摘要
提出了将模拟退火算法融合于遗传算法中的组合方案,对传动装置监测信号小波能量谱进行编码,以克服利用BP算法进行样本训练时网络可能会陷入局部极小点的缺点.以某型传动装置监测信号的小波能量谱为训练样本,识别传动装置带有缺损的齿轮的故障征兆.计算结果表明,组合方案能够提高神经网络训练效率和故障征兆识别精度.
A method for identifying the gear fault of the transmission system is presented, where the simulated annealing algorithm and genetic algorithm are combined in order to prevent the BP network going into the local minimum points during the training work. By using the wavelet energy spectrum of the signal measured from a certain transmission system, the symptom of the gears with damnification is detected. The results show that both the training efficient and the identifying precision are increased obviously.
出处
《车辆与动力技术》
2006年第4期53-56,共4页
Vehicle & Power Technology
基金
武器装备预研基金项目(514571201)
关键词
模拟退火
遗传算法
神经网络
传动装置
genetic algorithm
neural network
simulated algorithm
transmission
