Adaptive Genetic Algorithm for Hydro-thermal Unit Commitment Considering the Security Constraints

Abstract

This paper proposes a new approach to two efficient metaheuristic algorithms, combined with quadratic programming, to solve the nonlinear optimization problem Unit Commitment (UC) in a complex hydro-thermal power system (HTUC). The main purpose is to minimize the total production costs (which are a very non-linear and non-convex problem), while satisfying the many hydro-thermal constraints. Such constraints, together with the nonlinear non-convex and mixed-integer criterion function, make the search space extremely complex.

To solve such a complicated system, the paper proposes a hybridization of a developed binary coded genetic algorithm, which integrates quadratic programming, and particle swarm optimization (PSO) algorithm. PSO is applied to final i.e. main Economic Load Dispatch (ELD) based on the obtained optimal binary combination from the genetic algorithm.

Given the complexity of the problem (the many constraints that are strongly correlated with the control variables), the initialization is not done randomly, ie. by standard uniform distribution but by generating a population of chromosomes (a combination of 0 and 1 at random) that meet the standard condition for UC.

Given the change in genes resulting from the selection and mutation operators, it is possible that many chromosomes do not meet the UC condition. Therefore, such a standard methodology can be a serious constraint on the diversity of the population, as many chromosomes will be discarded, which may lead to premature convergence of the algorithm or stuck in a local optimum. In this paper, a new approach is proposed, with the implementation of a repair mechanism based on the priority list method (which is formed at the very beginning of the algorithm), based on the principle of economics of thermal power plants.

The main purpose of this paper is to propose a new approach to the HTUC problem, which simultaneously includes the key constraints of thermal and hydroelectric power plants and system constraints (compared to other proposed methods), in order to obtain a realistic and physically acceptable solution. On the other hand, a procedure for step-by-step constraints consideration is proposed, in order to increase robustness and convergence, ie to prevent premature convergence of the genetic algorithm and obtain the global optimal solution.

The entire algorithm was developed in MATLAB, tested, and then applied to the IEEE 30 BUS test system. Experimental results show better performance of the proposed algorithm compared to recently published algorithms, in terms of convergence, constraint handling, and better solution quality.