REACTIVE POWER COMPENSATION AND LOAD FREQUENCY CONTROL OF GRID CONNECTED MICRO GRID SYSTEM USING STATIC VAR COMPENSATOR (CASE STUDY: DILLA SUBSTATION

Abstract

n the modern power system, reactive power compensation and load frequency control are two of the main issues. The aim of this thesis is to study reactive power compensation and load frequency control of the grid-connected micro grid system (GCMG) under variable load .In this work, an Artificial Neural Network based Static VAR compensator (ANN-SVC) and Load Frequency Controller (LFC) will be proposed for reactive power compensation and load frequency control of the grid-connected micro grid system, The artificial neural network (ANN) is used to control the the SVC gete signal for proper controlling of the thyristor valve in the circulated system current management. Power system Planning and management provide the strategy for reactive power compensation and load frequency disturbance control. The ANN-controlled SVC is used for managing and compensating of reactive power in the system and LFCis control the system load frequency disturbance and the LFC controlling loop component is control and managed by ANN signal for proper controlling of the load frequency error. The real and reactive powers before applying the SVC at the peak load conditions are 40.56MW and 27.16MVAr, respectively, before reactive power compensation. During contingency conditions, the real and reactive powers are 36.03MW and 30.42MVAr, respectively, and the frequency is highly disturbed for a fraction of a second, but after compensation, the real power is improved to 50.54MW, while the reactive power is reduced to 17.34MVAr. Therefore, real power is improved by 52.18% while reactive power is compensated by 87.02%. The steady-state load frequency error is reduced by 95%, while the system power factor is improved by 94%. A detailed comparative analysis of the proposed ANN-based SVC with the fuzzy-based and multi-objective fire fly (MOFA) algorithm SVC is presented, which shows that the ANN-based SVC has better performance, fast and accuracy, in case of the above justification ANN is more prifereble than MOFA and FUZZY algorithem. The general GCMG structure with the compensation and controlling process was designed by ETAP and MATLAB/Simulin