This project is concerned with the application of a coevolutionary augmented Lagrangian method (MCE) for solving an economic dispatch generation problem. The basic objective of this research is analyzing and comparing the results obtained by the MCE method with the ones obtained by conventional genetic algorithm methods and primal dual interior point methods. Aspects such as computational effort, time for computational convergence (number of generations/iterations), and also optimal values and energy prices will be compared for all such methods. The application of the MCE method is supposed to obtain solutions for the studied dispatch problem involving non-differentiable and/or non continuous objective functions (such as valve point loading for thermal units) and also to calculate duality gaps related to primal and dual problems, which may provide subsidies for analysis of the solution optimality (which is generally not possible for conventional genetic algorithm approaches).
News published in Agência FAPESP Newsletter about the scholarship: