Research On Method And Application Of Vision-based Temperature Field Measurement For Continuous Casting Billet

The essence of continuous casting production is the heat transfer and solidification pro-cess of high temperature molten steel. In this process, the solidification speed is the most im-portant factor in determining surface and internal quality of billets, while surface temperature of billets is an important token to indicate the solidification state. Too fast upswing in surface temperature of billets will lead the internal stress concentration and induce the internal quality problem of billets. Moreover, too high and too low surface temperature of billets in the straightening zone both will also cause cracks. Therefore, controlling billets to have a good solidification speed by online monitoring surface temperature distribution of billets is an im-portant way to stabilize and improve the quality of billets.However, due to disturbances of randomly scattered iron oxide scale on surface of billets, uncertain emissivity for different steel grades and harsh environment in the field (such as wa-ter vapor and dust), reliable online surface temperature measurement for billets still remains an academic difficulty which has not been well solved yet in the metallurgical field. Concern-ing this issue, a novel approach of online temperature field measurement for billets based on the CCD-array detector with high resolution is proposed in this dissertation, and the corre-sponding main works include:the establishment of the narrow-band spectrum radiation tem-perature measurement model based on CCD, the research on temperature field correction for vision-based temperature measurement of billets, the research on the stability of temperature measurement results for billets based on multi-source information fusion and the experiment research of this temperature measurement system. The main research works and innovations are as follows:(1) Establishment of vision-based high temperature measurement model and temperature calibration error correction modelA vision-based radiation temperature measurement model has been established for high temperature field, and it is the precondition to realize online temperature measurement of bil-lets based on CCD-array detector. In this dissertation, according to the operation principle of the CCD-array detector, and combining with the radiation thermometry theory and geomet-rical optics theory, a narrow-band spectrum radiation temperature measurement model based on CCD is established. In the model, the influence of optical system parameters on tempera-ture sensitivity and measuring range are analyzed systematically, and based on these analysis results, optical system parameters in this system have been determined. Moreover, in order to reduce temperature measurement error aroused by the difference between measurement dis-tance and calibrated distance, a model of temperature calibration error correction is built, and after correction by means of this model, temperature measurement error can fall from 4℃ to less than 1℃ under the condition that the measure-ment distance is larger than calibrated dis-tance (2m).(2) Distortion correction for vision-based temperature field measurement of billets based on CCD sensorsThe accuracy is an important index of temperature field measurement system. However, the response non-uniformity of the photosensitive cells in CCD sensors and vignetting in op-tical system both can arouse serious photometric distortion of temperature field measurement, and the corresponding error of temperature field measurement can reach up to 70℃. Consid-ering the response non-uniformity of the photosen-sitive cells, the traditional segmented non-uniformity correction algorithm is employed in this dissertation and gray value non-uniformity could fall to 1.4‰ after correction. Meanwhile, the temperature field distor-tion aroused by vignetting has been also considered. In this dissertation, the specific reasons of such distortion is interpreted and a mathematical model for describing this distortion is given. According to the model, the influence of optical system parameters on the distortion of temperature field measurement is analyzed. On basis of these analysis results, a novel sce-ne-based method for vignetting coefficient estimation is presented. In this method, a simple polynomial model is introduced to fit of the photometric distortion function, whose parame-ters are obtained by maximizing the sparsity of neighborhood gray gradient distribution. Compared with the standard calibration method using a integrating sphere, the maximum ab-solute error and relative error for vignetting estimation by this method are respectively 0.052 and 10%, and the corresponding error of temperature measurement is less than 1℃.(3) Research on the stability of temperature measurement results for billets based on multi-source information fusionThe stability of temperature measurement results for billets is one of the key factors to determine the long-term stable operation of this system. However, temperature fluctuation aroused by randomly scattered stripped iron oxide scale on surface of billets can reach ±50℃ and even above, and so it brings a great challenge on the stability of temperature measurement results. In order to solve this problem, the inherent distribution character of surface tempera-ture field for billets is analyzed and extracted based on the heat transfer and solidification model, and combined with the temperature field measurement results by the high resolution CCD sensors, an online algorithm for reconstructing temperature field of billets is proposed in this dissertation. According to the algorithm, the real surface luminance temperature of billets, contaminated by stripped iron oxide scale, can effectively be reconstructed as the experi-menttal results show the temperature fluctuation can be reduced to ±5℃. Meanwhile, in order to overcome shortcomings of merely using the CCD camera in temperature measurement, such as not capable of realization of emissivity compensation for different steel grades, and being susceptible to dust and water vapor, a hybrid method of temperature measurement for billets, which combines the single-spot colorimetric thermometry with the CCD-array detector, is also proposed in this dissertation. This method is based on narrow-band spectral radiation temperature measurement thermometry and matching algorithm for multi-source data at dif-ferent scales. Experimental results show that this method could effectively reduce temperature measurement errors aroused by uncertain emissivity for different steel grades, and can pro-long the maintenance period of this system to over 3 months.(4) Analysis for influence factors on the accuracy of temperature measurement of billets and the field test of this systemAt the end of the dissertation, the influences of the CCD dark current and the emissivity of billets surface on the accuracy of temperature measurement of billets are further studied both in theory and experiment. The theoretical and experimental results show that the dark current can cause large error at low temperature of billets, and this error will increases with the environment temperature of CCD sensors. For this system, which is cooled by an air freezing dryer and semiconductor cooler chips, the maximum absolute temperature measure-ment error and relative error due to the dark current are respectively 2℃ and 2.7%o, and the average absolute and relative error of temperature measurement within the whole temperature measurement range (800～1200℃) are respectively only 0.6℃ and 0.7%o. Moreover, based on Tohru Iuchi’s emissivity calculation model and the temperature measurement experiment for billets with a thermocouple, the error aroused by emissivity is also discussed in this disserta-tion with an example of GGr15 steel grade, and analysis results show that the temperature measurement deviation between the colorimetric thermometer used in this system and the thermocouple is 9℃ when the true temperature of billets is 1200℃. Finally, many experi-ments on the system are done for temperature measurement of billets and plate blank in sec-ondary cooling zone end-piece, and the stability of temperature measurement for this system is illustrated by concrete examples.Until now, the billet surface temperature measurement system, which is based on this dissertation’s work, has been tested in field for almost one year. Experimental results show this method can realize the online temperature measurement of billets steadily and can pro-vide a temperature measurement criterion with a high workability for closed-loop control of secondary cooling. This system has a good foreground for application, and four steel compa-nies have already signed technical cooperation agreements. Furthermore, this vision-based temperature measurement method proposed in this dissertation has a certain degree of versa-tility and it can be also used in other similar situations of high temperature field measurement.