应用BP神经网络建立了电池模型,并应用自适应遗传算法对其权值阈值进行了优化,最后利用MATLAB编写了其仿真程序进行多组数据的测试,结果与纯BP网络和GA-BP网络进行了对比.结果表明,AGABP网络具有训练时间短、精度高的特点,对电池任一状态下的剩余容量预测均有效.%A model is created by adopting BP neural network to predict the state of charge of battery, and adaptive genetic algorithm (AGA) is utilized to optimize its weights and bias, to analyze many factors that affecting the battery residual capacity. Finally, with the emulation program written by MATLAB, multiple sets of data are tested and compared with pure BP network and GA-BP network. The results show that the AGA-BP network has a short training time and high accuracy. It is effective for prediction of the any state battery capacity.
展开▼
