The high redundancy of the measured data from heating, ventilation, and air conditioning (HVAC) systems significantly reduces the computational efficiency of model calibration. To address this challenge, a model calibration method based on mining feature operating conditions and a priori probability guidance was introduced in this study. Correlation analysis was conducted on the operational data for mining feature operating conditions. Feature variables related to HVAC system operation were selected, and a grid sampling technique based on these characteristic variables was employed to obtain representative operating conditions, enhancing the efficiency of the model calculations. Additionally, a prior probability model was established for the parameters to be calibrated during the model calibration process. A priori interval estimation was then performed, and the objective function was improved based on the prior probability to guide the model towards faster convergence. The proposed method was validated using a one-month operational dataset from a cooling plant in an industrial building located in Wuhan, China. The results indicated that the proposed method achieved significant improvements in performance metrics. Specifically, mean absolute percentage error (MAPE) and cross-validated root mean square error (CV-RMSE) were reduced by 16.0% and 12.0%, respectively, compared to the K-means clustering-based method, and by 20.9% and 15.2%, respectively, compared to the baseline data-based method. Furthermore, the normalized mean bias error (NMBE) was closer to zero, and the coefficient of determination (R²) increased by 4.7% and 8.5%, respectively, compared to the two aforementioned methods. Additionally, our method enhanced the computational efficiency by approximately 39.3%. This method provides technical guidance and data support for achieving an efficient and accurate modeling of HVAC systems.