Abstract: In scenarios such as harmonic distortion, electromagnetic disturbance, and link attenuation, electric energy metering systems are prone to structural deviations. The metering output presents a time-frequency coupling form, with ratio difference offset, phase drift, pulse variation, and communication frame disorder forming a correlated distribution, which prompts the diagnostic task to shift from single-domain determination to multi-domain modeling. This paper constructs a multi-domain diagnostic framework around the abnormal measurement of electric energy meters. It forms cross-domain representations through hierarchical feature fusion, extracts and analyzes harmonic trajectories, jump fragments, and link jitter features with dynamic patterns, generates multi-stage discrimination domains with adaptive discrimination structures, and conducts verification and analysis using ratio difference, phase offset, pulse consistency rate, and frame efficiency rate. Enable the measurement anomalies to form analyzable structures at different disturbance levels.