False Positive in Acoustic Leak Detection of a Flooded Heating Pipeline: A Case Study in Hybrid AI Diagnostics

This paper presents a field case study of a false positive leak verdict issued by conventional acoustic inspection equipment during the examination of a flooded district heating pipeline (L = 336 m, 426×7 mm, P = 1.6 MPa). Both a third-party analog defectoscope and all primary AI modules of the RIVIXI Diagnose platform — 2D-CNN (100.0%), 1D-CNN (98.3%), and classical cross-correlation (Z-Score 6.95) — independently classified intense hydrodynamic background noise as an active pipe leak. The system's correct final verdict — no excavation required — was produced exclusively by its hybrid physical layer: Peak-to-Noise Ratio (PNR = 8.64 < 15.0 threshold) and phase coherence (0.4% < 60.0% threshold). The excavation performed by the third-party crew under their emergency order confirmed no active leak existed. The case demonstrates the fundamental limitation of acoustic pattern classifiers and provides empirical justification for hybrid DSP + AI architectures combining physical signal metrics with neural network classification in pipeline diagnostics. Open audio recordings are provided for independent verification.