Bayesian Decision Network Based Optimal Selection of Hardening Strategies for Power Distribution Systems

Power distribution systems, composed of wood poles and wires, are susceptible to damage caused by strong winds during hurricanes or winter storms. Power outages induced by failed poles result in millions of revenue and restoration costs and impact millions of customers. Failure of aged wood poles is a major reason for power outages during hurricanes or winter storms. Replacing aging poles is an easy and effective approach but with a high cost to enhance radial power distribution poles. In this paper, the optimal enhancement strategy of the power distribution system was obtained through the Bayesian decision network (BDN), which is an ideal tool to deal with trade-off problems. Besides the economic benefit of the electricity utility company, the benefits of residential customers, namely the power supply reliability, were also integrated into the reference of decision makers through BDN. To determine priorities of the pole replacements, the component importance index of individual poles, instead of the commonly used component importance index of a line, was explored using physics-based reliability analysis rather than an identical mathematical model applied to all poles to describe their vulnerabilities. A surrogate model established by Bayesian regularization neural network (BRNN) was employed to save the computational cost in the Monte Carlo simulation of reliability analysis. The cost of the hardening strategy using the component importance index of each individual pole was about 80% of that with the component importance index of a line, which indicates that the component importance index of each individual pole is more effective. The optimal choice of strategies would vary with the utility function that describes the decision-making references because benefits between the utility company and customers might conflict with each other. When the utility company and customers are given the same priority in the hardening process against Category 5 hurricanes, the optimal pole replacement percentage was 30%. (C) 2022 American Society of Civil Engineers.