An Integrated Health Index, Remaining Life Assessment, and Probability of Failure Framework for Critical Electrical Assets

Abstract

Unplanned outages of critical electrical assets such as transformers, gas-insulated switchgear (GIS), motors, and power cables impose severe operational, financial, and safety penalties on modern power networks. Conventional condition assessment practice relies on discrete, periodic inspections that fail to capture the dynamic operational and environmental stresses that accelerate ageing. This paper presents an integrated, microservice-based framework that unifies three complementary analytics: (i) a Dynamic Health Index (HI) Service that fuses heterogeneous diagnostic measurements into a single weighted condition score and adjusts it for real-world stress through a Conditional Factor (CF) and a Weibull survival function; (ii) a Remaining Life Assessment (RLA) Service that fits a hybrid Weibull–polynomial degradation model to smoothed historical HI trajectories and projects Remaining Useful Life (RUL) by threshold crossing; and (iii) a Probability of Failure (PoF) Service that couples survival analysis with Monte Carlo uncertainty propagation to express failure risk over finite horizons. We further describe a partial-discharge (PD) monitoring subsystem that supplies a key diagnostic input through five-minute phase resolved partial discharge (PRPD) snapshots. The HI engine is realised as a configuration-driven pipeline: JSON-defined parameter rules drive data simulation and scoring, a gradient boosted regressor (LightGBM) learns the mapping from raw measurements to HI, and the model is exported to the ONNX format for portable, low-latency inference behind a REST API. The resulting platform enables risk based maintenance planning, prioritisation of ageing assets, and improved capital-expenditure forecasting.