A Hybrid Group Method of Data Handling–Modified Particle Swarm Optimization Algorithm‐Local Outlier Factor for Short‐Term Load Forecasting in Microgrids

Precision tools are required because load forecasting is essential for power system planning and optimal performance. Data machine learning techniques and data mining methods are essential instruments for evaluating data and uncovering concealed information in data. In this study, a novel approach to estimating the grid‐connected micro grid load is put forth. The group method of data handling (GMDH), on which the successful approach is based, was used to estimate the load using an adaptive learning network. Using the technique of modified particle swarm optimization (MPSO), the GMDH network was developed. The major goal of providing combination techniques is to concurrently leverage single‐model benefits in structuring complex systems while also overcoming single‐model constraints. The MPSO algorithm is used to improve GMDH polynomial coefficients. In this paper, GMDH‐MPSO combines with the local outlier factor (LOF) to account for potential outliers. The outcomes compare favourably with those of standard GMDH, the autoregressive integrated moving average (ARIMA) approach, and GMDH taught using PSO, demonstrating the GMDH‐MPSO‐LOF method's tolerable efficiency. The GMDH‐MPSO‐LOF model with a correlation coefficient (R) value of 0.9968, mean square error (MSE) value of 0.1156, and root mean square error (RMSE) value of 0.3532 is competitive with similar algorithms.