Research On Molten Steel Temperature Measuring Model Based On Infrared Thermography

Usually in the process of steel production workers generally use a thermocouple to measure temperature. Its disadvantage is that when the temperature measurement is directly contact with the target, the target temperature field changes. In order to make up for this defect, according to the principle of infrared thermal radiation, a system of measuring the temperature of liquid steel by infrared thermal imaging system is designed.For steelmaking environment complexity, its physical, chemistry, kinetics are difficult to determine. Molten steel surface temperature field distribution varies fast with time. It is difficult to use directly ordinary mathematical formula to establish an accurate model of molten steel temperature measurement, and is put forward a method by using BP neural network method to establish the model of molten steel temperature.Firstly, based on the understanding of the radiation temperature measurement principle and the characteristics of infrared detector, using Mallat wavelet algorithm is used to decompose the image obtained in the experiment, the high frequency components of the image after image decomposition, and then after wavelet transform to get the image, then the image feature data collection, for the subsequent establishment of temperature measurement model of liquid steel.Then, using least square method and BP neural network, curve fitting model is established respectively and comparing the temperature of the two models results for comparison. neural network fitting results will be better than the results of least square method. But two temperature measurement accuracy of all have reached the design requirements, so that the temperature measurement error is within 1%. So in real application, if the high accuracy is required and time allows the use of BP neural network to establish the model can be applied for the measurement. In general, the use of the least square method to establish the model is enough.Finally, after analyzing the functional requirements of the software, the MATLAB software is used to design the module function, the software work flow and friendly manmachine interface operation platform, which can be used to query the temperature of the molten pool surface temperature field, and the temperature field can be displayed clearly and accurately through display temperature field isotherm, morphology and pseudo color image. It is proved by experiments that the detection system can accurately measure the surface temperature field of molten steel through non-contact type in the molten steel,and can be used to improve the accuracy of temperature measurement. By doing experiments we can know that the detection system can accurately reflect the temperature in the production of the actual situation, to be further improved after it can be applied to the actual production of steel mills.