Objective With the advancement of intelligent natural gas pipeline networks in China, online simulation has become a core technology for optimizing pipeline operation and management. However, low-fidelity measurements caused by sensor drift, sensor stagnation, and other faults significantly undermine simulation accuracy. Robust data reconciliation mechanisms are urgently needed to ensure accurate and stable simulation inputs.

Methods A high-fidelity data reconciliation method was proposed for the online simulation of compressor station process systems. An online simulation model for compressor stations was first constructed. Physical balance equations describing in-station gas flow were adopted as equality constraints, and redundant measured data were used to enhance simulation reliability of the compressor station’s actual operating state. To address sensor faults, correntropy was introduced as an index for evaluating measurement fidelity. By maximizing the similarity between redundant measurements and the mechanistic model, abnormal data were automatically identified and suppressed during iterative optimization, thereby effectively ensuring the accuracy and stability of input data for online simulation.

Results A compressor station along the Shaanxi–Beijing natural gas pipeline was selected for validation using measured data. Validation results indicated that incorporating redundant pressure measurements into the online simulation model reduced the overall root mean square error of simulated values by 22％ compared to the model without redundant data. Based on the correlation between process flow and sensor measurements, multiple test cases involving anomalies in single and multiple correlated measurements were conducted. In all cases, low-fidelity fault measurements were effectively identified and eliminated, confirming the efficacy of the correntropy-based high-fidelity data reconciliation method in multi-sensor fault scenarios.

Conclusion For normal measurement data, the online simulation model effectively weakens the impact of random noise on simulation reliability by incorporating redundant data. When low-fidelity measurements exist, the correntropy-based high-fidelity data reconciliation method significantly improves the reliability and interpretability of abnormal measurement identification. The proposed method effectively enhances the robustness of online simulation for compressor station process systems, providing technical support for the stable operation of online simulation in natural gas pipeline network systems.