基于声信号特征谱提取的发动机故障诊断系统设计

Design of Engine Fault Ddiagnosis System Based on Acoustic Signal Feature Spectrum Extraction

针对当前发动机故障造成的危害,提出一种基于小波阈值去燥和PNN分类的故障诊断方法。针对声信号中噪声问题,提出采用加权平均值函数的方式对声信号特征进行提取,然后结合PNN算法的优势,通过训练样本的训练,对发动机故障进行分类。最后以摩托车发动机故障为例,通过搭建发动机诊断系统,实现对发动机不同类型故障的诊断。

Aiming at the harm caused by engine faults,a fault diagnosis method based on wavelet threshold de-drying and PNN classification is proposed.Aiming at the problem of noise in acoustic signals,a weighted average function is proposed to extract the characteristics of acoustic signals.Then,combined with the advantages of PNN algorithm,engine faults are classified by training samples.Finally,taking motorcycle engine fault as an example,the engine fault diagnosis system is built to realize the diagnosis of different types of engine fault.