The Methodology Of Innovation Development Evaluation Using Neuro-Fuzzy Modeling

In a globalized economy the tasks of the control, monitoring and evaluation of the level of innovative development, prediction of its dynamics are very important for both individual companies and territories on the whole. However, the implementation and evaluation of innovative economic activities take place under the conditions of considerable uncertainty, so the use of intelligent decision support systems based on mathematical and computer processing of uncertainty becomes a necessity. New modeling possibilities are provided by the combination of the fuzzy set theory tools and neural network instruments which is supported by relevant software. To solve the problem, the use of hybrid neuro-fuzzy systems which combine the elements of fuzzy logic and neural networks has been proposed. The subsystem ANFIS from the Matlab simulation environment has been considered and used as a tool for the creation of an appropriate model of such neuro-fuzzy systems. The structuring of factors determining the level of innovation has been developed and a formalized statement of the problem has been given in the modeling. The architecture of the corresponding system of fuzzy inference in the form of a neuro-fuzzy network ANFIS, which implements the fuzzy inference system Sugeno has been presented and described. The sequence of the development and implementation of the computer model has been presented in detail. An example of practical calculations of the developed model has been supplied and the main results have been analyzed. On the basis of the constructed model the impact of factors and conditions on the rating of innovation development has been estimated. The proposed approach to the use of the neurofuzzy model of innovative development evaluation makes it possible to predict the rating in the future. The results have confirmed the possibility of practical application of such models to predicting the total rating of innovative development.