Research On MPC Of Fuel Cell Emergency Power Supply Based On Load Power Prediction

In order to speed up the process of energy conservation and emission reduction and achieve the strategic goals of "carbon peaking" and "carbon neutrality",the development of a new energy emergency power supply system represented by fuel cells has become an urgent task for the development of the current emergency power supply industry.In order to reduce the hydrogen consumption of the fuel cell hybrid emergency power system,taking the emergency power system composed of two fuel cell modules and a lithium battery as the object,this thesis proposes a load power prediction model based on BP neural network,which is applied to the energy management strategy based on model predictive control.Then,a model predictive control energy management strategy based on load power prediction is proposed.The main research contents of this paper are as follows:Firstly,the energy source of the fuel cell emergency power supply system is analyzed,the system structure of "two fuel cells + lithium battery" in parallel is determined,and the simulation model of each component of the system is built,mainly including fuel cell,lithium battery and DC/DC convertor.The accuracy of model is verified according to the actual test data.Secondly,the load power of the fuel cell emergency power supply system is predicted.Three short-term load forecasting models were established,namely,the metabolic grey model(MGM),the metabolic grey markov model(MGMM)and the BP neural network forecasting model.Compare the prediction results of the three prediction models on the same dataset.The results show that,compared with the MGM and the MGMM,the BP neural network prediction model has higher prediction accuracy for the prediction of load power.Then,a model predictive control(PMPC)energy management strategy based on load power prediction is designed.The objective function of equivalent consumption minimization(ECMS)control strategy is extended from instantaneous to local as the objective function of model predictive control(MPC),thus realizing the transition from instantaneous optimization to local optimization.The load power in the control horizon predicted by the BP neural network is combined with the MPC,and the PMPC energy management strategy is proposed.The PMPC control strategy is simulated and compared with the MPC,ECMS,and logic threshold control strategies.The results show that when the load power fluctuates greatly,PMPC can reduce hydrogen consumption by 5.45%,7.63% and 10.33%,respectively,compared with MPC,ECMS,and logic threshold control strategies.Finally,for the problem of solving the objective function of model predictive control,a nonlinear programming algorithm based on sequential quadratic programming(SQP)is proposed in this paper,which can achieve better solution speed and solution accuracy.In order to verify the feasibility and effectiveness of the energy management strategy of the emergency power system,a rapid-control-prototype simulation platform for fuel cell hybrid emergency power supply based on d SPACE is built in this paper.The experimental results show that PMPC can reduce hydrogen consumption by 5.02%,7.48% and 10.07%,respectively,compared with MPC,ECMS and logic threshold control strategies,which improves the working efficiency of fuel cells and reduces the fluctuation of output power of lithium batteries.