基于ELM–Takagi Sugeno的车载锂电池优化充电策略的研究

Research on optimized charging strategy of vehicle lithium battery based on ELM–Takagi Sugeno

针对目前车载锂电池充电慢、充电效率低以及对电池损害大等问题,提出一种基于ELM–Takagi Sugeno(T–S)模型的锂电池梯级式优化充电策略.首先通过极限学习机(ELM)获得锂电池的最佳充电电流与温度、内阻、极化电压等特征参数之间的数学模型,将离线训练好的参数存储在STM32处理器中,系统每采集1次数据,处理器即调用ELM程序计算当前时刻的最佳充电电流.其次利用T–S模糊模型将恒压、恒流、脉冲3种不同充电方式优化处理实现动态最优充电.仿真结果表明锂电池实际充电电流能够实时跟踪最优充电电流,其充电时间比三阶段充电模式缩短20%,充电效率比CC–CV充电模式提高约25%.

Aiming at the problems of slow charging, low charging efficiency and great damage to the battery,a lithium battery step –optimized charging strategy based on ELM –Takagi Sugeno(T–S) model is proposed.Firstly, lithium is obtained by the extreme learning machine(ELM). The mathematical model between the optimal charging current of the battery and the characteristic parameters such as temperature, internal resistance and polarization voltage, and the offline training parameters are stored in the STM32 processor. Each time the system collects data, the processor calls the ELM program to calculate the optimal charging current at the current moment.Secondly, the T–S fuzzy model is used to optimize the three different charging modes of constant voltage, constant current and pulse to achieve dynamic optimal charging. The simulation results show that the actual charging current of the lithium battery can track the optimal charging current in real time. The charging time is 20% shorter than the three-stage charging mode, and the charging efficiency is increased by about 25% compared with the CC–CV charging mode.