PROGNOSTIC ALGORITHM FOR CONDITION MONITORING AND ESTIMATION OF REMAINING USEFUL LIFE OF CRITICAL ELECTRICAL INFRASTRUCTURE

Abstract

Aerial Bundled Cables (ABCs) are multi-layer insulated bundled cables. XLPE insulation makes these cables less prone to pilferage. However, these cables are degraded rapidly in a coastal environment. The ABCs are insulated cables and are more frequently used and preferred over the traditional bare conductors in power-associated networks as they are highly reliable and safer than bare conductors with better operational expenses and minimum power losses. Despite all these benefits, ABCs are prone to various categories of failures after installation in coastal regions. Since ABCs conductors are concealed under the XLPE insulation, the visual inspection of most damages that occur in it becomes impractical. This limitation creates a need for a highly advanced and sophisticated inspection method to assess the degradation growth estimation of such electrical infrastructures. This thesis has proposed a genuine framework to categorize the sound and degraded ABC cables of various degrees positioned in areas near the shoreline. The progressive damage (ABC cable insulation degradation and erosion of hidden bundled conductors) is detected utilizing the ultrasonic probe listening method in installed operational ABC cables. The built-in superheterodyne module of UT-probe translates the inaudible supersonic partial discharge signal into audible range. The captured NDT-based UT data is utilized in this research work for the diagnosis and prognosis of onshore ABC cables. The UT probe's recorded partial discharge corona signal is highly scattered, discontinuous, and movable (or non-stationary) across the ABC cable. The Empirical Mode Decomposition (EMD) along with the Hilbert Huang Transform (HHT) scheme is utilized to extract the time-frequency-energy characteristics present in the recorded UT signal. The presented research approach can be utilized to assess the cumulative deterioration or degradation growth rate of ABC cables placed in different areas near seashore under harsh and corrosive climatic conditions. Further, the benefits of other states of the art Intrinsic Mode Functions (IMFs) extraction techniques such as Ensemble EMD (EEMD) and Complementary Ensemble EMD (CEEMD) production process is investigated for assessment of damaged live ABC cables. Hence, corresponding IMFs utilized to determine the Hilbert Huang Transform (HHT) spectrums. The periodically obtained ultrasonic signals data from various locations of the coastal region are used for the computation of HHT spectrums. A continuous progressive shift towards lower frequency ranges is observed in all three IMF extraction modes in the signal energy spectrum. This assists in determining and classification of the health state of operational ABCs. Although all the above-mentioned IMF-extraction techniques are very effective for the damage identification of in-service ABCs, the CEEMD is considered the most suitable technique among the aforementioned three techniques. The presented scheme is equally beneficial for the diagnosis and identification of the level of damage and for determining the health state classification and prediction of the probable remaining useful life of in-service ABCs. Furthermore, the advanced knowledge of insulation damage is crucial for the upkeep of electrical infrastructure, including insulated power distribution lines (Aerial Bundled Cables). The preliminary damage information helps power distribution companies reduce their maintenance costs by avoiding unnecessary inspections and maintenance activities. The preliminary damage information can be obtained by utilizing a prognostic model that can predict the future degradation growth rate (Remaining Useful Life Factor) in the insulation of active Aerial Bundled Cables operational cables installed in coastal regions. This thesis presents the prognosis study of the energized ABCs failures under the harsh climate of coastal areas for the first time. The novel Hybrid resampling scheme-based particle filter (PF) algorithm is proposed in the second phase of research. The hybrid-resampling scheme is a combination of classical Multinomial resampling and Residual resampling schemes. This scheme is introduced in the resampling step of the Particle Filter algorithm. The selection of each resampling scheme is based on the degeneracy rate of each particle. The performance evaluation of the Hybrid resampling scheme-based PF algorithm is done by inserting the artificial noise into the measurement data. The promising results highlight the efficacy of the proposed predictive algorithm. In the third phase of the research, the posterior density function of degradation growth rate (Remaining Useful Life Factor) in the cable insulation is estimated using the proposed novel Hybrid resampling scheme-based particle filter (PF) algorithm. Furthermore, this improved scheme under the framework of the /-step prediction algorithm is applied on the actual super heterodyned Ultrasonic listening-based field acquired data of installed ABCs. The use of actual measurement data introduces more accuracy as it captures the degradation trend in the cable insulation without any assumptions. PF well handles the nonlinear state transitions and measurement functions for non- Gaussian / multimodal noise distributions. The proposed Hybrid resampling scheme- based PF algorithm is further complemented by a step error calculation method to analyze the prediction accuracy in the absence of measurement data. The promising results depict the benefits of the proposed technique.