Prediction of industrial VOCs based on LSTM for multi- monitoring stations

The prediction technology of VOCs (volatile organic compounds) from industrial sources is very important for monitoring and early warning. At present, few studies have focused on prediction using multi-site VOCs data, and there are few relevant studies on whether the statistics of VOCs monitoring data have an impact on the prediction effect of deep learning model. This study selected the appropriate multi-monitoring-site data to train the LSTM (Long Short-term Memory) model, and statistical data of VOCs for four monitoring stations was used to analyze the correlation between prediction performance, so as to improve the prediction effect of the LSTM model. The results showed that the statistics of VOCs observation stations can provide guidance on the predictive performance of the model to a certain extent, thereby improving the predictive performance of the model.