Research on meta heuristic algorithms for optimal reactive power dispatch in power system

Abstract

Meta-heuristic algorithm plays a vital role in the Optimal Reactive Power Dispatch (ORPD) of the complex power system planning and operation. Most of the ORPD problems considered metaheuristic algorithms for their minimum number of parameters and operators. The reactive power generation changes for every load variation in the power system operation and inturn tends to vary the load voltage. By proper management of reactive power, the voltage profile will be maintained. The objectives of ORPD is to minimize the real power loss and voltage deviation in the transmission network. The objective is subjected to equality and inequality constraints. The minimization of real power loss and voltage deviation are achieved through control variables which consists of generator voltage magnitude, transformer tap settings and reactive power of shunt capacitors. Therefore, proper handling of control variables leads to minimization of real power loss and voltage deviation in the transmission lines easily. In this research work, three meta-heuristic algorithms namely Self-Balanced Differential Evolution (SBDE) algorithm, hybrid Water Cycle Moth-Flame Optimization (hWCMFO) algorithm and hybrid modified Path-Finder Differential Evolution (hmPFDE) algorithm, are proposed to solve ORPD problem and tested on the IEEE 30-bus system, IEEE 57-bus system and IEEE-118 bus systems. The simulation results are tested on two different cases: Case I-100 % (normal) loading condition and Case II-130 % (stressed) loading condition. Being first, the evolutionary based algorithm called Self-balanced differential evolution (SBDE) is proposed to solve ORPD problem. Differential Evolution (DE) is a population-based meta-heuristic technique that optimizes a problem based on evolutionary process. The primary thought behind DE was to create new offspring. The performance of DE gives the global search ability but it is sensitive to the mutation function, crossover function and population size. The difficulty of DE is to set the parameter values according to