IJCSNS International Journal of Computer Science and Network Security, VOL.8 No.6, June 2008 61 Optimum Scheduling of Generators Using Genetic Algorithm G K Joshi*, Sanjay Mathur** and Sumit Mathur*** * Principal and Professor, Institute of Engg. & Technology, Alwar (Raj.) – 301030, India ** Senior Faculty Marwar Engg. College & Research Centre, Jodhpur (Raj.), 342001, India ** * Senior Faculty Marwar Engg. College & Research Centre, Jodhpur (Raj.), 342001, India Summary Variation in load demand does not allow a fixed number of generators working in parallel to share the load in proportion to their capacity and therefore lead to an uneconomical operating cost. Presently the economic load dispatch in a group of generators is decided by the criteria of constant derivative state achieved among all the generators in the group of generators. Thus with n-generators operating together the partial derivatives of fuel cost F to the capacity of generator P, should always prevail with equality for each of the generators. Since the derivative approach for economic load dispatch suffers from the draw back of local maxima, the present work has implemented genetic algorithm, which yields global maxima. Also it has been found that the scheduling of generators with regard to its operation on maxima and minima of their capacity in a particular hour over 24 hours is decided by genetic algorithm GA. It is proved that the GA approach for optimum scheduling of generators, Yields a substantial saving/ economy in operating cost over the derivative approach. Software has been developed to implement GA for selection of best string. Here the best string refers to the optimum scheduling of generators in a group. Key words: Genetic algorithm, Reproduction, crossover, Mutation, chromosomes, total objective function (TOF), Generations. 1.0. Introduction The modern power system needs to operate its generating units in parallel to meet the specific load demand under variable operating conditions. The maintenance engineer is always under a pressure to ensure economic dispatch of load. The engineer has always to plan about the generators that will remain in operation at a given time and under given load conditions. The economic dispatch involves two separate steps namely the unit commitment and the online economic dispatch. By the term Unit Commitment [3] its meant that the load will be supplying by the generator at minimum cost and besides it the generator will keep a specified margin known as operating reserve. By the online economic dispatch provides for proper load allocation to different generating units in such a manner as to minimize the total cost of supplying the load. The operating cost is mainly the cost of fuel, which is a non-linear function of plant generation given by Ci = αi + βi Pi + γi Pi 2 Since the cost of generation is a nonlinear function of the power generated by any generating unit. The use of mathematical model for solving optimal operating conditions doesn’t yield a global optimal conditions, rather its provides local optimal conditions, in order to overcome this difficulty the most versatile which was noticed after study the GENETIC ALGORITHM. 1.1 Why Genetic Algorithm? Basically Genetic Algorithm popularly known as GAs, are search algorithms based on the mechanics of natural selections and natural genetics. GA combined survival of the fittest string among string structures with structure randomize information exchange to form a search algorithm. Genetic algorithm has been developed by John Holland at university of Michigan. It was aimed to abstract and give a rigorous explanation to adaptive processes in natural systems. He also aimed to design artificial system software that strictly follows the basic mechanism of natural system. This powerful tool is going to prove wonders in the field of search, optimization and machine learning. The objective of genetic algorithm is to ensure robustness [2] of the best fit generation. It strikes a balance between two or more competing features for example effectiveness and power. Consider a function that varies in a pattern as shown in figure. f(x) X Manuscript received June 5, 2008. Manuscript revised June 20, 2008.