A Causal Auditing Paradigm for Sub-health Diagnosis of Offshore Wind Assets via Multi-terminal Harmonic Fingerprinting

Core Content Description Research Context: The paper addresses the limitations of traditional reactive maintenance and black-box stochastic prediction models in highly power-electronic offshore wind systems, where marine environments and capacitive coupling introduce significant spectral turbulence. Proposed Framework: It introduces the "Electromagnetic Ledger" framework under the Clark Paradigm. This establishes a deterministic causal link between the generator (Node A), converter (Node B), tower-base switchgear (Node C1), and the remote collector entrance (Node C2) by extracting harmonic fingerprints from the 1st to the 20th orders. Technical Methodology: Causal Auditing: The framework transitions diagnostic focus from passive observation to the deterministic verification of internal physical laws, utilizing an audit residue () to track health states in real-time. Algorithmic Engine: An online iterative calibration is executed via a Recursive Least Squares (RLS) operator for dynamic equilibrium. Multi-head attention mechanics are also deployed to disentangle intricate multi-dimensional resonance cross-talk within the feature space. Decision Philosophy: The system institutes a "90% Prediction + 10% Safety Margin" threshold. A breach of the 10% critical boundary by the residual triggers preemptive maintenance, maximizing residual asset value while neutralizing the risk of collapse. Hardware Architecture: The implementation features a distributed architecture decoupling the perception layer (industrial MCUs handling high-sync sampling) from the audit layer (edge hosts running the causal engine). Computational latency for a single audit iteration is kept below 5ms. Experimental Results: Validation via a Digital Twin platform of an demonstrates that the Clark Paradigm exhibits superior robustness compared to conventional LSTM models under extreme wind transients. It effectively isolates early precursors of XLPE insulation degradation and IGBT anomalies.