A Robust Hybrid Filtering Method for Accurate Battery Remaining Useful Life Prediction

Abstract

Accurate remaining useful life (RUL) prediction under the noisy environment is a big challenge for the health management of modern industrial systems since the extraction of the accurate data structure from heavily corrupted data is difficult. In recent years, the kernel adaptive filter (KAF) has been widely adopted to solve the robust regression problem due to its low-complexity and high-approximation capability and robustness while the applications in battery RUL prediction are still few and far between. Thus, this paper is concerned with long-term RUL prediction using the KAF method. At first, concretely speaking, a robust KAF algorithm is derived based on the double-Gaussian-mixture (DGM) cost function, which is used to learn the capacity degradation mechanism from contaminated capacity data and so as to build the long-term prediction model. Second, a robust unscented Kalman filter (UKF) algorithm employing the DGM-based cost function is developed, which is then combined with the KAF-based prediction model to realize a more accurate and reliable prediction. Under the hybrid prognostic framework, the proposed UKF algorithm is applied to filter the noisy observations. When the observation data are inaccessible, the predicted data from the off-line trained KAF-based prediction model are adopted as the approximated value of the real observations for the UKF algorithm to optimize the prediction results and to provide the uncertainty representation. The experimental results reveal that the proposed method has great robustness when the measurements contain noise and large outliers, which makes it possible to get satisfactory prediction performance without preprocessing the data manually.INDEX TERMS Kernel adaptive filter, battery, unscented kalman filter, remaining useful life (RUL), robustness.

Improved Modeling of Lithium-Ion Battery Capacity Degradation Using an Individual-State Training Method and Recurrent Softplus Neural Network

“…Past capacity was also excluded because the past conditions were already learned by training the initial RSNN states. We also argue that the exclusion of past capacity is in fact favorable for long-term capacity modeling: using past capacity as an input tended to result in short-term models that rely on capacity data from very few cycles ago, which tended to produce poor long-term prediction due to the accumulation of model errors [25], [31], [38]. In contrast, without short-term capacity information to rely on, our method optimized the long-term prediction of the RSNN instead.…”

“…Since LIB degradation is a dynamic process with pathdependent behavior [37], a recurrent data-driven model is crucial to the precise modeling of the degradation process. For example, studies such as [23] and [38] implement recurrence by using the capacity of consecutive time steps as the model input and output, but other studies find that this solution tends to produce poor long-term prediction due to the accumulation of model errors [25], [31], [38]. Recent studies instead implement recurrence by using recurrent models such as long short-term memory neural networks (LSTMs) [1], which are recurrent neural networks (RNNs) that are designed to learn long-term dependence in time series [1].…”

Simulator Assessment Theory for Remaining Useful Life Prediction of Lithium-Ion Battery Under Multiple Uncertainties

“…According to the predictive lines, we can calculate the RUL by (24) and (26). The results are tabulated in Table 1, it can be seen that the method proposed in the paper is more accurate than BMC and GPR.…”