Research On Online Monitoring System For Each Fuel Cell Impedance And Soft-sensing The PEM Water Content

Fuel cell is one of the most potential green generator appliances and is the hot topic in new energy sources research. Fuel cell stack is generally made of multi single cell chip connected in series. In its operation period, the abnormal of single chip will affect the performance and safety of whole fuel cell stack. So for the evaluation of the performance of single chip and find out the abnormal in time, it is must to monitor the internal impedance and voltage of every single chip. Ohmic resistance of PEMFC is most decided by PEM water content so that it can affect the efficiency of generate electricity. Focus on the problems above, this paper develops the design of the internal impedance and voltage of every single chip online monitoring system and soft-sensing research of average PEM water content. The main contents as follows:In this paper, the equivalent circuit model and experiment model of single fuel cell chip are illustrated, the formation of PEMFC internal impedance and the characteristics of which are analyzed first, and then the author compared the advantages and disadvantages of domestic and international methods of the fuel cell internal impedance measurement in presence, and basing on the EIS method developed the general online measuring system to monitor the internal impedance and voltage of every single chip.Upon the general system, the hardware platform basing on the core of high properties DSP is constructed, high accuracy signal inspection and process circuit, AD transfer circuit, SCI communication extent circuit and CAN communication circuit are designed detailedly, the design principles and choosing standards of parameters in these circuits are analyzed.On the basis of analysis of software requirement, the author divided system software to some function subprograms. Basing on DSP software develop platform CCS, the author designed the system initial program, AD transfer program and interrupt program, made the detail application layer protocols of SCI and CAN communication. At the same time, the author used VB language to design the perfect function software of PC.Because the PEM water content is difficult to measure directly, the author analyzed the mathematic relationship between PEM water content and its ohmic resistance firstly, and then proposed the soft-sensing method to measure PEM water content basing on RBF neural network, researched data preprocess method of auxiliary variables in the model, the soft-sensing model of RBF network is simulated lastly, the simulation results show good effectiveness of the proposed method to forecast the average PEM water content in whole stack.The whole circuit is simple and clear, the system has high reliability, high measure accuracy and high speed, so better realizes the real-time and high precision measure on internal impedance and voltage of every single chip for minitype stack.