Input-delay-based Support Vector Machine for Pressure Prediction in Nitrogen Pipeline Networks

During the consumption of low-pressure nitrogen in a steel industry energy system, the influence of nitrogen consumption units on the system pressure can cause delay due to their dispersion and different locations. In order to predict the pipeline network pressure, a factor input-delay-based multiple kernel least squares support vector machine (LSSVM) is reported here. A causality theory-based method is proposed to calculate the delay time. This first calculates the delay time of the factors influencing the system pressure and constructs training samples for different factors with the corresponding delay time. Then, the LSSVM-based model is established for prediction. To verify the effectiveness of the proposed method, two different practical conditions are considered, a normal and an abnormal one, and numerical experiments are conducted to validate the high accuracy of the proposed method.