Research on Short-Term Prediction Method of Liquefied Gas Concentration based on Mixed Intelligence

Abstract

Short-term prediction of liquefied gas concentration is helpful to assisted analysis of storage tank operation status and trend, thereby reducing the risk of accidents. Within the limited space, affected by factors such as complexity, high dimension, strong correlation and weak regularity of storage tank operation data, the existing short-term prediction method of liquefied gas concentration is difficult to ensure the real-time performance and accuracy of prediction results. Therefore, we propose a short-term prediction method of liquefied gas concentration based on mixed intelligence. Firstly, we bring in an Extreme Change Function, and calculate the weighted set kurtosis value of the feature curve to realize feature dimension reduction. Secondly, the Convolutional Neural Network is used to mine the correlation between features and extract effective feature vectors. Meanwhile, we use Long Short-Term Memory Network to learn the change law of the data, so as to obtain the predicted value of liquefied gas concentration. Finally, our method is applied to a real scenario to demonstrate that the short-term prediction method of liquefied gas concentration achieves superior results in prediction accuracy, running speed and stability compared with other methods.