Research And Design On Safety Monitoring System And Fault Diagnosis Of Hydrogen Fuel Cell

As the monitoring module of hydrogen fuel cell stack,the safety monitoring system of hydrogen fuel cell undertakes the task of collecting real-time parameters in the operation of the stack,and plays a very important role in the monitoring,control,and fault diagnosis of the operation of the stack.The fault diagnosis ensures the timely discovery and treatment of the fuel cell system fault.Therefore,the safety monitoring system and fault diagnosis research of hydrogen fuel cell has practical significance to realize.As a typical multi input and output nonlinear system,the normal operation of the hydrogen fuel cell power system and the function realization of each subsystem depend on the real-time parameters collected by the safety monitoring system and the feedback results of fault diagnosis.To realize the relevant requirements of the system for monitoring and diagnosis,the fuel cell power system of heavy truck is taken as the research object in this paper,designs and develops the hydrogen fuel cell safety monitoring system.In the aspect of fault diagnosis,the fault diagnosis method based on machine learning algorithms which are studied and analyzed.Based on the feature extraction ability of unsupervised extreme learning machine(US-ELM),the clustering ability of K-means and the incremental learning ability of online sequential extreme learning machine(OS-ELM),the fault diagnosis system of hydrogen fuel cell is established to realize the incremental learning and online diagnosis of fault diagnosis model It has the function of power off and so on.In the aspect of safety monitoring system of hydrogen fuel cell,the basic principle of proton exchange membrane fuel cell and the structure and function of fuel cell power system are studied and analyzed.The overall structure of the safety monitoring system of hydrogen fuel cell is designed.Using the overall structure design as the backbone,based on the parameters of the standard stack and the calculation results of the related parameters,the sensors of the safety monitoring system of the hydrogen fuel cell are selected.The schematic design of the relevant modules is completed,the protocol of serial communication is analyzed,and the data frame used in the communication of the system is designed to realize the software and hardware design of the whole safety monitoring system of the hydrogen fuel cell.Based on the hardware design,the lower computer is optimized by Altium Designer,and the schematic diagram is drawn,and the PCB of the lower computer is designed and manufactured.The monitoring system of the upper computer is built based on Lab VIEW,which realizes the functions of parameter monitoring,data recording and fault diagnosis.In the field of fault diagnosis,based on the original data of hydrogen fuel cell stack collected in the experiment,the relevant research is carried out to solve the needs of on-line diagnosis and model updating for fault diagnosis.Based on the analysis of relevant diagnosis literature and algorithm principle,a fault diagnosis system is built to meet the needs of incremental learning and online fault diagnosis.The system extracts manifold regular features through US-ELM,realizes the training of fault diagnosis model and model update based on incremental learning through OS-ELM.At the same time,K-means clustering algorithm is introduced to assist in the solution of data marking problem in incremental learning.Based on the 20-d original data,the validation sample set is constructed.The comparison between US-ELM and OS-ELM algorithm with their similar algorithms is carried out.The analysis results prove that US-ELM and OS-ELM have advantages in the visual results of feature extraction,clustering accuracy,fault diagnosis accuracy and diagnosis time,thus verifying the effectiveness of the whole diagnosis system.