基于ICEEMDAN模糊熵与Bi-LSTM的工业设备健康状态预测

工业设备健康状态关系着工业生产能否正常进行，为此提出一种基于改进自适应噪声完备经验模态分解（ICEEMDAN）和双向长短期记忆网络（Bi-LSTM）的工业设备健康状态预测方法。ICEEMDAN用于将原始音频信号进行分解得到若干个固有模态函数（IMF）分量，通过计算相关系数选取最佳分量组进行信号重构，然后计算重构IMF分量的模糊熵值构造特征向量集合，最后再输入到Bi-LSTM网络进行模型训练和预测。实验结果表明：相较于其他模型，基于ICEEMDAN模糊熵和Bi-LSTM的工业设备健康状态预测方法，能够有效提取音频信号特征，并准确进行健康状态预测。

Prediction of Industrial Equipment Health Status Based on ICEEMDAN Fuzzy Entropy and Bi-LSTM

The health status of industrial equipment is related to the normal operation of industrial production.Therefore,a method based on improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) and bidirectional long-term short-term memory network (Bi-LSTM) method was proposed for predicting the health status of industrial equipment.ICEEMDAN was used to decompose the original audio signal to obtain several intrinsic mode function (IMF) components,the best component group was selected by calculating the correlation coefficient for signal reconstruction,and then the fuzzy entropy structure of the reconstructed IMF component was calculated to reconstruct the feature vector set.The set of feature vectors was finally input to the Bi-LSTM network for model training and prediction.The experimental results show that,compared with other models,the health status prediction method of industrial equipment based on ICEEMDAN fuzzy entropy and Bi-LSTM can effectively extract the audio signal features and accurately predict the health status.