遗传算法和BP网络在电池电量预测中的研究

Research on Capacity Predication of Battery based on BP network and Genetic Algorithm

蓄电池剩余容量为汽车可持续行进提供有力的判据,所以,对它的准确估计有重要的意义。该文在BP网络的基础上采用一种组合方法对荷电状态进行预测;并利用BP网络学习能力与泛化能力满足的不确定关系确定隐层节点数;利用遗传算法,确定初始权值和阀值,使网络的初始条件得到优化,使神经具有更好的收敛速度和收敛质量;通过实验表明网络不仅收敛速度快,而且易达到最优解,证明网络对MH-N i电池剩余电量的预测是有效的。

The surplus capacity of battery is a reasonable criterion for the continuable travel of the vehicle. Thus , it is significant to predict it exactly. In this paper, a combined solution for prediction based on BP network is used. By giving the undefined relation between learning ability and generalization ability of BP neural network, the hidden notes are obtained. The original weights and bias are defined by using the genetic algorithm. It can improve the search efficiency and global optimization. The simulation result shows this method has high convergent speed, can obtain global optimization easily. It proves the network is successful.