中文说明：使用遗传算法对电池soc估计.ＲBF 神经网络法有很强的非线性拟合能力，可映射任意复杂的非线性关系，具有很强的鲁棒性和记忆能力，且学习规则简单、学习能力强大，便于计算机实现。但如何合理确定网络的结构和参数，目前尚未有系统的规律可循，网络的逼近性能因此受到影响。GA 借鉴了自然界遗传中适者生存法则，在问题空间进行全局并行、随机的搜索优化，使得种群全局最优的收敛。与传统算法相比，GA 训练神经网络无需先验知识，而且对初始参数不敏感，不会陷入局部最小点。所以具有全局搜索能力的 GA 可以对ＲBF 神经网络进行优化，寻找到最优的网络结构和参数，保证最佳的网络性能。
English Description：
RBF neural network method has strong nonlinear fitting ability, can map any complex nonlinear relationship, has strong robustness and memory ability, and has simple learning rules and strong learning ability, which is easy to be realized by computer. But how to determine the structure and parameters of the network, there is no systematic law to follow, so the approximation performance of the network is affected. GA draws lessons from the survival of the fittest rule in natural genetics, and conducts global parallel and random search optimization in the problem space to make the global optimal convergence of the population. Compared with the traditional algorithm, GA training neural network does not need prior knowledge, and is not sensitive to the initial parameters, and will not fall into the local minimum. So GA with global search ability can optimize RBF neural network, find the optimal network structure and parameters, and ensure the best network performance.