Optimized PID controller and model order reduction of reheated turbine for load frequency control using teaching learning-based optimization.

Load frequency control (LFC) systems in power grids face challenges in maintaining stability while managing computational complexity. This research presents an optimized approach combining model order reduction techniques with Teaching Learning-Based Optimization (TLBO) for PID controller tuning in single-area LFC systems. Three reduction methods-Routh Approximation, Balanced Truncation, and Hankel Norm Approximation-were implemented to reduce system order from 4th to 2nd order, achieving a 47.3% reduction in computational time. The TLBO-optimized PID controller was compared with conventional tuning methods (Ziegler-Nichols, AMIGO, S-IMC, and CHR), demonstrating superior performance with a 38.2% decrease in settling time and 42.7% reduction in peak overshoot. The Routh Approximation method exhibited optimal performance with minimum settling time (2.8s) and peak overshoot (8.4%). Sensitivity analysis revealed stable system behavior with phase margin maintained at 84.25 degrees across parameter variations. The proposed approach achieved a 56.8% reduction in Integral Square Error compared to conventional methods, establishing its effectiveness for modern power grid applications. This research provides a robust framework for implementing efficient load frequency control in power systems while maintaining system stability and performance.