The integration of distributed power sources lends more challenges to the reliability assessment of the grid. To address issues of time-consuming and high error in existing distribution network reliability assessment methods, this paper proposes a methodology for grid reliability assessment based on an improved Ali Baba and the Forty Thieves (AFT) optimization algorithm and support vector machine regression (RSVM) to enhance the accuracy and efficiency of distribution network reliability assessments. The proposed method combines the developed IAFT-RSVM model with a time-series production simulation using the Monte Carlo method for distribution network reliability assessment. This approach improves the convergence speed of the optimization algorithm to obtain more accurate RSVM model parameters. Numerical simulation results demonstrate that the reliability assessment accuracy of the proposed IAFT-RSVM is higher than that of AFT-RSVM, indicating the superior applicability of the IAFT-RSVM method for grid reliability assessments.

AFT algorithm, Monte Carlo method, reliability assessment, support vector machine regression, time-series production simulation.