Research On Raw Coal Production Cost Based On PSO-SVR

Raw coal production cost forecast is the basis for coal cost accounting, costdecision-making and workout cost analysis, is an important measure to reduceproduction costs, and is also the principal means to improve the economic benefit andenhance enterprise competitive ability. At present, Coal Enterprises in our countrycalculates the raw coal cost on the basis of enterprise accounting standards and financialgeneral rule issued by the Ministry of Finance in2006, namely the cost spent in theproductive process. However, considering the factors affecting coal production cost isvery complex, calculating the raw coal cost with the help of the production process costforecast is more scientific.This paper makes a deep research on the factors affecting coal production costfrom the three respects of space, time and policy, the main factors affecting the raw coalproduction cost is summarized. Collecting data According to the main influence factors,When support vector machines are used for regression prediction, it is not only to solvethe global optimal solution in the case of small-size training samples, but also to solvedimension problems. However, selection of element type’s defects exists in supportvector machines. This paper analyzes in depth the learning rule and training of supportvector machines, another more advanced style of particle swarm optimization algorithmis introduced to optimize the parameters of support vector machines, the morereasonable index system of raw coal production cost forecast is established.Unascertained measure model is introduced to quantify the quantitative indexes in theinfluencing factors. Meanwhile, Considering there have not been unified theoreticalprinciple on the selection of parameter and kernel function for support vector machines,three kinds of method of parameters optimization such as cross validation, particleswarm optimization method and genetic etc. and four kinds of kernel function are addedto SVM regression model separately, comparing the optimum results, the best SVMregression model which is suitable for the raw coal production cost forecast is obtained,that is PSO-SVR prediction model. Finally, producing data that is collected by Liuhecoal mine is applied to make predicted analysis scientifically and the result is perfect,which can provide a reliable reference for the final control of the coal production cost.