence of water on a fuel cell power generation system in a gaseous state;
25) the basic principles of pressure balance and material conservation are followed.
8. The proton exchange membrane fuel cell fault diagnosis device of the hydrogen energy storage power station as claimed in claim 6, wherein the proton exchange membrane fuel cell model parameter identification unit performs parameter identification on all the electric piles in parallel by using the chaotic particle swarm optimization at the cloud end by the specific steps of:
31) the purpose of algorithm optimization is to make the objective function smaller, and the objective optimization function is specifically defined as follows:
Figure 584749DEST_PATH_IMAGE002
，
wherein the content of the first and second substances,RMSEis the root mean square error of the received signal,Nis the number of cells in the stack,V cellis the output voltage of the stack of cells,V i is the output voltage of the fuel cell output characteristic model;
32) defining the size of a population, iteration times and input parameter dimensions, wherein the size of the population is positively correlated with the number of galvanic piles;
33) giving upper and lower limits of parameters to be identified based on historical data experience;
34) creating a chaotic sequence based on the SkewTent mapping, fusing the chaotic sequence with upper and lower parameter limits to generate an initial population, and increasing the optimization probability of a global optimal solution;
35) creating a chaotic disturbance sequence based on Logistic mapping, and fusing the chaotic disturbance sequence with the initial position of each particle individual to obtain a disturbance increment;
36) calculating the fitness value of each particle, and updating the historical individual optimal position and the group optimal position of each particle;
37) updating the speed and position of the particle, calculating the positionXCorresponding fitness value is recorded asfFor updated positions of particlesX′Applying disturbance increment to calculate positionX′Corresponding fitness value is recorded asf′，ComparisonfAndf′the size of (a) is (b),
if it isf<f′The final position of the particle is not changed; if it isf>f′The final position of the particle is updated toX′；
38) Carrying out-of-range processing on the speed and the position, setting parameters exceeding the upper limit as boundary upper limits, and setting parameters exceeding the lower limit as boundary lower limits;
39) judging whether the iteration times meet the requirements or whether the errors meet the requirements, if so, terminating the algorithm and outputting parameters and a final fitness value; and if not, continuing to repeatedly execute the steps 36) to 38) until the iteration termination condition is met.
9. The proton exchange membrane fuel cell fault diagnosis device of the hydrogen energy storage power station as claimed in claim 6, wherein in the one-dimensional T-S fuzzy model construction unit, the identified parameters of the fuel cell output characteristic model are normalized and used as the input quantity of the one-dimensional T-S fuzzy model, the number of fuzzy reference subsets is L, and an optimization algorithm is adopted to identify the parameters of the membership functionAnd input into fuzzy model, belonging to fuzzy reference subset input by each one-dimensional T-S fuzzy modelkMembership function ofu Bn,k Comprises the following steps:
Figure 834465DEST_PATH_IMAGE006
，
wherein the content of the first and second substances,Brepresenting the identified fuel cell output characteristic model parameters,B n is a specific value corresponding to the fuel cell output characteristic model parameter,nis the number of identified parameters;m Bn,k ands Bn,k is a membership function parameter to be determined;
according to the inference rule of the T-S fuzzy model, when L fuzzy reference subsets are input, L fuzzy reference subsets are output, and the output is as follows:
Figure 6689DEST_PATH_IMAGE007
；
wherein the content of the first and second substances,u output,m a function of membership representing the output,mis a fuzzy reference subset output by a one-dimensional T-S fuzzy model;
the method for identifying the membership function parameters by adopting the optimization algorithm comprises the following steps:
41) optimizing the objective function of the algorithm as the output of the fuzzy modeloutputThe purpose of algorithm optimization is to find the optimal parameters to maximize the objective function;
42) the input parameters are normalized, the parameter range is limited to (0,1), and two significant digits are reserved for each parameter;
43) identifying parameters of the membership function by adopting an optimization algorithm;
44) judging whether the iteration times reach the maximum iteration times or whether the error meets the requirement, if so, terminating the algorithm and outputting an optimization parameter and a corresponding objective function value; if not, continue to 43) until the iteration termination requirement is met.
10. The proton exchange membrane fuel cell fault diagnosis device of the hydrogen energy storage power station as claimed in claim 9, wherein in the fuel cell fault diagnosis unit, the variation trend of the output quantity of the fuzzy model is analyzed and compared, and the variation trend of the output quantity of the fuzzy model is expressed as:
Figure 316447DEST_PATH_IMAGE008
，
wherein, the first and the second end of the pipe are connected with each other,output(t) Is at a certain momenttThe output of the fuzzy model is then used,Tis the period of the sampling, and,output(t+T) Is the output of the fuzzy model spaced one sampling period apart,Doutputis the amount of change in output;
when the fuel cell fails, the change trend of the output quantity of the fuzzy model is abnormal, the abnormal conditions respectively correspond to different fault states of the fuel cell, the abnormal states are recorded at the cloud end to serve as a historical knowledge base, and in the following working period, when the output quantity of the fuzzy model is abnormal, the historical knowledge base is inquired for matching, so that accurate fault diagnosis of the fuel cell is carried out, and accurate fault positioning of the fuel cell is carried out; if the fault is reversible, automatically issuing an instruction to a fuel cell stack controller, and adjusting operation parameters to enable the fuel cell to recover the normal working state; and if the fault is irreversible, issuing a shutdown instruction in time to avoid further damage to the fuel cell.
CN202210342798.2A 2022-04-02 2022-04-02 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station Active CN114492087B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210342798.2A CN114492087B (en) 2022-04-02 2022-04-02 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210342798.2A CN114492087B (en) 2022-04-02 2022-04-02 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station

Publications (2)

Publication Number Publication Date
CN114492087A true CN114492087A (en) 2022-05-13
CN114492087B CN114492087B (en) 2022-07-19

Family

ID=81488560

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210342798.2A Active CN114492087B (en) 2022-04-02 2022-04-02 Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station

Country Status (1)

Country Link
CN (1) CN114492087B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115966739A (en) * 2023-03-13 2023-04-14 湖北工业大学 Fuel cell parameter identification method and system based on improved hippocampal optimization algorithm
CN116504122A (en) * 2023-06-29 2023-07-28 武汉理工大学 Proton exchange membrane fuel cell teaching experiment platform based on metauniverse
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 (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6735576B1 (en) * 1999-09-14 2004-05-11 Yamaha Hatsudoki Kabushiki Kaisha Method of real-time optimization of control device equipped with fuzzy inference system
DE102011007434A1 (en) * 2011-04-14 2012-10-18 Suzlon Energy Gmbh Method for creating software-simulation module for simulating correlation between input and output variables and/or behavior of wind energy plant in energy supply network, involves forming model from transfer and combination functions
US20140176145A1 (en) * 2012-12-20 2014-06-26 Hyundai Autron Co., Ltd. Method and apparatus for diagnosing fault of fuel cell stack based on measurement of stack current
CN106230377A (en) * 2016-07-01 2016-12-14 重庆大学 A kind of photovoltaic battery panel hot spot fault detection method
CN110579709A (en) * 2019-08-30 2019-12-17 西南交通大学 fault diagnosis method for proton exchange membrane fuel cell for tramcar
CN110705657A (en) * 2019-11-21 2020-01-17 北京交通大学 Mode identification fault diagnosis method of proton exchange membrane fuel cell system
CN111261903A (en) * 2020-01-21 2020-06-09 同济大学 Model-based proton exchange membrane fuel cell impedance online estimation method
WO2021213989A1 (en) * 2020-04-24 2021-10-28 Robert Bosch Gmbh Fuel cell diagnostic apparatus and corresponding diagnostic method, vehicle, and vehicle system
CN113780537A (en) * 2021-11-12 2021-12-10 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 power generation system
US20220029184A1 (en) * 2019-04-02 2022-01-27 Zhejiang University Active fault-tolerant temperature control method for proton exchange membrane fuel cell system

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6735576B1 (en) * 1999-09-14 2004-05-11 Yamaha Hatsudoki Kabushiki Kaisha Method of real-time optimization of control device equipped with fuzzy inference system
DE102011007434A1 (en) * 2011-04-14 2012-10-18 Suzlon Energy Gmbh Method for creating software-simulation module for simulating correlation between input and output variables and/or behavior of wind energy plant in energy supply network, involves forming model from transfer and combination functions
US20140176145A1 (en) * 2012-12-20 2014-06-26 Hyundai Autron Co., Ltd. Method and apparatus for diagnosing fault of fuel cell stack based on measurement of stack current
CN106230377A (en) * 2016-07-01 2016-12-14 重庆大学 A kind of photovoltaic battery panel hot spot fault detection method
US20220029184A1 (en) * 2019-04-02 2022-01-27 Zhejiang University Active fault-tolerant temperature control method for proton exchange membrane fuel cell system
CN110579709A (en) * 2019-08-30 2019-12-17 西南交通大学 fault diagnosis method for proton exchange membrane fuel cell for tramcar
CN110705657A (en) * 2019-11-21 2020-01-17 北京交通大学 Mode identification fault diagnosis method of proton exchange membrane fuel cell system
CN111261903A (en) * 2020-01-21 2020-06-09 同济大学 Model-based proton exchange membrane fuel cell impedance online estimation method
WO2021213989A1 (en) * 2020-04-24 2021-10-28 Robert Bosch Gmbh Fuel cell diagnostic apparatus and corresponding diagnostic method, vehicle, and vehicle system
CN113780537A (en) * 2021-11-12 2021-12-10 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 power generation system

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
YANG YANG等: "Modeling of PEMFC and Analysis of Multiple Influencing Factors on Output Characteristics", 《JOURNAL OF THE ELECTROCHEMICAL SOCIETY》 *
全睿: "车用燃料电池系统故障诊断与维护若干关键问题研究", 《中国优秀博硕士学位论文全文数据库(博士)工程科技Ⅱ辑(月刊)》 *
叶宗俊等: "车用燃料电池系统建模及参数辨识", 《机电一体化》 *
郭家兴等: "质子交换膜燃料电池故障诊断", 《电源技术》 *
陈维荣等: "质子交换膜燃料电池故障诊断方法综述及展望", 《中国电机工程学报》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115966739A (en) * 2023-03-13 2023-04-14 湖北工业大学 Fuel cell parameter identification method and system based on improved hippocampal optimization algorithm
CN115966739B (en) * 2023-03-13 2023-06-02 湖北工业大学 Fuel cell parameter identification method and system based on improved Hippocampus optimization algorithm
CN116504122A (en) * 2023-06-29 2023-07-28 武汉理工大学 Proton exchange membrane fuel cell teaching experiment platform based on metauniverse
CN116504122B (en) * 2023-06-29 2023-10-24 武汉理工大学 Proton exchange membrane fuel cell teaching experiment platform based on metauniverse
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
CN114492087B (en) 2022-07-19

Similar Documents

Publication Publication Date Title
CN114492087B (en) Fault diagnosis method and device for proton exchange membrane fuel cell of hydrogen energy storage power station
Lu et al. On-line fault diagnosis for proton exchange membrane fuel cells based on a fast electrochemical impedance spectroscopy measurement
CN108344947A (en) A kind of fuel cell diagnostic method of non-intrusion type
CN114899457B (en) Fault detection method for proton exchange membrane fuel cell system
CN108957335B (en) SOC online estimation method of 2kW/10 kW.h all-vanadium redox flow battery
WO2013083872A1 (en) Method and arrangement for indicating solid oxide cell operating conditions
CN112117475A (en) Fault detection device and method for water management subsystem of fuel cell
CN111916803A (en) Fuel cell system based on cloud intelligent monitoring and edge computing
CN114551944A (en) Method and system for rapidly controlling water content in proton exchange membrane fuel cell
Peng et al. Generalized spatial–temporal fault location method for solid oxide fuel cells using LSTM and causal inference
CN113611900A (en) Membrane dry fault diagnosis method of proton exchange membrane fuel cell
CN108680867B (en) Online SOC calibration method of all-vanadium redox flow battery based on capacity correction
Xiao et al. Water management fault diagnosis for proton-exchange membrane fuel cells based on deep learning methods
CN114583219A (en) Water flooding fault diagnosis method for proton exchange membrane fuel cell
Quan et al. Fault diagnosis in a current sensor and its application to fault-tolerant control for an air supply subsystem of a 50 kW-Grade fuel cell engine
WO2013083873A1 (en) Method and arrangement for diagnosis of solid oxide cells operating conditions
Li Data-driven fault diagnosis for PEMFC systems
CN116231013A (en) Method for establishing fault embedded model of fuel cell
Kahia et al. Neural network based diagnostic of PEM fuel cell
CN115799673A (en) Energy storage battery management system with multidimensional sensing
Shi et al. PEMFC Fault Diagnosis Based on an Equivalent Circuit and OS-ELM Framework
Wang et al. Diagnosis of Gas Leakage in SOFC Stack Based on Gas Sensing Data
Mora et al. Development and experimental validation of the dynamic model of an electrolyser for its integration with renewable generation
CN113659174B (en) Method for evaluating capability index of membrane electrode manufacturing process
CN213692124U (en) Fuel cell control system

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