The Research Of Evaluate Model Of The Industry Cluster Risk

Nearly twenty years, industry cluster has become a regional economy, industries, the field of economics to study the subject.。Industry cluster in the region and innovation ability to raise the competitive advantage and Also exposed to many factors of instability, the development of industrial clusters are faced with great risks. Economists risks of industrial clusters to carry out a lot of research, Especially the cluster risks, risks of its origins and risk prevention and resolution of measures to further explore. These studies provide risk analysis for the industry cluster multi-level research perspective and research methods。Most of these studies are qualitative, difficult to capacity at risk of cluster size, the strength to resist risks, an accurate evaluation.Previous studies of this subject in the reference, based on the major risks through the analysis of industrial clusters, according to a comprehensive, simple, easy to operate the principle of reciprocity, to extract the cluster of risk evaluation index system; and gray theory and BP neural network, the industry Quantitative risk assessment model clusterGrey prediction model GM (1,1) form is more simple, less data can be used to establish prediction model. Modeling and calculation is very convenient, but the gray model forecast volatile effects of the physical system is not satisfactory, and the prediction accuracy of the increase gradually decreased with time. In this thesis, the above GM (1,1) model of the limitations of using BP neural network is improved.Since the forecasting performance of nonlinear systems is obvious, but the time needed to train the neural network to a large number of sample data as the basis. Therefore, BP neural network according to its characteristics, in this paper a simple algorithm based on gray model, modeling the required data with less features and BP neural network has good performance characteristics of nonlinear prediction, the above two models of organic combine the cluster proposed for risk prediction model. Finally, specific high-tech zones in Changsha machinery industry cluster instance, the model established in this paper conducted a risk prediction. Predicted results show that the combined model prediction models in two separate full play the advantages of forecasting the same time weakening the monotony of their respective shortcomings of forecasting, and forecast results than a single prediction with higher accuracy, and more cluster of risk factors consistent with trends.