Research On Soft Sensor Model For The Inner Temperature Of Work In Furnace

Because of the limitation of the testing technique,there is a little difficulty to the detection of the work's inner temperature in the process of delivery. To gain the inner temperature of work,researchers process the description with concentrated parameters,regression analysis and so on,but majority methods are too complex or they think the reality of delivery little,so they can not be used by the reality of delivery.The work's temperature and its distribution are one of the important rules to judge the work's heating quality and to realize the furnace's self-control in furnace,controlling the work's temperature rationally can increase the heating quality of work and reduce fuel's consumption in furnace. But the process of delivery in reality,the sensing element's technology can't measure the work's inner temperature in-line at the present time,how to measure the temperature of work in precision and in time is the premise to realize the self-control and the guarantee of the product's quality. To settle the detection and control of the work's temperature,I use the soft sensor's technology and establish the monitoring model of the work's temperature and forecast it.This paper focuses on the study of the work's superficial temperature,the inner temperature's distribution in furnace. The research contents mainly contain the following three aspects:(one),The creation of the soft sensor model to the work's superficial temperature. The soft sensor on neural network can create the model according to the target's input and output data and it has not prior knowledge,it applies for the system that is high nonlinear and very indeterminate. I select RBF neural network to create the soft sensor model and examine the model.(two),The creation of the soft sensor model to the work's inner temperature field. I calculate the work's inner temperature field applying for the mechanism model and emulate the work's inner all kinds of cases.(three),The influence of the error of the work's superficial temperature to the change of the work's inner temperature. I mainly analyze the influence of the error of the work's superficial temperature to the change of the inner temperature of work and analyze the varying regularity of the inner temperature's error to the superficial temperature's error and the correction of total convective heat transfer coefficient.All model algorithms are implemented on PC under VB and Matlab. The results demonstrate that the using of the soft sensor model can realize the difficult detection of the work's inner temperature and forecast the work's temperature distribution efficiently...