Research On Demand Forecasting Of Coal Port Spare Parts Based On LS-SVM

Along with the quick development of the construction of coal port, port enterprisesare facing increasing competitive situation. Port equipment is the key to the portoperations. In order to avoid downtime of the equipment due to spare parts shortage, portenterprises tend to store large amounts of spare parts, which take up a lot of inventorycapital, but even so, spare parts shortage phenomenon still occurs frequently. Therefore,improving the ability to forecast spare parts demand is very significant to the coal portoperation as well as the economic benefit.However, the demand forecasting of port spare parts has the characteristics ofmultiple influential factors, nonlinear and less historical data, which brings great difficultyto forecast. Traditional statistical prediction method is based on asymptotic theories, whichis suitable to deal with situations where the amount of samples is tremendous and eveninfinite. But it is difficult to solve the problem of small sample. The paper chooses LeastSquares Support Vector Machines (LS-SVM) to forecast, and through the case of a certaintype of spare parts of Qin Huangdao port, the viability and effectiveness of the model isverified. The main contents of this paper are as follows:First, the thesis introduces the research status of the spare parts demand predictionand least squares support vector machine algorithm, and the relative theory and method ofleast squares support vector machine.Second, some influential factors which have an effect on spare parts consumption aredeeply analyzed in this thesis. And Analytic Hierarchy Process (AHP) is used to sieve outthe more influential factors as the inputs of LS-SVM model.Third, aiming at the parameter optimization problem in LS-SVM, Adaptive MutationParticle Swarm Optimization (AMPSO) algorithm was adopted to optimize the parameter.And the establishment of the prediction model based on AMPSO-LSSVM is explained indetail.Finally, the paper chooses the real data of spare parts in Qin Huangdao port to makecase analysis, and compares with other prediction models. Based on the proposed model, a new Forecasting Support System (FSS) is developed for port spare parts, which provides anew method for coal port spare parts forecasting.