Inverse Dynamics Modeling And Control For SOFC System Based On Support Vector Machines

Solid oxide fuel cell (SOFC) is a kind of high efficient and clean new fuel cells.SOFC has great application on co-supply,distributed generation,transportation,movablepower source and so on.There is a great significance on alleviating the energycrisis,protecting the living environment of human,satisfing requirements of the humansociety and ensuring national safety.At the present time,researchs which have been carried on SOFC are mainly onchemistry,material and structure,there is little on the control of SOFC,but it is anecessary procedure before the practicality and industrialization of SOFC.According tothe problems,the main achievements of this thesis are given below:1．In allusion to SOFC stack objects,based on the electrochemical theory, energyconservation theory and Nernst equation.the model of SOFC stack isestablished.then,qualitative analysis is made for the factors which affect the outputcharacteristics of the SOFC,such as temperature, pressure, current, fuel efficiency, fuelcomponents.According to the correlative analysis of the object dynamic process,thedynamic response characteristics of temperature and the voltage is obtained for SOFCsystem.2．LSSVM is difficult to realize on-line identification for complexity systembecause it need solve inverse matrix when updating model in dynamic modelingprocess.While considering the complexity of the SOFC system,This paper adopt arecursive least square support vector machine (RLSSVM) method to establish SOFCsystem inverse dynamics model.The parameter of model are updated by recursive leastsquare(RLS) algorithm.To speed up the computational speed,the number of supportvectors is reduced by doing pruning.it has realized on-line identification for inversedynamic model of SOFC system.3．Since the system of SOFC is highly non-linear, conventional linear controltheories are not suitable for designing its controllers.This paper employ an adaptiveinverse control system based on inverse dynamic model of RLSSVM for SOFCsystem.In this adaptive inverse control mechanism,inverse dynamic model were updatedby RLS algorithm.The parameters of controller were adjusted on-line by ε-filtering.Thecontrol of temperature and voltage of SOFC is realized through on-line identificationfor inverse dynamic model. The simulation experiment results show that the compression operation or increasethe temperature operation can effectively improve the SOFC output characteristics.Thehigher the proportion of hydrogen, the higher the partial pressure of anode hydrogen, itcan effectively increase the ideal potential.Usually keeping fuel efficiency a constantvalue.Hydrogen gas flow and electric current can lead to a mutation process for voltage,then gradually converge to the stable value, temperature response is a gradualconvergence process;the inverse dynamic model identified has high precision andtractability,can effectively avoid the internal complexity of SOFC.The controllerdesigned has good control performance and Can realize the comprehensive control ofthe temperature and the voltage of SOFC system.