Optimal distribution system reconfiguration incorporating dg and variable load profile using artificial neural network / Hesham Hanie Youssef

Hesham Hanie, Youssef (2020) Optimal distribution system reconfiguration incorporating dg and variable load profile using artificial neural network / Hesham Hanie Youssef. Masters thesis, Universiti Malaya.

	PDF (The Candidate’s Agreement) Restricted to Repository staff only Download (729Kb)
Preview	PDF (Thesis M.A) Download (1800Kb) | Preview

Abstract

Optimal network reconfiguration is a common method used in distribution systems to ensure minimum power losses are always attained. This is very important task for achieving cost effective operation. Due to varying load demands, conventional network reconfiguration techniques have to be repeated whenever system loading changes to find a new configuration that has minimum power losses. This task is time consuming and ineffective approach for a real time application. Therefore, this research proposes an Artificial Neural Network (ANN) technique for optimal distribution network reconfiguration to overcome long processing time, mainly in load variation case. The proposed method involves; (1) Implement optimal network reconfiguration with variable load profile and DG generation using meta-heuristic techniques for ANN modelling (2) Designing an ANN model for optimal network reconfiguration (3) Train the proposed ANN model on the generated data using different split ratios for optimal network reconfiguration. The applied meta-heuristic techniques in this work are Evolutionary programming (EP) and Particle swarm optimization (PSO). To evaluate the performance of the proposed ANN method, simulation conducted on MATLAB were conducted on IEEE 16-bus, IEEE 33-bus and IEEE 69-bus system. The proposed network reconfiguration based on ANN significantly reduces the computational time to find the optimal solution while avoiding additional calculations. The results show that the proposed ANN technique is more than 90% faster than the conventional methods for varying load profile.

Actions (For repository staff only : Login required)