基于匹配滤波模型神经网络在数控机床故障诊断中的应用

Application of MBAM Neural Network in Computer Numerical Control Machine Fault Diagnosis

为了改善双向联想记忆(BAM)神经网络的性能,提出了一种修正模型.该模型能增强神经网络的记忆容量和容错联想能力,具有渐进稳定的特征,并且改进了网络平衡状态的稳定性和吸引性能.理论分析和实验结果证明,这种修正模型不仅能正确完成数控(CNC)机床的故障诊断,而且对于存在干扰的输入信号序列具有很好的容错联想能力.

To improve the bidirectional associative memroy (BAM) performance, a modified model was used to enhance neural networks' memory capacity and error correction capability. The modified model (MBAM) is shown to be asymptotically stable. Theoretical analysis and experiment results demonstrate that the modified model performs much better than the original BAM in terms of memory capacity and the modified model not only can complete fault diagnosis correctly but also have fairly high error correction capability for disturbed input information sequence. Moreover, MBAM model is more convenient and effective for solving the problem of computer numerical control (CNC) electric system fault diagnosis.