Battery lifetime prediction is crucial in industrial applications. However, the lack of diversity in training data often poses challenges regarding the robustness and generalization of lifetime predictions for batteries from different batches. Motivated by the early cycle data from lithium-ion batteries, this article proposes a robust transfer learning method by employing a model average framework, where the weights are determined based on the distance between the source domain and the target domain. Kernel regression is used to build the prediction of battery lifetime using early cycle data, and transfer component analysis is utilized to transfer knowledge between different domains. The case study on lithium-ion phosphate/graphite cells demonstrates that the proposed method can mitigate the impact of negative transfer and has superior performance compared to traditional methods.
In portfolio optimization, the computational complexity of implementing almost stochastic dominance has limited its practical applications. In this study, we introduce an optimization framework aimed at identifying the optimal portfolio that outperforms a specified benchmark under almost second-degree stochastic dominance (ASSD). Our approach involves discretizing the return range and establishing both sufficient and necessary conditions for ASSD. We then propose a three-step iterative procedure: first, identifying a candidate portfolio; second, assessing its optimality; and third, refining the discretization scheme. Theoretical analysis guarantees that the portfolio identified through this iterative process improves with each iteration, ultimately converging to the optimal solution. Our empirical study, utilizing industry portfolios, demonstrates the efficacy of our approach by consistently identifying an optimal portfolio within a few iterations. Furthermore, comparative analysis against other decision criteria, such as mean-variance, second-degree stochastic dominance, and third-degree stochastic dominance, reveals that ASSD generally leads to portfolios with higher out-of-sample average excess returns but also entails increased variations and risks. This paper was accepted by Agostino Capponi, finance. Funding: C. Luo acknowledges financial support from the National Natural Science Foundation of China [Grant 72101070] and the Zhejiang Provincial Natural Science Foundation of China [Grant LY23G010001]. P. Chen acknowledges financial support from the National Natural Science Foundation of China [Grant 72401253]. P. Jaillet acknowledges financial support from the Office of Naval Research [Grant N00014-18-1-2122 and N00014-24-1-2470] and the Air Force Office of Scientific Research [Grant FA9550-23-1-0182 and Grant FA9550-23-1-0190]. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2022.01092 .
1 March 2018 in European Journal of Operational Research
Influenced by dynamic environmental factors, the degradation rates of products often vary with initial levels. However, most existing degradation models assume a fixed or zero starting point and treat it as independent of subsequent process, leading to biased failure-time predictions, poorly timed maintenance, and elevated costs or risks. To fill this gap, we propose a Wiener-process degradation model that explicitly links the random initial phase to its ongoing deterioration, allowing both the degradation rate and the failure threshold to adapt to the initial state. We derive explicit expressions for the lifetime and remaining useful life (RUL) distributions, along with confidence intervals for model parameters. To construct confidence intervals for reliability metrics and prediction intervals for RUL, we propose tailored generalized inference and objective Bayesian inference, which ensure accurate coverage even in small samples. Through simulations and a real-world case study, we demonstrate that our approach more accurately captures individual variability and delivers tighter, more reliable life-time forecasts than traditional models.
14 November 2019 in Naval Research Logistics (NRL)