Maintenance scheduling of nuclear components under reliability constraints using adaptive parallel particle swarm optimization

Abstract

Preventive maintenance is a critical element of maintenance policies in a wide range of industries, including the power sector. To achieve reasonable and effective maintenance of nuclear power plants (NPPs), proper aging management is critical and should be optimized from both safety and economic perspectives. Thus, in this paper, we propose a maintenance-scheduling model based on an adaptive parallel particle swarm optimization (PSO) to minimize the total number of maintenance activities over the lifetime of an NPP while ensuring the reliability of safety-critical functions. The proposed model recognizes that effective maintenance activities differ depending on the cause of the latent failure. In addition, the applied PSO algorithm, which is based on the dynamic exchange of hyperparameters between adjacent swarms, allows us to optimize inertia factor and learning factors adaptively during the solution search process. The proposed model is verified by applying it to a representative case in which the best maintenance schedules for the components constituting a water injection function are produced.