Multi-objective optimization of generation maintenance scheduling

Abstract

For the generation maintenance scheduling (GMS) problem, a producer hopes to maximize its profit while ISO is to guarantee the system reliability. Thus, the GMS is a multi-objective optimization problem. In the GMS, there are large numbers of both continuous and integer variables, which complicates the resolving of the GMS. This paper proposes a new GMS model, which is suitable to be solved by the nondominated sorting genetic algorithm-II (NSGA-II). In the GMS model, the maintenance status of a generator is encoded into an integer variable and both the online status and the start-up status are represented by the generation variables. The GMS model on the IEEE reliability test system is solved by NSGA-II with a set of Pareto-optimal solutions obtained. The simulation results show that the GMS can be efficiently solved by NSGA-II. The simulation results also show that one producer's profit conflicts with another one's, and that the reliability objective is independent of the other objectives.

Comparison of Multi-objective Optimization Methods for Generator Maintenance Scheduling

“…The study of [10] consider optimization method to obtain solution for GMS with three types of objective functions; comprising producer profit, system reliability which is defined as the minimization of the standard deviation of the reliability index, and total generation cost. The study of [20] considers optimization method to obtain solution to GMS with two objectives i.e., profit minimization and reliability maximization. The system reliability objective is defined to be the average value of the reliability index.…”

“…Aggregation methods are reasonably straightforward, and no modifications are required for the basic algorithm [16]. The most common intelligent methods that are used in the domain of multi-objective GMS are the Non-dominated Sorting Genetic Algorithm II (NSGAII), Strength Pareto Evolutionary Algorithm2 (SPEA2), Pareto Ant Colony Optimization, Multiobjective Simulated Annealing (MOSA), and Multiobjective Particle Swarm Optimization (MOPSO) [9,[18][19][20][21][22][23][24][25][26].…”

Generation maintenance scheduling based on multiple objectives and their relationship analysis

“…In this paper, the generation variable P G ij (t, s) is encoded into the real number (Chen et al, 2014).…”

“…Besides, the GMS is also suitable to be solved by other MOEAs, e.g., NSGA-II (Chen et al, 2014) and MOPSO.…”

“…Another disadvantage of the mixed integer programming based methods is that the computational burden increases exponentially as the dimension of variables increases. The presented GMS model employs an encoding technique for the variables, which makes the model suitable to be solved by MOEAs which use real number encoding, including the NSGA-II (Chen et al, 2014), multi-objective particle swarm optimization (MOPSO) (Tripathi et al, 2007), and group search optimizer with multiple producers (GSOMP) (Guo et al, 2012). In our previous work (He et al, 2009;Guo et al, 2012), it has been proved that GSOMP has better performance than NSGA-II and MOPSO, especially on complex high-dimensional multi-objective optimization problems.…”