【摘要】 针对传统埃尔曼(Elman)神经网络在预测过程中初始权值和阈值随机性选取,易陷入局部极小化问题,为提高锂电池剩余寿命的预测精度,提出一种基于自适应权重的改进粒子群(IPSO)-埃尔曼(Elman)神经网络预测锂电池剩余寿命的方法。针对锂电池测量数据中伴随的噪声,利用高斯去噪,削弱数据中的噪声影响,提取原始数据;再利用IPSO全局搜索的能力对Elman神经网络的初始参数进行优化;最后基于美国国家航空航天局(NASA)提供的锂电池测量数据,对提出的方法进行有效性验证,并与常规的BP,Elman算法进行对比。预测结果表明,IPSO-Elman预测误差在不同训练样本下都小于BP,Elman算法,表现出较强的适应能力。更多还原

【Abstract】 The random selection of initial weights and thresholds in the prediction process of traditional Elman neural network is easy to fall into the problem of local minimization. A method of lithium battery remaining life prediction based on IPSO(improved particle swarm optimization)-Elman neural network is proposed to increase the prediction accuracy of the remaining life of the lithium battery. In allusion to the concomitant noise in the lithium battery measurement data,Gaussian is used to reduce the noise and the impact of noise in data,and extract the original data;the initial parameter of the Elman neural network is optimized by using the IPSO global search capability;and then the effectiveness of the proposed method is verified based on the lithium battery measurement data provided by NASA(national aeronautics and space administration),and compared with conventional BP and Elman algorithms. The prediction results show the prediction error of IPSO-Elman is smaller than that of BP and Elman algorithms under different training samples,which shows a stronger adaptability.更多还原