Optimizing Hybrid Ultrasonic Testing Algorithms for District Heating Networks under Incomplete Spatial Metadata

This paper addresses false positives generated by classical cross-correlation algorithms in acoustic leak detection systems when operating over long distances (L > 60 m) in municipal district heating networks. Under high attenuation and low signal-to-noise ratios, cross-correlation Z-score values saturate and generate spurious alarms driven by common-mode environmental noise. A systematic audit of a five-year operational database revealed that only 2.9% of historical diagnostic records contain verified distance labels, making full meta-classifier retraining intractable. We propose a physics-informed post-decision arbitration layer — a hybrid algorithm integrating sensor separation distance as a physics-derived confidence constraint applied after the neural ensemble verdict, overriding the analog correlator alarm when acoustic propagation physics predicts unreliable correlation at the observed span. Experimentally verified on 110 real-world acoustic records from a district heating network, the method reduced false positives by 20.9% (from 43 to 34) while fully preserving leak detection sensitivity (Recall: 63.3%), with overall accuracy improving from 50.9% to 59.1% (p = 0.0039, McNemar's test).