Modeling Method Of Virtual Instrument Design

With the ever-increasing testing requirement in the test control domain, the traditional testing instrument increasingly can't meet the demand of the computer time. In certain industries (for example petroleum chemical industry), the situation is especially serious. Since some variables in the industrial process have close relation to the production ,yet the traditional testing instrument can not actualize on-line measurement of these variables by reason of technical level, working conditions, cost, and so on .The problem affects the production efficiency and the product quality to a certainty. In order to resolve this problem and accurately estimate and control these variables, the people focus their attention on the Virtual Instrument (VI) for its unique superiority. Virtual Instrument, with the specialities such as miniature, automation, flexibility, and individual-computer managing has become the developing trend of the testing equipment today. This article mainly study how to constitute a model of VI. The paper is made up of the following two parts.The first part of this paper gives the analysis and appraisal to the design of VI as a whole. To start with, the papers shows research background of the topic, the objective and the significance for developing VI, and concisely recount its phylogeny. And then the paper introduced its concept, characteristic and basic structure. Thereafter,the paper summarized the necessary condition of VI development, and elaborate on its engineering design steps which emphasize particularly on modeling. Finally, the paper analyses its development trend and application prospect.The second part, which includes the core of the paper, mainly study modeling method of VI. In the beginning of this part, common modeling methods are introduced, among of which the statistical regression analysis and the Back Propagation neural network (BPNN) are expatiated improved according to the analysis of their shortages. Thereafter, the paper takes example for measuring polymer viscosity and makes the further research on the modeling and design of VI .based on the analysis for the sample data, the paper preliminary designs four kinds of models : multiple linear regression analysis (MLR)model, multiple stepwise regression analysis (MSR)model,improved polynomial regression model and BPNN model. In three statistical regression analysis model, improved polynomial regression model is the best in the aspects of the model fitting and survey precision, therefore this model is chosen as a representative statistics regression analysis model. Finally, the paper is focused on building the primary variables two models which are Polynomial Regression Model and BPNN Model using Visual C++ software. The end of this part, based on the comparison and the analysis of modeling result, a conclusion is drawn and some suggestions to the next study period are briefly presented.