【摘要】 在机械故障诊断中,对并发的故障进行诊断是一个备受关注的问题。为此,提出了用并行遗传神经网络诊断并发的故障的新方法。介绍了并行神经网络的结构、遗传算法的建模原理及并行遗传神经网络的结构和诊断机理。以某大型旋转机械为诊断对象讨论了该方法的实现技术,并与传统的BP(back-propagation)网络用于故障诊断的方法进行了比较。实验仿真结果表明,该方法在训练网络过程中能够避免网络陷入局部极小,加快网络训练的速度,减少诊断时间。更多还原

【Abstract】 Diagnosis of concurrent malfunction is a concerned problem in diagnosing fault of mechanical equipment. A method based on parallel genetic network with which the problem had been solved, was presented. The configuration of parallel neural Network, the modeling principle of genetic, the configuration of parallel genetic neural network and the law were all expounded. Then, the technique of utilizing the method was discussed, a diagnosis system to the circumvolving machine developed and the traditional way of BP network for the fault diagnosis compared. The result of emluator expatiated that this method can avoid getting into local minimum during educating the network, quicken the velocity and reduce the time of diagnosis.更多还原