Dataset Information

A generic battery-cycling optimization framework with learned sampling and early stopping strategies

ABSTRACT: Summary Battery optimization is challenging due to the huge cost and time required to evaluate different configurations in experiments or simulations. Optimizing the cycling performance is especially costly since battery cycling is extremely time consuming. Here, we introduce an optimization framework building on recent advances in machine learning, which optimizes battery parameters efficiently to significantly reduce the total cycling time. It consists of a pruner and a sampler. The pruner, using the Asynchronous Successive Halving Algorithm and Hyperband, stops unpromising cycling batteries to save the budget for further exploration. The sampler, using Tree of Parzen Estimators, predicts the next promising configurations based on query history. The framework can deal with categorical, discrete, and continuous parameters and can run in an asynchronously parallel way to allow multiple simultaneous cycling cells. We demonstrated the performance by a parameter-fitting problem for calendar aging. Our framework can foster both simulations and experiments in the battery field. Graphical abstract Highlights • A battery-parameter optimization framework is proposed with a pruner and a sampler• The framework can optimize categorical, discrete, and continuous variables• An early-stopping strategy is introduced to reduce the high cycling cost• Parameter fitting problems are investigated to demonstrate fast optimization The bigger picture There are many parameters to optimize for a battery, in both simulations and experiments, from design to manufacturing. It is time consuming and costly to evaluate the lifetime performance of batteries since it takes a long period to cycle them. We introduce a generic framework leveraging machine-learning algorithms. The framework is designed to optimize battery parameters to enhance cycling performance in a systematic and efficient way, which allows parallel cyclers, stops unpromising cycles, and automatically yields new configurations of parameters. The framework could reduce the average cycling time per battery from years to months/weeks for cycling experiments or from weeks to days/hours for cycling computations. This method is flexible to scale up for many applications, from fundamental research to industrial development in batteries and other similar fields. Evaluating battery performance typically requires cycling the cells hundreds or thousands of times. Thus, it is time costly to evaluate the effect of battery parameters on its performance and even more expensive to optimize the parameters. We introduced a generic framework leveraging machine learning incorporated with a pruner and a sampler to efficiently optimize the battery parameters. It allows parallel cyclers, stops unpromising cycles, and automatically yields new configurations of parameters. The framework showed excellent results in our demonstration.

SUBMITTER: Deng C

PROVIDER: S-EPMC9278511 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BACKGROUND:Delivery of behavioral interventions is complex, as the majority of interventions consist of multiple components used either simultaneously, sequentially, or both. The importance of clearly delineating delivery strategies within these complex interventions-and furthermore understanding the impact of each strategy on effectiveness-has recently emerged as an important facet of intervention research. Yet, few methodologies exist to prospectively test the effectiveness of delivery strategies and how they impact implementation. In the current paper, we describe a study protocol for a large randomized controlled trial in which we will use the Multiphase Optimization Strategy (MOST), a novel framework developed to optimize interventions, i.e., to test the effectiveness of intervention delivery strategies using a factorial design. We apply this framework to delivery of Family Navigation (FN), an evidence-based care management strategy designed to reduce disparities and improve access to behavioral health services, and test four components related to its implementation. METHODS/DESIGN:The MOST framework contains three distinct phases: Preparation, Optimization, and Evaluation. The Preparation phase for this study occurred previously. The current study consists of the Optimization and Evaluation phases. Children aged 3-to-12 years old who are detected as "at-risk" for behavioral health disorders (n?=?304) at a large, urban federally qualified community health center will be referred to a Family Partner-a bicultural, bilingual member of the community with training in behavioral health and systems navigation-who will perform FN. Families will then be randomized to one of 16 possible combinations of FN delivery strategies (2 × 2 × 2× 2 factorial design). The primary outcome measure will be achieving a family-centered goal related to behavioral health services within 90?days of randomization. Implementation data on the fidelity, acceptability, feasibility, and cost of each strategy will also be collected. Results from the primary and secondary outcomes will be reviewed by our team of stakeholders to optimize FN delivery for implementation and dissemination based on effectiveness, efficiency, and cost. DISCUSSION:In this protocol paper, we describe how the MOST framework can be used to improve intervention delivery. These methods will be useful for future studies testing intervention delivery strategies and their impact on implementation. TRIAL REGISTRATION:ClinicalTrials.gov, NCT03569449. Registered on 26 June 2018.

Dataset Information

A generic battery-cycling optimization framework with learned sampling and early stopping strategies

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets