ABSTRACT
This chapter aims to compare several prognostics methods for lithium-ion batteries. It carries out more experiments on capacity prognostics of lithium-ion batteries by applying Gaussian Process Regression (GPR) with varying basis functions. Due to the characteristics of high energy density, flexible and lightweight design, long lifespan, and low cost, lithium-ion batteries are becoming an increasingly key device for storing and transferring energy, ranging from mobiles and electric vehicles to space crafts. Extensive works have been conducted to perform battery prognostics. Approaches for battery prognostics can be categorized into three classes: physics-based, data-driven, and hybrid approaches. GPR is based on, but more generic than Bayesian linear regression. High-performance prognostics of battery health is of interest to both users and producers. Support vector machine has been utilized for prognostics of battery health. Support vector machine solves nonlinear tasks by transforming data into a higher feature space, where a problem becomes linear.
