基于CEEMD-AKF的锂电池剩余使用寿命预测方法

Remaining Useful Life Predictionmethod for Lithium-ion Batteries Based on CEEMD-AKF

针对锂离子电池剩余使用寿命(RUL)预测存在建模复杂、预测误差大等问题,提出一种基于CEEMD-AKF的锂电池RUL预测方法。首先,基于补充的总体平均经验模态分解(CEEMD)将电池历史容量分解为固有模态函数(IMFs)及余项,并基于排列熵(PE)与均方根误差(RMSE)建立最优低通滤波器,以此消除原始容量的随机性波动与噪声。其次,自适应卡尔曼滤波(AKF)用于更新自回归(AR)模型参数。最后,基于蒙特卡洛(MC)模拟得到概率密度函数(PDF),用于评估RUL预测结果的不确定性。通过在NASA测试数据上进行试验分析,结果表明CEEMD-AKF方法既能够降低建模复杂性,又能够有效地提高RUL预测精度。

Aiming at the problems of complex modeling and large errors in the prediction of remaining useful life(RUL)of lithium-ion batteries,a novel RUL prediction method was proposed.Firstly,the battery historical capacity was decomposed into a set of intrinsic mode functions(IMFs)and one residue based on the complementary ensemble empirical mode decomposition(CEEMD).Based on the permutation entropy(PE)and root mean square error(RMSE),an optical low-pass filter was established to eliminate the random fluctuation and noise of the raw capacity.Secondly,the adaptive Kalman filter(AKF)was used to update the parameters of the Autoregressive(AR)model.Finally,a probability density function(PDF)was obtained based on Monte Carlo(MC)simulation,which was used to evaluate the uncertainty of RUL prediction.The experimental analysis on the NASA data set shows that the CEEMD-AKF method can not only reduce the modeling complexity,but also can effectively improve RUL prediction accuracy.