A combined system based on data preprocessing and optimization algorithm for electricity load forecasting

Creating steady models for predicting electricity load can enhance the equilibrium between power supply and demand, a critical factor in advancing precise distribution management and optimizing economic advantages at a granular level. Electricity load forecasting is a challenging research area, and the accuracy improvement of existing single-point load forecasting models is limited by the randomness and volatility of electricity load data. As such, this research introduces a combined system. Firstly, based on the optimized Variational Mode Decomposition method, the system utilizes the Tuna Optimization Algorithm to optimize two key parameters of VMD (the penalty factor α and the number of mode decomposition K) with the objective of minimizing the envelope entropy and obtaining smoother and more stable signals. Secondly, a combination model consisting of multiple single models is proposed, and the Chef-Based Optimization Algorithm is employed to search for the combination weights that minimize the prediction errors, thereby enhancing the precision and consistency of the predictive model. To validate the superiority of the combined system, experiments are conducted using electricity load data from Queensland, Australia, with a time interval of 5 min. The numerical findings demonstrate that the system not only exhibits a substantial performance advantage over the single model in various assessment criteria like mean absolute error and root mean square error but also confirm the efficacy of the proposed method.