摘要
提出了一种基于改进遗传算法（Improved Genetic Algorithm,IGA）优化的径向基函数（RBF）神经网络,将实数编码的自适应交叉和变异操作的遗传算法与梯度下降法混合交互运算,作为RBF网络的学习算法,并应用于非线性函数的逼近和导弹故障模式的识别问题。仿真结果表明,基于IGA算法的RBF神经网络不仅结构简单,而且具有较好的网络泛化性能。
In this paper,a radial basis function （RBF） neural network based on improved genetic algorithm （IGA） was proposed. A hybrid learning algorithm that incorporated the real-coded genetic algorithm with adaptive crossover and mutation into the gradient-dropping algorithm was presented to optimize the RBF neural network. And the simulation experiments about approximation problem of nonlinear function and pattern recognition of missile＇s failure were done. The simulation results show that the RBF neural network based on IGA not only has the advantages of simple structure and fast learning,but also has better generalization performance.
出处
《海军航空工程学院学报》
2010年第3期271-275,共5页
Journal of Naval Aeronautical and Astronautical University
关键词
RBF神经网络
梯度下降法
遗传算法
自适应
RBF neural network
gradient-dropping algorithm
genetic algorithm
adaptation
