A BP Neural Network-Based Hierarchical Investment Risk Evaluation Method Considering the Uncertainty and Coupling for the Power Grid

Investment decision-making is affected by the uncertain and highly coupled risks in the power grid, and the inaccurate risk evaluation results in the great economic losses of power companies. In order to improve the accuracy of risk evaluation and reduce the economic losses, a hierarchical risk evaluation method considering the uncertainty and coupling of risks is proposed in this paper. At the lower level, the uncertainty and time response of risks are taken into consideration for evaluating individual risks accurately in the power grid. Through the data processing of historical risk factors based on BP neural network, the distribution regularities of risk loss and probability of occurrence are performed. According to the partition of time period, risk losses expressed by the interval and corresponding probabilities in different time periods are identified. The evaluation results at the lower level are considered as the basis for risk management and the inputs at the upper level. At the upper level, a comprehensive evaluation of multiple risks is performed for evaluating the investment scheme accurately. A discretization method is developed to transform the inputs into the probability sequences, and the sequence operation theory is applied in the comprehensive evaluation of multiple risks that considers the coupling among risks. Extensive case studies are presented to validate the effectiveness of the proposed method for risk evaluation.