基于HPSO-BP神经网络融合的锂电池SOC预估研究

为实现锂离子电池荷电状态(SOC)的高精度预测,采用混合粒子群(HPSO)与BP神经网络相结合的联合优化算法,通过优化神经网络的初始权值和阈值克服了种群易陷入局部极小的缺点,加快了收敛速度,减小了SOC预估的误差,通过分析磷酸铁锂(LiFePO4)电池充、放电机理,将电池电压、电流、内阻和温度作为SOC的影响因子。MATLAB仿真结果表明,HPSO-BP神经网络算法的预测精度和收敛速度较传统BP神经网络算法更优。

Estimation for SOC of Li-ion Battery Based on HPSO-BP Neural Network Fusion

In order to make high-precision prediction of Li-ion battery State of Charge (SOC), a joint optimization algorithm combining Hybrid Particle Swarm Optimization (HPSO) and BP neural network is adopted, the combined algorithm, that optimizes the initial weights and biases of neural network, overcomes the drawback of swarm that is easy to fall into local minimum, accelerates the convergence speed, and reduces the error of SOC estimation. Through the analysis of the lithium iron phosphate battery (LiFePO4) charging and discharging mechanism, battery voltage, current, resistance and temperature are used as the impact factors of SOC. MATLAB simulation result indicates that HPSO-BP neural network algorithm performs better than the traditional BP neural network algorithm in prediction accuracy and convergence speed.