Study On Humidity Soft-sensing Of Vehicular PEMFC

Internal humidity is a very important state parameter during vehicular fuel cellsystem operates, which severely affects cell’s operational property and its life.However it is too hard to use available means and methods to directly measure andobtain the humidity parameter. So launching the key technology and methodresearches of humidity soft-sensing in proton exchange membrane fuel cells haveimportant theoretical significance and realistic significance. By taking theself-developed60kW vehicular proton exchange membrane fuel cell system as thestudying objective, the dissertation has done comparatively systematic researches onthe humidity mechanism, the experimental model based on internal resistance,soft-sensing model and its modeling optimization based on Fuzzy Sets and ArtificialNeural Network and humidity soft sensor system as well. The main researchachievements of this dissertation are as follows:The dissertation has proposed a kind of vehicular fuel cell humidity soft-sensingmodel based on internal resistance. By according to the high coherence betweenmoisture content (humidity) and internal resistance of the fuel cell, this method hasestablished soft-sensing based on internal resistance. And by adopting the on-linetesting real time internal resistance when fuel cells operate, the model establishedexperimental equation between internal resistance and humidity. By using theoptimized resistance data, the humidity soft-sensing results conformity with thetheoretical value very well.The dissertation has established a kind of soft-sensing model for the humidity offuel cells based on Fuzzy Sets. That with using three parameters which including withinternal resistance value as the input values of fuzzy logical model and the humidityvalue as model output and adopting Normal Gaussian membership functions, andadopting Standard Additive Model as learning algorithm, the dissertation hasestablished and trained soft-sensing model of fuel cell humidity based on Fuzzy Sets,the outputs and the actual humidity accorded with high degree.The dissertation has established a kind of dynamic fuzzy Neural Networksmodeling for the soft-sensing of vehicular fuel cell humidity. Based on the neural networks and Fuzzy Sets, the dissertation has proposed a dynamic fuzzy neuralnetworks modeling method of the soft-sensing for the vehicular fuel cell humidity.And the dissertation has done the analysis and comparison based on the simulatingresults of the actual testing data, which proves that the dynamic fuzzy neuralnetworks work better.The dissertation has designed and realized a kind of humidity states monitor softsensor system when fuel cells operate. Combining with humidity soft-sensingmethods based on internal resistance and dynamic neural networks, the dissertationhas finished the design of the software and hardware system and the design of thecommunication and operational interface between up and down machine. And thehumidity soft instrument outputs based on the real data conformed well to the actualsituation, which has testified the effectiveness of the soft sensor operation.Above all, this dissertation has proposed relevant research methods from severalperspectives: humidity mechanism analysis, soft-sensing experimental model basedon internal resistance, soft-sensing model based on Fuzzy Sets, soft-sensing modelingbased on neutral networks, the humidity soft-sensing model based on the dynamicfuzzy neutral networks and humidity soft sensor system design. Actual operationaldata and humidity soft-sensing results have testified the feasibility and effectivenessof the design of this dissertation, which provides some methods and references for thehumidity soft-sensing of the vehicular fuel cells and the researches involved in otherrelevant problems.