基于小波、WAE和LSTM的压裂车故障诊断

Fracture Truck Fault Diagnosis Based on Wavelet,WAE and LSTM

动力系统作为压裂车的关键部件,其工作状况直接影响着压裂车的性能,压裂车工况多变,使得其动力系统故障诊断更加复杂。为解决压裂车动力系统振动信号的强时变性和强噪声特性而造成其故障难以辨识的问题,提出了一个基于提升多小波包(LMWP)、小波自编码器(WAE)和长短时记忆网络(LSTM)方法。首先对压裂车动力端采集的振动信号进行3层提升多小波包分解;其次计算各子频带的相对能量,构成原始特征向量;最后将原始特征向量经WAE降维,并输入LSTM网络实现压裂车动力系统故障诊断。试验结果表明,提出的故障诊断方法在不同工况下能够实现99%以上的诊断准确率,具有优于传统方法较强的泛化能力、特征提取能力和识别能力。所得结论可为压裂车动力系统诊断方法的进一步发展提供参考。

As the key component of the fracturing truck,the power system directly affects the performance of the fracturing truck.The changing working conditions of the fracturing truck makes the fault diagnosis of the power system more complicated.To address the difficult fault diagnosis due to the strong time-varying and strong noise characteristics of the vibration signal of the fracturing truck power system,a method based on lifting multi-wavelet packet(LMWP),wavelet auto-encoder(Wavelet)and Long Short-term Memory(LSTM)methods are proposed.Firstly,the vibration signal acquired at the power end of the fracturing truck is subjected to 3-layer lifting multi-wavelet packet decomposition.Secondly,the relative energy of each sub-band is calculated to form the original feature vector.Finally,the original feature vector is dimensionality reduced by WAE and input into the LSTM network to realize fracture truck power system fault diagnosis.The experimental results show that the proposed method can achieve more than 99%diagnostic accuracy under different working conditions,and has better generalization ability,feature extraction ability and recognition ability than traditional methods.The study can provide references for the further development of the diagnostic method of the fracturing truck power system.