Novel Version of Horse Herd Optimization for Enhancing Electric Load Forecasting Capabilities of Neural Networks

M Mishra and P Mahajan and R Garg, ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING, 50, 17537-17554 (2025).

DOI: 10.1007/s13369-025-09996-w

Developing steady electric load forecasting model can ensure the synchronization between supply and demand, marking the attainment of efficient power distribution system and securing economic benefits. The accurate electric load prediction with a conventional model is a challenging task due to the dynamic nature of electric load consumption. As such this research proposes a sequential model involving adequate pre-processing of dataset and the novel optimized load forecasting model which yields higher accuracy. The novel optimized load forecasting model has been developed by proposing the improvement in conventional horse herd optimization (HHO) algorithm. The development of improved HHO (iHHO) is based on the integration of conventional HHO with the prey tracking characteristic of the grey wolf optimization (GWO) algorithm. Consecutively, this novel iHHO has been used to enhance the forecasting capability of the artificial neural network (ANN) and cascade neural network (Cas-NN). Initially, important meteorological parameters have been extracted using XGBoost technique and the K-means clustering algorithm has been used to divide the dataset into clusters based on their similarity pattern. Additionally, to test the computational capability and robustness of the proposed algorithm, the non-clustered original dataset has also been analysed. The Diebold-Mariano test along with error matrices analysis has been used to highlight the best performing model (Cas-NN-iHHO model). The further benchmarking and validation of the proposed models have been done by making a comparison with various metaheuristic-based Cas-NN models and other recently develop electric load forecasting models present in the literature.

Return to Publications page