Prediction of Lithium battery remaining life based on fuzzy least square support vector regression

Batteries are essential components of any aircraft electrical system and exhibit aging and health degradation during operation. Therefore, the correct estimation of the battery remaining useful life (RUL) is important to aircraft operators. The prediction methods of existing Lithium battery remaining life mostly have no learning capabilities and nonlinear prediction ability. In order to predict the remaining life of Lithium battery more accurately, an algorithm based on fuzzy least square support vector regression (FLS-SVR) is presented. This algorithm reconstructs the phase space of multivariate time series using improved embedding dimension time delay automatic algorithm. This algorithm determines the embedding dimension m and the delay timeτ. Then, a FLS-SVR model is built according to m and τ. The parameters of SVR are optimized by adaptive chaotic particle swarm optimization (ACPSO). Comparing with the Logistic regression method, the simulation result demonstrates that the FLS-SVR prediction model has smaller prediction error.