The measurement accuracy of ultrasonic flow meters installed in multiproduct pipelines is reduced under complex operating conditions, due to fluid composition, temperature-pressure fluctuations, and installation factors. To address this problem, an intelligent online calibration method was proposed based on Stacking ensemble learning. The mass flowmeter-derived volumetric flow was used as the benchmark to achieve dynamic calibration and enhance measurement precision and reliability. A Stacking ensemble framework was constructed with nine machine learning algorithms including random forest (RF), K-nearest neighbors (KNN), and extreme gradient boosting (XGBoost) as candidate base learners. Over 80 000 field measurements from a southwest China multiproduct pipeline were collected. Models were trained and evaluated after data preprocessing. Based on coefficient of determination (R²) and other metrics, the three optimal models were selected as base learners. Linear regression was adopted as meta-learner for weighted integration to construct the Stacking calibration model. Performance was assessed via R², mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) metrics. Case studies demonstrate that an R² of 0.999 1 is achieved by the Stacking model, outperforming individual base learners. The MAPE is merely 0.216%, indicating that flow measurement bias is effectively reduced. Absolute errors are mainly distributed within the 0~1 m³/h range, exhibiting a right-skewed distribution with minimal tail extension, indicating superior prediction accuracy and stability. The advantages of multiple heterogeneous learners are effectively integrated by this intelligent calibration model for ultrasonic flow meters. Therefore, measurement accuracy is enhanced and a viable solution is provided for online flowmeter calibration in complex industrial environments. This approach also offers reliable technical support for safe and economical pipeline operations while providing theoretical foundation and practical guidance for intelligent industrial instrument calibration.