启发式BP神经网络估算铅酸电池的SOC

SOC estimation of lead-acid battery by heuristic BP neural network

针对BP神经网络估算电池荷电状态(SOC)时存在易陷入局部最优、估算精度低等问题,提出基于启发式BP神经网络的电池SOC估算方法。将粒子群优化算法(PSO)和遗传算法(GA)相结合,构成混合启发式算法,并引入混沌机制,优化BP神经网络来对电池SOC进行估算。以阀控式铅酸电池(VRLA)为实验研究对象,与传统BP神经网络估算方法比较,发现基于启发式BP神经网络的SOC估算误差可控制在2%以内。

Aiming at the problems of easily trap into local optimization and low estimation accuracy in estimating the state of charge (SOC) by BP neural network, the estimation method of SOC of battery based on heuristic BP neural network was proposed. Particle swarm optimization (PSO) and genetic algorithm (GA) were combined to form hybrid heuristic algorithm,and chaos mechanism was introduced to optimize BP neural network algorithm to estimate the SOC. The valve regulated lead-acid (VRLA) Battery was used as the experimental research object. Compared with the traditional BP neunil network estimation method, it was found that the estimation error of SOC based on the heuristic BP neural network could be controlled within 2%.