An Assistant Decision-Making Method for Multi-Stochastic Variable Programming Based on Case-Based Reasoning

Multi-stochastic variable programming (MSVP) is an important tool to solve the decision-making problem with multiple random variables. However, when there are a large number of nonlinear constraints and complex objective function, directly solving the programming model will consume a lot of computing resources and time costs, making the solution results difficult to meet the requirements of online decision-making, real-time decision-making and other specific conditions. In this paper, an assistant decision-making method for MSVP based on case-based reasoning is proposed to explore and utilize the similarity between various historical uncertain conditions and current uncertain condition, so that the historical optimal decisions can be applied to obtain the current optimal decision. Considering the distribution characteristics and correlations of random variables, the methods for formally representing the cases, measuring the similarity of different attributes in cases and measuring the overall similarity between different cases are designed. The effectiveness and superiority of the proposed method in decision quality and efficiency are verified by the application of microgrid stochastic optimal scheduling problem.