The Optimal Combination Forecasting Method Base On The Kuhn-tucker Conditions And It Apply In R&D's Input Of Funds Research

R&D's funding for academic research is not only a hot issue, but also is a measure of a country's international competitiveness, one of the key indicators. On the one hand, funding for R&D investment promote the rapid progress of science and technology, promote economic health, sustainable and stable develop; on the other hand, the adjustment of industrial structure promote its optimization and upgrading, enhance industry correlation, and promote industrial technology innovation and so on have made outstanding contributions. In view of the R&D funding shortage caused by inefficiencies or excess waste of resources put into such issues as how to use the model to determine the effective R&D funds into the reasonable rangeAnd basing on available data on R&D Funds for scientific prediction has become a major issue of this paper.In general, the current prediction methods are more mature for the individual prediction and the combination prediction method. Single forecasting method is a relatively simple prediction method, it is mainly through a single model specific data for the existing linear or nonlinear prediction, the method is simple, workable, intuitive and strong advantages, but there are also some limitations, such as the model prediction of high precision, and limiting to conditions caused by too much prediction have a wide gap and so on. The combination forecasting rule effectively overcome the disadvantages of individual forecasts, such as the single forecast their prediction accuracy is not high, and there is greater uncertainty in-the overall forecast, etc.Forecasting individual by a variety of forecasting methods as representing different fragments of information collection, through integration and decentralization of information overcome the above drawbacks. Therefore, the combination forecast is more precise, predictable stability and meet the future ability to predict environmental change better.In view of R&D expenses in the projections of input information is not complete under the complex economic system, combined forecast theory and modeling method has strong practicability and in a mix of different characteristics of combination forecasting model based on the selection of the final paper In Kuhn-Tucker conditions, the smallest error sum of squares principle, and can translate into non-linear programming model linear programming model of optimal combination forecast model. The optimal combination forecasting model to solve the general combined forecasting model in solving the combined weight of no analytical solution of finding the core issue.And this also through the use of qualitative and quantitative methods will be exponential smoothing forecasting method, the gray prediction method, BP neural network forecasting method of three individual forecasting methods and the combination of three analysis and forecasting software, and R&D funds to predict, and then the use of optimal combined forecasting model for its combination of weight distribution, weight distribution and thus overcome the subjectivity and arbitrariness, reducing forecast error, and the establish R&D funds into the system optimal combination forecast model. Finally, using the years R&D funding in Chongqing input data, for example, the establishment of the optimal combination forecasting model analysis and test R&D ChongqingR&D funds were predicted from the final results and these can be seen both for the short-term forecast and the long-term forecast, optimal combined forecasting model is better than the single prediction model, and from 2008-2012 Chongqing investment in R&D funding forecast result show that the optimal combination forecasting model has a higher stability and forecast accuracy.