International Journal of Computer Applications (0975 – 8887) Volume 20– No.2, April 2011 7 Environmental Economic Dispatch Optimization using a Modified Genetic Algorithm Simona Dinu Sr. Lecturer Constanta Maritime University Mircea cel Batran Constanta ROMANIA Ioan Odagescu Professor The Bucharest Academy of Economic Studies ROMANIA Maria Moise Professor Romanian American University, Bucharest ROMANIA ABSTRACT This paper aims to develop a new Genetic Algorithm based approach to solve the Combined Environmental Economic Power Dispatch Problem. The essential features of our proposed algorithm include a diploid based complex-encoding with meiosis specific attributes and new mutation operators that performs global search during the initial generations and local search in the later generations. Using the parallel searching mechanism and the new defined mutation operators, the local searching ability of the algorithm is improved, as well as the algorithm’s efficiency. Results of comparative tests on a sample power system are presented, showing the better computation efficiency and convergence property of the proposed methodology. General Terms Dispatch strategies. Evolutionary Computation. Energy Systems. Keywords Multiobjective optimization. Environmental Economic Dispatch, Genetic Algorithms. Diploidy. 1. INTRODUCTION The dispatch strategy for an energy system is a primary objective in the operation of power systems. It is a computational process based on a set of rules controlling the interaction among various system components in order to operate the power system in an economic and efficient manner. In accordance with the objectives, the dispatch strategies can be divided in: - economic dispatch, attempting to schedule the committed generating unit outputs to meet the load demand at minimum production and transmission (operating) cost; - environmental dispatch, attempting to reduce the environmental impact of power generation ; - economic/environmental dispatch, attempting to achieve both objectives (minimizing the operating fuel cost and emission cost) in a single dispatch. The classical Economic Dispatch (ED) is to allocate the total required generation among the available generating units in order to minimize the total generation cost while simultaneously satisfying all equality and inequality constraints. To balance the load variations, the power output of generators has to be adapted. This leads to minimizing system losses at all time and decrease the operational costs. Thus, it is a critical task of electric utilities to deliver power as demanded in order to maintain the reliability and continuity of electricity supply. The literature of the ED problems and its different numerical solution methods are investigated in [1] [2]. Fossil-fueled electric power plants produce harmful emission such as Sulfur Oxides, Nitrogen Oxides and Carbon Dioxide. Recently, in order to meet severe environmental standards imposed by legislation, pollution minimization has become another important operational goal. Thus, improvements in scheduling the unit outputs must result in both monetary profits and reduced emissions of gaseous pollution. A survey of the commonly environmental dispatch algorithms has been given in [3] and [1]. The Combined Environmental Economic Dispatch (CEED) problem is a bicriterial optimization problem with two conflicting objective functions: operating costs and environmental impact of emissions. Due to the contrasting/conflicting goals and non-commensurable natures of fuel cost and emission minimization objectives, conventional approach which optimizes the integrated two objective functions seems not appropriate for this class of multiobjective optimization problems [4]. Therefore, conventional optimization methods based on derivatives and gradients are not suitable for this nonlinear and multimodal optimization problem. Not longer after considering the environmental feature in the ED problem, different solution methods have progressively been reported in the literature concerning the CEED problem. Researchers’ methods considered emissions either in the objective function (by converting into a single-objective problem and assigning relative weights to each objective [5], [6]) or treated them as additional operational constraints that must be satisfied [7], [8]. With increasing size and complexity of the problem, many researchers have proposed the use of heuristic optimization approaches. Evolutionary computation techniques such as genetic algorithms (GA), evolutionary strategies (ES), evolutionary programming (EP), genetic programming (GP) and related fields such as swarm intelligence (Ant Colony Optimization and Particle Swarm Optimization) and other