Unit commitment by artificial intelligence techniques

The present work deals with thermal generation scheduling, which could be considered the major part of the overall scheduling problem of hydrothermal power systems. The scheduling problem of thermal generating units can be considered as two linked optimization problems. It comprises the solution of both the Unit Commitment Problem (UCP) and the Economic Dispatch Problem (EDP). The former is a combinatorial optimization problem with very hard constraints, while the later is a nonlinear programming problem.; The growing interest in the application of Artificial Intelligence (AI) techniques to power system engineering has introduced the potentials of using this state-of-the-art technology in the thermal generation scheduling of electric power systems. AI techniques, unlike strict mathematical methods, have the apparent ability to adapt to nonlinearities and discontinuities commonly found in power systems. The best known algorithms in this class include evolution programming, genetic algorithms, simulated annealing, tabu search, and neural networks.; In the present work, seven different AI-based algorithms have been developed to solve the UCP. Two of these algorithms namely, simulated annealing and genetic algorithms, are implemented in a novel way. The other five proposed algorithms are applied for the first time to solve the UCP. The algorithms are a Simple Tabu Search Algorithm (STSA), an Advanced Tabu Search (ATSA), a hybrid of Simulated annealing and Tabu search algorithms (ST), a hybrid of Genetic and Tabu search algorithms (GT), and a hybrid of Genetic, Simulated annealing, and Tabu search algorithms (GST).; As a first step to solve the UCP, some modifications to the existing problem formulation have been made to render the formulation more generalized. An augmented model including all the problem constraints is presented.; A major step in the course of solving the UCP, is the solution of the EDP. In this regard, an efficient and fast nonlinear programming routine is implemented and tested. The implemented routine is based on a linear complementary algorithm for solving the quadratic programming problems as a linear program in a tableau form. Comparing the results of our proposed routine, it is found that the results obtained are more accurate than that obtained using an IMSL quadratic programming routine. The application of this routine to the EDP is original.; The corner stone in solving the combinatorial optimization problems is to come up with good rules for finding randomly feasible trial solutions from an existing feasible solution, in an efficient way. Because of the constraints in the UCP this is not a simple matter. The most difficult constraints to satisfy are at the minimum up/down times. A major contribution of this work is the implementation of new rules to get randomly feasible solutions faster.; All the proposed algorithms have been tested on several practical systems reported in the literature, with different complexities. The numerical results obtained by the proposed algorithms are superior to the results reported in the literature.