The construction of energy system digital twin relies on accurate models. This paper proposes a new modeling method for natural gas station hydraulic systems by integrating physical models and data-driven models to improve the accuracy of models. Taking a natural gas station of long-distance gas pipeline as an example, this paper builds the physical model of a station, including compressors and regulating valves. Then a hydraulic calculation algorithm of the station is developed. A data-driven model Back Propagation Neural Network (BPNN) is introduced for physical model error compensation. Finally, the calculation results show that the hybrid model has better accuracy than the physical model and the energy consumption of key equipment, such as compressors, air coolers in the station is monitored. Moreover, the hybrid model better integrates the advantages of the two types of models, it can serve as a soft sensor to enrich the status monitoring data of station equipment and lay the foundation for further optimization of station energy consumption.
Keywords energy consumption monitor, natural gas transportation, intelligent model, digital twin