| 优化汽车故障诊断神经网络的改进遗传算法 |
| |
| 引用本文： | 苏子林,车忠志,邱景炜.优化汽车故障诊断神经网络的改进遗传算法[J].鲁东大学学报,2011,27(1). |
| |
| 作者姓名： | 苏子林 车忠志 邱景炜 |
| |
| 作者单位： | 鲁东大学交通学院;中国农业大学烟台研究院; |
| |
| 摘 要： | 为了优化汽车故障诊断神经网络的权值和网络结构,改进并设计了遗传算法.在保证种群多样性的条件下,将种群规模降低到最小;引入BP算子,与交叉和变异算子共同竞争,自适应调整运算概率;引入个体间的广义海明距离用于保持种群的多样性,引入连接权的敏感度用于优化网络结构.算法对比测试表明,改进后的遗传算法能够在优化权值的同时优化网络结构,减少了内存占用量,保证了种群的多样性,提高了算法的运行速度和收敛效果.
|
| 关 键 词： | 神经网络 改进遗传算法 BP算法 汽车故障诊断 |
An Improved Genetic Algorithm for Optimizing Automobile Fault Diagnosis Neural Network |
| |
| SU Zi-lin,CHE Zhong-zhi,QIU Jing-wei.An Improved Genetic Algorithm for Optimizing Automobile Fault Diagnosis Neural Network[J].Ludong University Journal (Natural Science Edition),2011,27(1). |
| |
| Authors: | SU Zi-lin CHE Zhong-zhi QIU Jing-wei |
| |
| Institution: | SU Zi-lin1,CHE Zhong-zhi2,QIU Jing-wei1(1.School of Traffic,Ludong University,Yantai 264039,China,2.Research Institute in Yantai,China Agricultural University,Yantai 264670,China) |
| |
| Abstract: | To optimize automobile fault diagnosis neural network's weights and structure,a genetic algorithm is improved and designed.On the condition of keeping population's diversity,the population size is minimized.BP operator is introduced,which mutually competes with crossover operator and mutation operator,adaptively adjust operation probability.To keep population's diversity,the generalized Hamming distance between individuals is introduced,and the weights' sensitivity is used to optimize neural network's struc... |
| |
| Keywords: | neural network improved genetic algorithm BP algorithm automobile fault diagnosis |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
