摘要
神经网络结构简单,但训练容易陷入局部最优解,而遗传算法依照生物进化理论将群体进行选择交叉变异从而求取全局解。首先讨论了神经网络与遗传算法的结合方法,并针对传统遗传算法中存在的问题,采取一种改进的遗传算法,将两种智能控制方法相结合进行研究,仿真结果表明此算法能有效改善系统的响应指标。
The struction of Neural network is simple,but it can easily get into local optimization in the Neural network training.While genetic algorithm(GA)can perform selection and crossover and mutation,thus it can get global solution according to the theory of biology evolution.Combination method of GA and neural network has been discussed.Aimed at the shortcomings of tradional GA,an improved GA method is applied combined with neural network.Simulations show that this algorithm can greatly improve performance specifications of the system.
出处
《火力与指挥控制》
CSCD
北大核心
2012年第6期170-172,共3页
Fire Control & Command Control
关键词
神经网络
遗传算法
自适应
交叉概率
变异概率
neural network
genetic algorithm
self-adaptive
crossover probability
mutation probability