Particles Swarm Optimization Algorithm And Its Application In Spare Part Management

Operation management with the low cost and high efficiency has become a theme of the modern enterprise along with the increasingly fierce market competition. So, spare parts management becomes more and more important. It impacts the enterprise's operation costs directly and becomes a key part for the enterprises to improve their competitiveness. Based on the spare parts management of a nuclear power station, the thesis discussed the key points of the spare parts management and the suppliers'selection & order quantity allocation problem.First, the limitations of the methods of the classification of spare parts and the suppliers'selection are concluded. The characteristics and application fields of the intelligence optimization algorithm (Particles Swarm Optimization, PSO) algorithm are systematically analyzed and the applications & the improvement strategies of the PSO algorithm in this thesis are discussed. Secondly, an artificial neural network (ANN) model for ABC classification is proposed. A hybrid learning algorithm for ANN train is designed combining PSO with back-propagation (BP) algorithm based on the analysis of the limitations of the widely-used ANN train algorithm. The proposed hybrid algorithm can make use of not only strong global searching ability of the PSO algorithm, but also strong local searching ability of the BP algorithm. The result gained by the proposed ANN classification model and the conventional methods are compared using the data collected from a certain nuclear power Station. Finally, the vendor selection and the order quantity allocation problems are studied. An uncertainty, multi-objective and stochastic constraint planning model is constructed with the objective functions under three criteria of quality, cost, delivery and the other goals as constraint conditions with the stochastic demand. With the weighted way and the penalty function, stochastic model with uncertainty and multi-objective is converted into a single target optimization model. At last, a new PSO algorithm with inertia and contraction factors is designed to solve the proposed model and comparative analysis with commonly-used genetic algorithm is given to verify the feasibility of the PSO applied to such issues.