基于布谷鸟算法优化BP神经网络的锂电池健康状态预测

Estimation of SOH for battery based on CS-BP neural network

摘要:

锂电池健康状态(SOH)的预测是电动汽车锂电池管理系统的最重要的关键技术之一。传统的误差逆向传播(BP)神经网络容易使权值和阈值陷入局部最优，从而导致预测结果不精确。结合布谷鸟搜索算法(CS)的全局优化能力，提出一种基于CS算法优化BP神经网络的锂电池SOH预测方法，该方法的核心在于优化BP神经网络的初始权值和阈值，从而减小算法对初始值的依赖。为了验证算法的泛化性，利用美国国家航空航天局开源锂电池数据集6号电池和7号电池进行仿真实验，仿真得到该算法预测SOH的均方根误差(RMSE)分别为0.2658和0.2620，平均绝对百分比误差(MAPE)分别为0.3319%和0.2605%。通过与BP神经网络、粒子群优化的BP神经网络(PSO-BP)、遗传算法优化的BP神经网络(GA-BP)对比，布谷鸟算法优化的BP神经网络(CS-BP)具有更小的预测误差。

Abstract:

Estimating the state of health(SOH) of lithium battery is one of the most important key techniques of lithium electric vehicle battery management system. The traditional error back propagation(BP) neural network is easy to bring the weight fall into local optimal solutions, which can lead to inaccurate prediction results. Combined with the cuckoo search algorithm(CS) which has global optimization ability. A method based on cuckoo search algorithm optimized BP neural network model for predicting the SOH of lithium ion battery is proposed, the core of the method is optimizing the BP neural network's initial weights and thresholds. This method can reduce the dependence of the algorithm on the initial value. At the same time, in order to verify the generalization performance of the algorithm, use the NASA open source lithium battery data set No. 6 battery and No. 7 battery for simulation experiments, and the CS-BP algorithm is simulated to predict the root mean square error (RMSE) of SOH. They are 0.2658 and 0.2620, and the mean absolute percentage error (MAPE) is 0.3319% and 0.2605%, respectively. Compared with BP algorithm, particle swarm optimization BP neural network (PSO-BP), genetic algorithm optimized BP neural network (GA-BP), cuckoo search algorithm optimized BP neural network(CS-BP) has smaller prediction error.