PEMFC的PSO优化LS-SVM动态建模仿真

质子交换膜燃料电池(PEMFC)作为一种电化学反应装置,湿度管理对于它的运行性能十分关键.为了建立相应的动态模型,采用了最小二乘支撑向量机(LS-SVM)算法,并使用微粒群算法(PSO)对LS-SVM模型参数进行了自动优化.这种方法建立的动态数学模型能够以进气湿度和工作电压为输入量预测输出的电流密度.仿真结果显示,模型的预测值和实验数据间的平均平方相对误差小于0.0001,能够达到很好的预测精度.因此,这种优化建模策略可以有效的应用到PEMFC研究中.

Dynamic Modeling and Simulation of PEMFC by PSO Optimized LS-SVM

Proton exchange membrane fuel cell (PEMFC) is an electrochemical reactor, and water management is paramount critical for its performance. In order to build a dynamic humidity model for PEMFC, least squares support vector machine (LS-SVM) is adopted, and particle swarm optimization (PSO) is used to select the best hyperparameters of the model. Using inlet humidification and cell voltage as input variables, the proposed dynamic model can predict the output current density very well with the mean square relative prediction error less than 0.0001. Therefore, the proposed modeling strategy can be used to PEMFC research fields.