Covariant offset correction method for water chilling unit LSTM fault diagnosis

The invention discloses a covariable offset correction method for water chilling unit LSTM fault diagnosis, and belongs to the field of water chilling unit fault diagnosis methods. Aiming at the defect that a single neural network, such as a 1DCNN network, cannot fully extract data features, a deep learning fault diagnosis method combining an LSTM network and the 1DCNN is selected. According to the method, the advantages of extracting sample local features by 1DCNN and processing a sample time sequence by LSTM are combined, and data sample features are fully extracted from space and time dimensions; aiming at the problems of unstable network training process and overfitting caused by internal covariable offset in a neural network, a layer normalization (LN) technology is utilized to perform normalization operation on data feature information before the data feature information enters an LSTM layer, so that the problem of network training overfitting caused by covariable offset is effectively solved, the diagn