<p>For the construction of natural gas transmission pipeline digital twin, real-time simulation is crucial for obtaining gas flow state from measurements. However, random noise and gross errors in the station measurements significantly influence simulation results, while the measurement redundancy along the pipeline is limited for gross error detection. To address this issue, this study systematically analyzes the intrinsic mechanism of correntropy-based robust data reconciliation in compressor stations, with the trace of the information matrix employed as a key numerical diagnostic indicator. First, a reconciliation framework is established to estimate flow states from redundant pressure and flow measurements that may contain gross errors, with the physical laws of the station incorporated as equality constraints in the optimization model. Then, by introducing quantitative diagnostics such as the trace of the information matrix and process variance, this work analyses the internal conflict-evolution mechanism of reconciliation process in the station. The effectiveness of reconciliation is validated in a real-world station. The results demonstrate that the gross error in outlet station pressure measurements is effectively suppressed. Theoretically, the trace of information matrix is reduced by shortening the window width of correntropy function, thereby enhancing the consistency between reconciled data and the physical model. This study provides valuable insights for sensor fault detection and suppression in compressor station operation and monitoring.</p>