Power system risk assessment strategy based on weighted comprehensive allocation and improved BP neural network

Abstract

During the operation and maintenance of the power system, power outages and supply‐demand imbalances can disrupt the normal power supply process. This issue must be mitigated or even resolved through the implementation of an appropriate power system risk warning. The article proposes a self‐assessment and early warning strategy for power system hazards based on an enhanced ant colony optimization algorithm (IACO) and a BP neural network. First, a combination of the Analytic Hierarchy Process (AHP) and the Entropy Weighting Method (EWM) is used to assign weights comprehensively to indicators that have a significant impact on the stability and safety of power system operation, thereby avoiding the negative impact of subjective experience or objective factors on the weight allocation results. Secondly, multiple regression analysis is used to calculate the risk assessment results of the selected indicators and weights corresponding to the power system. Training and testing samples for the BP neural network were calculated based on the weight allocation procedure described previously. Then, IACO is employed to global optimize the weights and thresholds of the BP neural network, and an enhanced BP neural network model for independent power system risk assessment is developed. The designed risk assessment and warning strategy was finally evaluated. The results indicate that the proposed power system risk assessment and early warning method can precisely predict the actual operating status of the power system based on weight values, thereby enhancing power supply quality by providing technical personnel with a data reference.