A Discrete Chaotic Jaya algorithm for optimal preventive maintenance scheduling of power systems generators

The main role of the optimal Generator Maintenance Scheduling (GMS) problem in power systems is to develop an optimal preventive maintenance scheduling of the generation part units. An optimal GMS provides power systems with higher operational reliability, extends the generators lifetime and reduces the cost of generators maintenance. The GMS problem is formulated as an optimisation problem. This problem should satisfy both load’s power demand and workforce constraints with ensuring the reliability of power systems at economical operation cost. The GMS problem has been studied for many years when exact mathematical methods have been used in the past to reach exact solutions for small-scale problems. However, these conventional mathematical approaches have many limitations and they suffer from unreasonable computational efforts as system dimension increases. Traditional approximate methods have been adopted to overcome the limitations of exact methods for medium-scale power systems. However, they provide approximate solutions and they require a large computational effort for wide-area systems of big dimensions. Recently, modern methods based on metaheuristics optimisation have taken a long part to solve the GMS problem and to overcome the limitations of approximate methods. In this paper, a proposed Discrete Chaotic Jaya Optimisation (DCJO) algorithm is employed to perform the preventive maintenance scheduling of electric power systems generators. The proposed algorithm is based on a cooperation between the discrete Jaya optimisation algorithm and a proposed move rule based on Chaotic Local Search (CLS) technique to improve both exploration and exploitation phases. The GMS problem is modelled based on the reliability criterion of an objective function of a sum of the squares of the reserves of generation. The optimisation process is performed through minimising an evaluation function of a weighted sum of the objective function and the penalty function for violations of the constraints. The proposed approach has been tested in a 21-unit test system over a planned horizon of 52 weeks in which the peak load is 4739 MW and the maximum generation is 5688 MW and there is a total number of 35 workforce available per week to perform the maintenance tasks. The proposed method has been compared through several statistical tests with recent algorithms of the related works. The obtained results show the effectiveness of the proposed algorithm for solving the GMS proble