Lithium battery health state monitoring method based on state attenuation and interactive learning

The invention relates to a lithium battery health state monitoring method based on state attenuation and interactive learning, and belongs to the technical field of batteries. The method comprises the following steps: constructing a missing data vector; extracting features of missing data of the lithium battery sequence; multi-resolution time feature extraction is carried out; and proportional fusion of the multi-scale features. A lithium battery time sequence belongs to one-dimensional data, various curves objectively and abstractly reflect working conditions of a lithium battery, and people cannot directly utilize a traditional data filling method to destroy nonlinearity of data. According to the research, DEC-LSTM and CIN are used as core modules, a hybrid neural network meeting SOH, RUL and other tasks is constructed, and accurate monitoring of the lithium battery health state is realized. According to the method, the authenticity of a data source and the accuracy of model prediction are both considered,