Font Size: a A A

Research On Water Fault Diagnosis Method For Fuel Cell

Posted on:2020-03-04Degree:MasterType:Thesis
Country:ChinaCandidate:L JiangFull Text:PDF
GTID:2381330599476004Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
Environmental issues,energy crises,and climate warming have made the study of new energy an important way to solve problems.To this end,countries have carried out various research and application studies to explore various possible sources of energy.Proton Exchange Membrane Fuel Cell(PEMFC)has become one of the most promising new energy batteries with its excellent performances such as fast start-up,low operating temperature,high power density and no pollution of products.At present,fuel cell applications have been gradually commercialized,but due to technical and theoretical limitations,there are few research methods for water fault diagnosis in fuel cells in China.In the development of fuel cells,a reasonable and feasible water fault diagnosis method is proposed to help effectively monitor the internal working state of the battery and respond in time.The timely fault handling can reduce the application cost of the fuel cell,and the reasonable maintenance can prolong the fuel.Battery life.Therefore,this paper studies the PEMFC water fault diagnosis method,the main work is as follows:(1)Analyze the intrinsic relationship between the reaction process of the fuel cell working electrode and the equivalent circuit model,and establish an equivalent circuit model with different complexity.The formation mechanism of Electrochemical Impedance Spectroscopy(EIS)was studied to determine the equivalence between EIS and equivalent model.In-depth analysis of the physical causes of water faults,determination of water faults,EIS and equivalent circuit parameters between the changes,based on this,established a PEMFC water fault diagnosis system topology.(2)Focus on the online EIS acquisition method,which is based on the traditional single-frequency sweep time,can not reflect the current status of the fuel cell in time,and the peak value of the mixed signal is too high,which is easy to destroy the real-time working state of the fuel cell.The Discrete Interval Binary Sequence(DIBS)pulse excitation signal is used to build the PEMFC working model through MATLAB/Simulink for simulation verification.Based on the simulation,the semi-physical test platform is built to verify the DIBS pulse excitation signal.The results show that: DIBS The experimental results of the pulse excitation signal can well follow the system power change and take a short time,which can be used for EIS online measurement.(3)Based on the EIS measurement experiment,the problem of the internal structure of fuel cell exchange membrane and electrode is difficult to observe.This paper uses Nelder-Mead optimization algorithm to identify the third-order RQ equivalent circuit.The identification result can be accurately Reflects the results of nonlinear experiments.By comparing with the least squares method and genetic algorithm in accuracy,speed and anti-interference ability,the advantages of Nelder-Mead algorithm and the feasibility of online identification are determined.(4)Combining the existing experimental data and fault characteristics to simulate the simulation of the model,using the Randles model parameters and the important feature points in the EIS and Bode diagrams as the sample feature attributes of the 90 sets of simulated faults,and the feature attributes after the variance screening After the sample data is processed,it is input into the Support Vector Machine(SVM)optimized by Particle Swarm Optimization(PSO)to classify the faults,and the classification results are good.Based on the above research and verification work,the accuracy and speed of each link of the fault diagnosis topology proposed in this paper can reach a high standard,and it is feasible to apply it to online fault diagnosis in practical engineering.
Keywords/Search Tags:Proton exchange membrane fuel cell, Electrochemical impedance spectroscopy, DIBS pulse excitation, Nelder-Mead optimization, Parameter identification, SVM fault classification
PDF Full Text Request
Related items