Dynamic Spatio-Temporal Graph Neural Network-Based Wet Purification Key Parameters Real-Time Visualization Prediction Model

Aiming at the key measurement points in the post-treatment wet purification process using sensor measurement lag, a real-time visualization prediction model of key parameters of wet purification based on dynamic spatio-temporal graph neural network is proposed. The model contains dynamic graph construction, graph attention, time series modeling, online learning and visualization window. Taking the process operation data of wet purification process as the research object, the adaptability of the model to multiple working conditions is guaranteed by dynamic graph construction, the spatial relationship of parameters is extracted by using graph attention network (GAT), the temporal dependence of input operation data is captured by long and short-term memory neural network (LSTM), and temporal attention is used to pay attention to the abnormal parameter values, and at the same time, the results are updated in real time by using the on-line learning mechanism, and the Dash realizes the visualization monitoring. The model training and validation show that the prediction accuracy, visualization and other functions meet the engineering requirements. Therefore, the real-time visualization prediction model based on dynamic spatio-temporal graph neural network proposed in this paper can assist the operator's judgment and improve the safe and stable operation of the system.