【摘要】 对柴油机表面振动加速度信号进行检测,经时频分析、小波分析提取信号的特征量,并将其作为BP神经网络的输入端,用神经网络方法对故障信号的冲击类型进行分类和诊断.分析表明,发动机多故障的冲击为进气门异常状态、排气门异常状态和爆燃冲击状态,这与实测的发动机运行状态完全一致.由此证实基于小波和神经网络相结合的分析方法能准确地识别发动机多故障状态的冲击类型,准确区分正常状态、故障状态及故障类型.更多还原

【Abstract】 The surface acceleration vibration signals of diesel engine are tested. The signal characteristic parameters are extracted through time-frequency analysis and wavelet analysis,and acted as the input of neural-net to classify and diagnose the impact styles of fault signals. The analysis results show that multi-fault impact signals of engine are inlet valve fault,outlet valve fault and deflagration impact,which is consistent with the fact. It makes sure that the method combining wavelet with neural-net can identify multi-fault impact styles of engine,and distinguish normal and fault state.更多还原