Energy Management Of Solid Oxide Fuel Cell/lithium Battery Hybrid Power System

Solid oxide fuel cell(SOFC)is a new type of power generation device that works at high temperature.It has the advantages of cleaness and high efficiency and has great development prospects.The SOFC system cannot track the rapid changes of the load in real time due to its soft output characteristics,so an auxiliary power supply(such as a lithium battery)needs to be introduced to help realize the fast tracking of the load.However,in the hybrid power generation system,the SOFC system and the lithium battery influence each other.Therfore how to realize the real-time energy management strategy in order to enhance the system performance and economy has become the key problem in the SOFC hybrid power system.This paper proposes a real-time energy management strategy based on power prediction for the SOFC/lithium battery hybrid power generation system.The main research contents are as follows:(1)A SOFC/lithium battery hybrid power generation system model is built.Firstly,based on the law of conservation of mass,law of conservation of energy,the principle of electrochemistry,and the law of thermodynamics,a mathematical model of the SOFC system is established,including core stacks,air compressors,heat exchangers,bypass valves,mixers,exhaust combustion chambers,and splitters.Secondly,the lithium battery second-order resistance-capacitance equivalent circuit model is established,and the model parameters are identified based on the least square method.This paper simulates and verifies the polarization curve of the SOFC system and the V-t curve of the lithium battery model.The results show that both the SOFC system and the lithium battery system can accurately reflect the actual output characteristics of the system.(2)A real-time energy management strategy for SOFC/lithium battery hybrid power generation system is proposed,which includes three parts: power prediction,energy distribution,and power tracking.In the part of power prediction,this paper established two power prediction model based on a nonlinear autoregressive neural network and a long-short-term memory neural network based on a household electricity data,and the impact of different parameter settings on the model's prediction accuracy has benn analyzed.Finally,the training speed and prediction accuracy of the two power prediction models are compared.The results show that the long and short-term memory neural network is better than the traditional nonlinear autoregressive neural network in the training speed and prediction accuracy of the model.(3)In the energy distribution part,based on the predicted power,SOFC system and lithium battery working status,rules-based and optimization-based energy distribution strategies are designed respectively to obtain the reference output power of SOFC system and lithium battery.In the power tracking part,the PID controller tuned based on the radial basis function neural network realizes the fast tracking of the SOFC system and the reference output power of the lithium battery by adjusting the fuel flow and charge and discharge current respectively.Finally,the above two energy management strategies are implemented on the MATLAB/Simulink platform.The results show that the optimized energy management strategy is superior to the rule-based energy in the three aspects of SOFC system power generation efficiency,fuel utilization,and hybrid power generation system operating costs.