A data-driven state identification method for intelligent control of the joint station export system.

As a necessary part of intelligent control of a joint station, the automatic identification of abnormal conditions and automatic adjustment of operation schemes need to judge the running state of the system. In this paper, a combination of Particle Swarm Optimization (PSO) and Gray Wolf Optimizer (GWO) is proposed to optimize the Backpropagation Neural Network (BP) model (PSO-GWO-BP) and a pressure drop prediction model for the joint station export system is established using PSO-GWO-BP. Compared with the traditional hydraulic calculation modified (THCM) models and other machine learning algorithms, the PSO-GWO-BP model has significant advantages in prediction accuracy. Based on the PSO-GWO-BP pressure drop prediction model, the determination method of state identification threshold is established, and a state identification method based on dynamic threshold is proposed, which realizes the intelligent identification of the system operation state by automatically adjusting the threshold. Through the analysis of the production and operation data of the joint station, the abnormal working conditions are successfully identified, and the effectiveness and accuracy of the method are verified. This method not only enhances the ability to discriminate abnormal working conditions but also adaptively adjusts the operation scheme, which effectively improves the intelligence level of the joint station export system.