MONITORING DATA PROCESSING METHODS, TREND FORECASTING METHODS AND SYSTEMS

The present disclosure discloses a monitoring data processing method and system, and a trend prediction method and system. The system includes an obtaining module, a prediction module, a determining module, and a processing module. First, the obtaining module preprocesses obtained monitoring data, and the prediction module predicts a changing trend of a monitored object by using a trained GRU model. Then, the determining module determines whether the changing trend is abnormal; when the changing trend is abnormal, the determining module determines nonlinear jump data in the monitoring data by using a self-attention mechanism network. The processing module then determines whether the jump data is noisy data; if yes, the processing module replaces the jump data through interpolation. Finally, the prediction module performs trend prediction again based on the monitoring data processed by the processing module. This can improve the trend prediction accuracy of the monitoring data and avoid false trend warnings.