Artificial Neural Network Research Into Measurement Of Enterprises In Technical Innovation Based On TOPSIS

Technological innovation is the first motive force of economic growth, is the main country economic growth wellhead, and has created large amount of wealth for the human beings and promoted the whole world economy making prosper, therefore estimating enterprise technological innovation status scientificly and rationally not only can provide rational basis to help the all of the various levels governments know our region and every industry enterprise technological innovation developing status timely and accurately, but also be easy for the enterprise operator to discover enterprise oneself deeper problems there existing in the processes of the technological innovation, then be beneficial for the region and industries scientificly and rationally to work out policy,law and regulation as well as corresponding technological innovation tactics which enterprise adopts.Based on the discussion and analysis of the measurement of enterprises in technical innovation in the paper, sets up an index system of it from the point of inputs and outputs; considering much of the existing deficiency of current measurement of enterprises in technical innovation, the paper provides an artificial neural network method of measurement of enterprises in technical innovation based on TOPSIS, its principle is using TOPSIS to calculate ideal outputs sample values in the method of neural network. First set up an index system of measurement of enterprises in technical innovation including 3 inputs and 4 outputs, then according to the overall evaluation request and the network studying training feasibility,validity design a BP neural network model of 3-10-1 topological structure, among them, network input is 3 measurement in technical input innovation, and network output is 1 estimating value of measurement in technical innovation, but ideal outputs for the network training and studying is a overall evaluation by using TOPSIS for calculation on the 4 measurement in technical output innovation. In the example, In the example, on the recent 4 years technical innovation inputs and outputs sample values of 9 marketing enterprises, apply the method in the paper, by the MATLAB network tools, through large amount of testing and training of the sample, make model error values reach the predetermined range, then set up a neural network model which can be used in the measurement of enterprises in technical innovation.