基于蚁群算法和神经网络的数控机床故障诊断技术研究

Research on CNC Machine Fault Diagnosis Based on Ant Colony Algorithm and Neural Network

为了克服BP神经网络收敛速度慢、易于陷入局部极小点的缺点,在研究蚁群算法优化神经网络训练算法的基础上,以数控机床的进给伺服系统故障诊断为例,建立其故障诊断模型。利用训练后的蚁群神经网络对其进行故障诊断,并把BP神经网络和蚁群神经网络的训练和诊断结果相比较。实验结果表明:蚁群神经网络比BP神经网络的收敛速度快、运算效率高、识别能力强。这说明蚁群神经网络应用于数控机床的故障诊断中,可有效地提高故障诊断的准确度和效率,具有良好的应用效果。

In order to overcome the shortcomings of slow convergence speed and easy falling into the local minimum points in the BP neural network,based on the research of ant colony algorithm to optimizate neural network training algorithm,it takes CNC machine tool feed servo system fault diagnosis as example to establish the fault diagnosis model.The fault of feed servo system is diagnosed by trained ant colony neural network,and the training and diagnosis results of the BP neural network and the ant colony neural network are comparied.The result shows that the ant colony neural network has the advantages of more quick convergence speed,higher operation efficiency,stronger identification ability than BP neural network.These show that the ant colony neural used in the fault diagnosis of CNC machine tool,which can effectively improve the accuracy of fault diagnosis and efficiency,has good application prospects.