| 一种基于遗传算法的神经网络控制方法 |
| |
| 引用本文： | 曲慧杨,王清,马广富.一种基于遗传算法的神经网络控制方法[J].自动化技术与应用,2003,22(9):8-11. |
| |
| 作者姓名： | 曲慧杨 王清 马广富 |
| |
| 作者单位： | 哈尔滨工业大学,控制科学与工程系,黑龙江,哈尔滨,150001 |
| |
| 摘 要： | 本文比较了传统的BP算法和遗传算法用于神经网络设计的优缺点，阐明了遗传算法和神经网络相结合的必要性，提出了一种用遗传算法同时优化网络的结构和权值的神经网络控制方法，通过对遗传算法基本参数及骗码方案，遗传算子的设计，实现了权值与结构的同时优化，成功地应用于二级倒立摆系统的控制，仿真结果显示了这种遗传算法能够有效抑制早期收敛，以较快的速度与较高的精度达到全局快速收敛。
|
| 关 键 词： | 遗传算法 神经网络 二级倒立摆系统 |
| 文章编号： | 1003-7241(2003)09-0008-04 |
A Method of Neurocontrol Combining with Genetic Algorithms |
| |
| QU Hui-yang,WANG Qing,MA Guan-fu.A Method of Neurocontrol Combining with Genetic Algorithms[J].Techniques of Automation and Applications,2003,22(9):8-11. |
| |
| Authors: | QU Hui-yang WANG Qing MA Guan-fu |
| |
| Abstract: | The comparison between conventional backpropagation(BP)algorithm and genetic algorithm in designing of neural networks is discussed.The necessity of combining genetic algorithm with neural networks is demonstrated.A kind of neurocontrol method is proposed in which the weights and the architecture of the neurocontroller are optimized by genetic algorithm.And the basic parameters,encoding scheme and genetic operators in genetic algorithm are all analyzed.The double inverted pendulum control is successfully realized by this method.The simulation result indicates that the premature convergence in genetic algorithm is restrained effectively.It achieves a rapid global convergence quickly and accurately. |
| |
| Keywords: | Genetic algorithm Neural networks Double inverted pendulum |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
