Predictive Reliability Analysis of Power Distribution Systems Considering the Effects of Seasonal Factors on Outage Data Using Weibull Analysis Combined With Polynomial Regression

In this study, the reliability of power distribution systems is analyzed using a novel strategy of predictive reliability analysis based on the lifetime failure rate cycle in a bathtub curve shape and considering the standard Weibull distribution to determine the trend of the failure rate in each period using median rank regression for parameter estimation. The proposed strategy consists of three processes. The first process involves separating the external and internal factors that influence power outages in the power distribution system from the seasonal multimodal shape in the empirical distribution of the dataset using a bisection algorithm of residuals of polynomial regression. Second, clustering and characterization of each component in the power distribution system according to the condition of the total factor bathtub curve leads to the introduction of the use of shape parameters as the total factor deterioration index (TFDI) with linear regression trends of log scale shape parameters of the useful period. A simple approximation of the system’s overall total factor bathtub curve using a sixty-year forecast is the final process presented that can be used in reliability planning to address lifecycle risks. The actual time-to-outage dataset between 2015 and 2020 of the Provincial Electricity Authority, Region 1, Northeastern Thailand, which covers the area of distribution line life in the three periods of the bathtub curve, was used as the test data. The numerical results obtained from the proposed process provide a comprehensive prediction of the reliability of the electrical distribution system for risk response planning. The results show the proportion and amount of internal deterioration versus external disturbances, helps to group components according to health and usability and prioritizes them according to risk. Furthermore, it clarifies the important moments of status transition. All of these factors make it possible to improve reliability in the right place at the right time. Every method that we have chosen to improve for use in analysis is simple, provides a clear visualization of every step, and can be used in practice.