基于人工智能的电网用蓄电池健康度评估

Evaluation on State of Health of Storage Battery for Power Grid Based on Artificial Intelligence

目前对蓄电池检测的研究大部分集中在剩余电量方面,基于人工智能技术的蓄电池健康度(state of health,SOH)评估研究相对较少。为此,对电网用蓄电池失效机理进行分析,列举出影响蓄电池SOH的主要参数,提出以数据驱动的改进列文伯格-马夸尔特最优化方法的反向传播(Levenberg-Marquardt back-propagation,LMBP)神经网络算法来评估蓄电池的健康程度。测试数据表明,改进的LMBP算法评估结果相对误差小于2%,说明基于数据驱动的人工智能技术可作为行之有效的蓄电池SOH评估方法。

Most of study on storage battery detection focuses on remaining capacity at present while there is relatively less research on evaluation on state of health (SOH) of the storage battery based on artificial intelligence technology.This paper analyzes failure mechanism of the storage battery for the power grid,lists main parameters affecting SOH of the battery and proposes the Levenberg-Marquardt back-propagation (LMBP) neural network algorithm driven by data to evaluate SOH of the storage battery.Test data indicates that relative error of evaluation result of the improved LMBP algorithm is less than 2%,which declares that the artificial intelligence technology based on data driven is a feasible and effective evaluation method for SOH of the storage battery.