Process-constrained performance monitoring and degradation identification for aluminum alloy heating furnace operation

This study proposes a process-constrained and load-compensated multidimensional monitoring method for degradation identification in aluminum alloy heating furnaces. Benchmark process groups are defined by furnace ID, alloy grade, heating setpoint, and soaking setpoint, and load-sensitive indicators are corrected using charge weight and the maximum furnace temperature before heating. Based on heating execution, temperature control, thermal-field uniformity, actuator load, and energy-efficiency performance, deviation scores and health indices are constructed for batch-level anomaly monitoring and process-group degradation assessment. Using full-year 2024 production data, 385 raw batches were screened into 374 valid batches, including 14 benchmark process groups and 306 analysis batches. Load correction reduced the within-group dispersion of total energy consumption, total duration, and specific energy consumption by 41.16%, 33.03%, and 31.99%, respectively. The proposed method identified 238 normal, 32 mildly abnormal, and 36 severely abnormal batches, and further detected 8 normal, 4 mildly degraded, and 2 severely degraded process groups. These results indicate that the method enhances the fairness, robustness, and interpretability of furnace performance monitoring under explicit process constraints.