Study On Hysteresis Compensation Of Continuous Temperature Measurement Sensor With Variable Time Constant

Disposable thermocouples have been replaced by the blackbody cavity continuous temperature measurement sensor due to its low cost and high accuracy in the metallurgical industry process. The sensor consists of two layers for extending the life of the tube, but it will lead to thermal inertia, so that the temperature measurement system measures temperature signal lags behind the true temperature of the object under test, resulting in the dynamic measurement error, and then affect the actual production.This paper studies the dynamic property of the blackbody cavity continuous temperature measurement sensor, and analyzes various factors that influence the property, then we have to get the conclusion that we cannot build mechanism based compensation model due to its indetermination of structure size, material and the depth that the sensor inserted into molten steel. The time constant of dynamic vary with each sensor and temperatures changes, and the dynamic property is non-linear. As is well known that ANN is famous for its ability to solve nonlinear problem, two sort of ANN based dynamic compensation methods are proposed to solve the problem discussed above. Firstly, two methods are proposed to describe the dynamic property of the sensor, secondly we train ANN to get the statistic model, thirdly, compensation models based on the method discussed above are proposed, at lase, experiments are conducted to test the performance of the proposed methods.Numerous results indicate that the proposed two compensation methods get excellent performance when applied to real-world data from production field. To be specific, the compensate error is below±7℃at the time of90s, while the proportion of data which meet the accuracy above can reach to about80%and90% respectively.