基于多种群遗传神经网络的船舶发电机故障诊断

Fault diagnosis of ship generators based on multi- population genetic neural network

为及时发现船舶发电系统的早期故障,通过多种群遗传算法与反向传播(BackPropagation,BP)神经网络算法相结合,提出一种基于多种群遗传神经网络算法的船舶发电机故障诊断方法.利用该算法对实例进行故障诊断,结果证明该算法能有效克服BP神经网络收敛速度慢和易出现局部极小值的缺点.该算法有全局搜索能力强、优化速度快的特点,具有一定的应用前景.

To detect early faults of ship power systems timely, through the combination of the multi-popu- lation genetic algorithm and the BackPropagation （BP） neural network algorithm, a fault diagnosis method for ship generators based on a multi-population genetic neural network algorithm is proposed. The fault diagnosis for real examples is done by the algorithm. The results show that the algorithm can effectively overcome the weaknesses of BP neural network that has a slow convergence speed and is prone to a local minimum value. The algorithm is of strong global search capability and a rapid optimization speed, and is of certain application prospect.