基于挤压-激励注意力机制的CNN-LSTM煤矿用电机异常状态检测

CNN-LSTM Coal Mine Motor Abnormal State Detection Based on Squeeze-Excitation Attention Mechanism

针对环境恶劣和复杂工况导致煤矿用电机故障频发,为保障煤矿用电机的可靠运行,提出一种基于挤压-激励注意力机制的CNN-LSTM煤矿用电机异常状态检测模型。首先引入卷积神经网络(CNN)获取多特征输入空间联系,采用长短期记忆网络(LSTM)提取序列时序变化特征,结合挤压-激励注意力机制(SE)为LSTM层进行自适应权重分配来增强电机定子电流的关键信息提取;然后通过均方根误差(RMSE)对电机定子电流进行残差分析,检测电机定子电流的异常变化;最后,以新疆某大型露天煤矿121带式输送机1#煤矿用电机实时运行状态数据验证所提方法的实用性,结果表明该方法能够精准检测到电机定子电流异常状态,为煤矿用电机可靠运行提供重要依据。

Aiming at the frequent occurrence of coal mine motor faults due to the harsh environment and complex working conditions,a CNN-LSTM coal mine motor abnormal state detection model based on the squeeze-excitation attention mechanism is proposed to ensure the reliable operation of coal mine motors.Firstly,a convolutional neural network(CNN)is introduced to obtain the multi-feature input spatial connection,and a long short-term memory network(LSTM)is used to extract the sequence temporal change features,which is combined with the squeezed-excited attention mechanism(SE)to enhance the key information extraction of the motor stator current with the adaptive weight allocation for the LSTM layer;the residual analysis of the motor stator current is carried out by the root-meansquare error(RMSE),and the abnormal state detection model is proposed to protect the reliable operation of coal mine motors,detect the abnormal change of motor stator current;finally,the real-time running state data of No.1 mining motor of 121 belt conveyor of a large open-pit coal mine in Xinjiang is used to verify the practicality of the proposed method,and the results show that the method can accurately detect the abnormal state of the motor stator current,which improves the important basis for the reliable operation of motors used in coal mines.