Appling modified genetics algorithm method by combined method in optimization of river engineering problems

In most operational optimization methods in river engineering, the optimization problem is simplified through assumptions, discretization or henristic rules. Such simplification makes it easier for specific optimization methods to determine the optimal solution, but introduces bias into the solution by excluding a large number of potentially good Solutions. Genetic algorithms (GA.s) do not require such simplification measures, giving them a significant advantage in finding a near global optimal solution over most other optimization methods.A hybrid optimization strategy was developed by combining a genetic algorithm search with a hillclimber search method. GAs have very good initial convergence rates, but are less efficient once the GA has found a near-optimal solution. Local search methods, on the other hand, are good at converging on the local optimum from a nearby starting point, but are not able to jump to other possibly better, areas of solution space. By combining the GA strategy with a local search strategy, the advantages of both methods are exploited to produce an optimization method which is both reliable and fast.The ideal optimization method is one which is both reliable in finding a near-global optimal solution, and fast in converging on that solution. However, these goals are often in conflict with each other, forcing a trade-off in selecting the parameters for the optimization method. When applied to water distribution systems, for instance, in an emergency situation, it is often more important to find a good solution speedily than it is to find the global optimum solution. For this reason the hybrid method was developed with the main emphasis on convergence speed rather than reliability.

کلیدواژه ها:

Genetic Algorithms ، Hill climber strategies