Optimal under-frequency load shedding (UFLS) scheme using evolutionary programming technique

• Main Author: Lu, Michelle
• Format: Thesis
• Language: English
• Published: Universiti Malaysia Sarawak, (UNIMAS) 2013
• Subjects: T Technology (General)
• Online Access: http://ir.unimas.my/id/eprint/8291/3/Optimal%20Under-Frequency%20Load%20Shedding%20%28UFLS%29%20Scheme%20Using%20Evolutionary%20Programming%20Technique.pdf; http://ir.unimas.my/id/eprint/8291/

In literature, gradient-based nonlinear programming technique has been used to design an optimal Under-frequency Load Shedding (UFLS) scheme. However, the objective function of the System Frequency Response (SFR) - UFLS model was piecewise discontinuous and obtaining optimal solutions with this method had posed some difficulties. A promising solution approach could be the use of Evolutionary Programming (EP) which has recently being increasingly applied to constrained nonlinear optimization in power system problems. In this thesis, the analytic SFR - UFLS model has been used to compute the system and UFLS performance indicators from derived closed-form expressions of the load frequency response. Simulation studies have been done based on third-order and fourth order SFR-UFLS model to represent Kuching Power Island and the Sarawak power system respectively. The reliability, accuracy and appropriateness of the SFR-UFLS models in representing the actual system scenarios have also been verified with a full-scale power system simulator called the Power System Simulator for Engineering (PSS/E) software. For both the third-order and fourth-order model, simulation results have shown that a fixed UFLS scheme has better performance than an adaptive UFLS scheme with respect to the objective function. In the effort to further optimize the parameter settings for the fixed UFLS scheme, various static and dynamic adaptation methods have been analyzed. For the thirdorder model, static adaptation by replacing elitist selection with direct selection has produced the best results with an objective function of 0.4322. For the fourth-order SFR-UFLS model, static adaptation by adding cloning function has given the best optimized results with an objective function of 0.2209. The main contribution of this thesis is in the incorporation in the heuristic research field, of novel optimization model based in EP with capacity to provide optimal UFLS settings for a power system.