Abstract: To enhance the calibration efficiency in phased array antenna production, a clustering-based calibration method is proposed. By employing hierarchical clustering algorithms to perform feature extraction on massive phased array antenna calibration datasets, channel grouping patterns with phase coherence characteristics are identified. Calibration optimization is implemented based on the established phase deviation-calibration grouping correlation model, effectively balancing the requirements for testing efficiency and accuracy. This approach reduces the required test points to 10% of conventional methods while significantly improving calibration throughput, thereby providing robust technical support for large-scale manufacturing of phased array antennas.