A Mutated Particle Filter Technique for System State Estimation and Battery Life Prediction

Abstract

When classical particle filter (PF) techniques are used for dynamic system state estimation, they have some limitations: for example, when the weights of simulated samples are not sufficiently large, these classical PFs may suffer from sample impoverishment. In addition, the degraded diversity in sampling particles will limit the estimation accuracy, since the particles cannot capture the entire probability density function (pdf) effectively. To tackle these problems, a mutated PF (MPF) technique is proposed in this paper to approximate the posterior pdf of system states. In the MPF, a novel mutation approach is proposed to search extended areas of the prior pdf using mutated particles to make more comprehensive exploration of the posterior pdf. In addition, a particle selection scheme is suggested in the MPF to detect and process low-weight particles so as to explore the high-likelihood area of the posterior pdf more thoroughly. The effectiveness of the developed MPF technique is verified by simulation tests using a benchmark test model. It is implemented for predicting remaining useful life of batteries. Test results show that the developed MPF can capture a system's dynamics effectively and track system characteristics accurately.

Battery state of health estimation: a structured review of models, methods and commercial devices

“…It is measured during the charging phase and the authors show that there exists an almost linear relationship between the capacity and the DC resistance. MPF transformation of the system posterior PDF [70]. The output membership functions are actually four thresholds of the health condition: healthy, acceptable, weak and bad.…”

“…In [70], the authors propose what they call a mutated particle filter (MPF) method for estimating the SoH and RUL. The new method is designed to tackle the shortcomings of the classical, Rao-Blackwellized and UPF methods.…”

An adaptive particle filter based on PSO and fuzzy inference system for nonlinear state systems

“…In general PF, the state transition is often chosen as the proposal distribution [10,11]. As this proposal distribution does not include information of the new observations, the most particles get negligible weights and it leads to particle degeneracy [12][13][14]. By selecting a good the proposal distribution, which contains the current measurement information, the particle impoverishment problem can also be alleviated.…”

An integrated unscented kalman filter and relevance vector regression approach for lithium-ion battery remaining useful life and short-term capacity prediction

“…Depending on the mathematical representations of the batteries degradation evolution, model-based filtering methods use particle filter (PF), extended Kalman filter (EKF) or unscented Kalman filter (UKF) to estimate the state-of-health (SOH), state-of-charge (SOC) or remaining useful life (RUL) of batteries [3][4][5][6][7]. Hu et al [8] applied a particle filtering and kernel smoothing approach (PF-KS) for simultaneously estimating the degradation state and the unknown parameters of degrading components, then RUL prediction is obtained by simulating future particles evolutions.…”