粗糙集-遗传神经网络在挖掘机故障诊断中的应用研究

Application Research of Rough Set-Genetic Algorithm-Neural Network Algorithm in Excavator Fault Diagnosis

针对当前单一的故障诊断方法不能满足实际需求的问题,提出了一种粗糙集-遗传神经网络分类器模型,实现对挖掘机故障分类.该模型首先利用粗糙集理论对神经网络的输入进行属性约简,以减少神经网络的工作量;利用遗传算法优化BP神经网络,解决神经网络易陷入局部极小和收敛速度慢的问题;最后利用约简结果和优化的BP网络进行网络训练.实验结果验证了该方法用于故障诊断的有效性.

Aiming at a single fault diagnosis method can′t meet the actual need,a classification model were proposed,which based on rough set-genetic algorithm-neural network algorithm,to come true excavator fault diagnosis classification.Firstly,the attributes were reduced using rough set theory to choose neural network′s input parameters,which reduced the work and calculation time.Then,in order to solve the shortcoming in the back propagation algorithm,such as trapping to the local minimum and slowness in training speed,genetic algorithm was integrated to optimizing the BP network parameters.Finally,the model carried on the training by the reduction results and the optimized BP network parameters.The experimental result shows the effectiveness of the new proposed model.