Fast Bayesian Inference of Reparameterized Gamma Process With Random Effects

In recent years, deep learning (DL) methods for state of health (SOH) estimation of lithium‐ion (Li‐ion) batteries have attracted great attention. However, most existing DL‐based methods for SOH estimation were designed using specific battery dataset. To gain better performance, the procedure can be labor‐intensive, involving designing intricate features and fine‐tuning complex DL model, which significantly limits the applications of these methods. Moreover, uncertainty quantification, that is how much uncertainty these DL‐based methods can have on their SOH estimation, cannot be addressed with only point estimation. To this end, this paper proposes a Bayesian DL pipeline for Li‐ion battery SOH estimation targeting at automatic feature extraction and uncertainty‐aware DL‐model developing for general application to batteries under different working conditions. The proposed pipeline is composed of three modules accounting for common feature extraction, automatic feature selection, and Bayesian neural network (BNN) model construction. Specifically, a frequently‐used feature set is designed using the voltage, current and time data of constant current and constant voltage (CC‐CV) charging mode, while the recursive feature elimination (RFE) with cross validation (RFECV) algorithm is used to automatically select features based on relevance. Based on the automatically selected features, DL model is automatically constructed without too much human intervention. The uncertainty of SOH estimation is quantified by BNN model using Monte Carlo (MC) dropout method, which is subsequently used to calculate the confidence of the estimation results. Four battery datasets, covering regular charging, fast charging, and second‐use applications, are used to demonstrate the performance of the proposed method. The proposed pipeline shows superior performance to state‐of‐art DL‐based methods in terms of application capability to different working conditions and uncertainty quantification for SOH estimation.

Section: Automatic Feature Selectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Bayesian deep learning pipeline for lithium‐ion battery SOH estimation with uncertainty quantification

Ke,

Long,

Yang

et al. 2023

Section: Introductionmentioning

confidence: 99%

“…(2) reliability assessment methods based on probability statistics, such as the LaPlace method, 35 universal generating function (UGF) method, 17 Bayes method, 36 and so on, 30,37 where the UGF method has great advantages in deriving the reliability of performance-sharing systems. 31 However, there is currently no research on the reliability assessment method of balanced systems with epistemic uncertainty, two-level architecture, and distance-dependent transmission loss.…”

Section: Introductionmentioning

confidence: 99%

Reliability assessment of two‐level balanced systems with common bus performance sharing subjected to epistemic uncertainty

Tian

Yang

Wang

2023

Two‐level balanced systems with common bus performance sharing (TLBS‐CBPS), represented by power battery systems, play an important role in various industries. Where the two‐level balance refers to when the system is unbalanced, redistributing performance between modules using a common bus for the intermodule balance, followed by the in‐module component performance redistribution to achieve global balance, that is, the performance differences among all components are within the balance degree threshold. However, the component performance has epistemic uncertainty due to measurement limitations, and the distance‐dependent transmission loss incurs during the performance redistribution. These problems make the reliability evaluation of TLBS‐CBPS challenging. To address these issues, a reliability evaluation method for TLBS‐CBPS is proposed by considering two‐level balance, epistemic uncertainty, and transmission loss. First, the system model is constructed from working modes, and the two‐level rebalancing process is described as three nonlinear programming problems. Then, based on the Dempster–Shafer evidence theory, the belief universal generating function (BUGF) method is proposed to evaluate the uncertainty of the system reliability. To improve the computation efficiency, a simplified BUGF algorithm is given through merging, judging, labeling, and expanding realizations. Finally, two examples of lithium‐ion battery packs demonstrate the effectiveness of the proposed method.

“…In the response stage, rail operators usually take actions such as rolling stock rescheduling, commuter contingency routing and bus bridging services to mitigate the adverse impact of disruptions. 8,9 After that, in the recovery stage, SP will gradually return to a normally functioned level given the implementation of effective response strategies against disruptions. In the literature, the resilience indicator (RI) is typically defined as the ratio of the SP during the survivability, response, and recovery period to the SP without disruption over the same period.…”

Section: Introductionmentioning

confidence: 99%

Enhancing resilience assessment and bus bridging service design for large‐scale rail transit systems under disruptions

Gong,

Song,

et al. 2023

This paper proposes a more computationally efficient approach for resilience assessment of rail transit system under disruptions. An improved linear programming model is developed to depict commuter flows and estimate system statuses. To address the computational challenge caused by the complexity of system, a four‐step approach is proposed based on the proposed commuter flow model. In the first step, Origin‐Destination (OD) pairs are divided into smaller groups and their flows under normal conditions are estimated by the proposed model separately, with the assumption that the railway capacity is sufficient relative to demand. Next, overall system statuses under normal conditions, including commuters on each train and spare capacities of each train are calculated by integrating results obtained in the first step. In the third step, system statuses under disruptions are estimated. In this step, we assume that unaffected commuters will not change their routes and flows of all affected commuters are estimated by a modified commuter model with given spare space of trains. Based on these outputs, several critical measures are introduced and calculated to quantify the resilience, resistance, and recovery ability of rail network systematically. We also demonstrate how our approach could be used to facilitate design and evaluation of bus bridging service. The proposed approach is demonstrated on the core part of Hangzhou rail transit network.