基于NSVB-MPC的PEMFC温度控制与输出特性优化OA

In order to the issues that the stack temperature of proton exchange membrane fuel cell(PEMFC)can affect the reaction rate during operation,and load variations tend to cause unstable output voltage,a nonlinear sparse variational Bayesian model predictive control(NSVB-MPC)strategy is proposed.Based on a multi-physical field coupling model integrating electrochemical,thermal,and dynamic equivalent circuit models,the optimized NSVB-MPC strategy is used to realize real-time system compensation.This strategy can integrate sparse Bayesian learning and variational inference to provide a probabilistic prediction model for model predictive control,thereby quantifying model uncertainty,realizing feature sparsification,reducing redundant features,and enhancing robustness and computational efficiency under small-sample conditions.The simulation experiments verify the effectiveness of the proposed control strategy.The experimental results show that the proposed control strategy can effectively reduce the voltage oscillation amplitude of the fuel cell system and stabilize the voltage of PEMFC.Under the same disturbance conditions,the temperature regulation has the minimum overshoot and the shortest settling time,which indicates that the system possesses better robustness and anti-interference capability.