A Comprehensive Dataset and Benchmark for Battery Life Prediction

View the PDF file from the Batterylife: a comprehensive data collection and a standard for predicting battery life, by Ruifeng Tan and 8 other authors

PDF HTML (experimental) view

a summary:Battery prediction (BLP), which depends on the data chains’ data produced by battery deterioration tests, is very important to use the battery, improvement and production. Despite impressive developments, this research field faces three main challenges. First, the limited size of current data groups impedes visions of modern battery life data. Second, most data groups are limited to small ion batteries that were tested under a narrow range of diversity in laboratories, raising concerns about the generalization of results. Third, non -consistent and limited standards via studies block the effectiveness of basic lines and leave them unclear whether common models in other time chains fields are effective in BLP. To face these challenges, we suggest Batterylife, a comprehensive and standard data set for BLP. Batterylife 16 merges a data set, offers a sample size 2.5 times compared to the largest previous data set, and provides the most diverse battery life supplier with 8 format batteries, 59 chemical systems, 9 degrees of operation, and 421 charging/discharge protocols, including laboratory and industrial tests. It is worth noting that Batterylife is the first to launch battery life data sets from zinc ion batteries, sodium Ion batteries, and Lithium Eyon batteries with a capacity that was tested in the industry. With the comprehensive data collection, we reconsider the effectiveness of popular foundation lines in these fields and other time series. Moreover, we suggest CyclePatch, which is the ingredient technique that can be used in different nerve networks. The wide analogy of 18 ways reveals that common models in other time chains fields can be inappropriate for BLP, and CyclePatch continuously improves the performance of the model that creates modern standards. Moreover, Batterylife evaluates the performance of the model through the conditions of aging and fields. Batterylife is available in this URL https.