Fan spatio-temporal data anomaly detection method based on diffusion model

The invention relates to a fan spatio-temporal data anomaly detection method based on a diffusion model, and the method comprises the following steps: carrying out the correlation analysis of collected SCADA time series data of a wind turbine generator, and dividing the data into a w * w matrix through a sliding window mode; a GRUfusion model is constructed, the GRUfusion model comprises a self-attention mechanism module, a GRU module and a diffusion model, the w * w matrix is input into the GRU module to obtain time features, a transrank matrix of the w * w matrix is input into the self-attention mechanism module to obtain space features, the time features and the space features are subjected to feature data splicing through multi-channel fusion and then enter the diffusion model, and the time features and the space features enter the diffusion model. And outputting a prediction error by a diffusion model, and then carrying out anomaly judgment. The method is higher in accuracy and more effective in data spa