Design and development of cluster algorithms for power system problems

Abstract

This thesis presents design and development of genetic based cluster algorithm application to mitigate the complex power system problem namely, Optimal Power Flow (OPF) problem. Two individual objective functions are chosen: 1) minimization of Fuel Cost and 2) minimization of Power Loss. The proposed algorithm is the application of a multi-objective genetic algorithm (MOGA), using the combination of High Density cluster and continuous genetic algorithm. The OPF is modeled as a nonlinear, non-convex and large scale constrained problem with continuous variables. The algorithm uses a local search method for the search of Global optimum solution. Binary coded Genetic algorithm is replaced with continuous genetic algorithm that uses real values of generation instead of binary coded data. An attempt is made to reduce the length. Inspired by the results of Genetic Algorithm (GA) method and to overcome the general difficulties in GA approach, a novel method is proposed in this work.. The method uses high density cluster DBSCAN and Continuous GA algorithms. The new technique for the solution of OPF based on Genetic search from a High Density Cluster named in short form as ?GSHDC? is proposed in this thesis The objective of GSHDC is to retain advantages of Mathematical Programming techniques and to encounter the difficulties of evolutionary methods like GA and PSO Methods. The GSHDC has mainly four stages.Stage-1: In the first stage a suboptimal solution for OPF problem is obtained by any of the following local search methods 1) Modified Penalty Factor Method 2) Primal-Dual Interior Point method and 3) Particle Swarm Optimization Method that considers Lagrange multipliers, equality constraints, transmission loss B-Coefficients and penalty factors. Owing to the limitations of the methods, this solution is taken only as suboptimal or local optimal. Because of this reason, this OPF solution cannot be taken as a global one.