A Hybrid PSO-GCRA Framework for Optimizing Control Systems Performance

Abstract

Optimization is essential for improving the performance of control systems, particularly in scenarios that involve complex, non-linear, and dynamic behaviors. This paper introduces a new hybrid optimization framework that merges Particle Swarm Optimization (PSO) with the Greater Cane Rat Algorithm (GCRA), which we call the PSO-GCRA framework. This hybrid approach takes advantage of PSO's global exploration capabilities and GCRA's local refinement strengths to overcome the shortcomings of each algorithm, such as premature convergence and ineffective local searches. We apply the proposed framework to a real-world load forecasting challenge using data from the Australian Energy Market Operator (AEMO). The PSO-GCRA framework functions in two sequential phases: first, PSO conducts a global search to explore the solution space, and then GCRA fine-tunes the solutions through mutation and crossover operations, ensuring convergence to high-quality optima. We evaluate the performance of this framework against benchmark methods, including EMD-SVR-PSO, FS-TSFE-CBSSO, VMD-FFT-IOSVR, and DCP-SVM-WO. Comprehensive experiments are carried out using metrics such as Mean Absolute Percentage Error (MAPE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and convergence rate.  The proposed PSO-GCRA framework achieves a MAPE of 2.05% and an RMSE of 3.91, outperforming benchmark methods, such as EMD-SVR-PSO (MAPE: 2.85%, RMSE: 4.49) and FS-TSFE-CBSSO (MAPE: 2.98%, RMSE: 4.69), in terms of accuracy, stability, and convergence efficiency. Comprehensive experiments were conducted using Australian Energy Market Operator (AEMO) data, with specific attention to normalization, parameter tuning, and iterative evaluations to ensure reliability and reproducibility.