Research On Surface Temperature Monitoring System For Rare Earth Electrolytic Melting Based On Infrared Thermal Imaging

In the rare earth electrolysis process,electrolysis temperature is one of the important parameters affecting the efficiency and quality of rare earth metal electrolysis.At present,the traditional electrolysis temperature measurement methods mainly include molten electrolyte observation method and thermocouple method.The melting observation method has a strong dependence on the operator ’s experience and has a large error.Although the thermocouple method can achieve high measurement accuracy,it has short service life and low intelligence,and it is difficult to achieve long-term real-time temperature monitoring.In order to solve the above problems and realize the accurate and real-time molten electrolyte temperature monitoring in the whole process of electrolysis process,this paper proposes a rare earth electrolysis molten surface temperature monitoring system based on infrared thermal imaging.The system uses the improved wavelet threshold function algorithm to process the infrared thermal image of rare earth molten electrolyte.The temperature prediction model is established by BP neural network optimized by seagull algorithm(SOA).Finally,the online monitoring software of electrolysis temperature is designed to assist the operator to monitor the electrolysis temperature in real time.The research contents of the subject are as follows:(1)The experimental platform is built according to the actual working conditions,and the collected infrared thermal image is preprocessed.Firstly,the weighted average method is used to transform the gray level of the image.Secondly,the feature processing effect of conventional method and traditional wavelet denoising method on infrared thermal image is compared,and the controllable decomposition factor is introduced to optimize the wavelet denoising method.Finally,the results of image processing are verified by experiments.The experimental results show that compared with the traditional wavelet denoising method,the image processed by the improved wavelet denoising algorithm increases the peak signal-to-noise ratio by 1.427 d B while retaining the complete details,showing a good trend.(2)The BP neural network temperature prediction model is established,and the model is optimized according to the prediction results.Firstly,based on the gray mean value of infrared thermal image,combined with the voltage,current,oxide concentration and temperature information at the time of the image,the BP neural network temperature prediction model is established.Secondly,the SOA algorithm is used to optimize the BP neural network model,and the prediction accuracy is improved by about 6 % after optimization.(3)The online monitoring software of electrolysis temperature is designed to assist the operator to monitor the electrolysis temperature.Through experiments,the predicted temperature results are compared with the actual values.The results show that the data with an error of less than 5 °C between the predicted temperature and the actual value can reach 90 %,which meets the actual production requirements.The experimental results show that the rare earth electrolysis melting surface temperature monitoring system based on infrared thermal imaging can assist the manual temperature monitoring of the rare earth electrolysis process,and has certain feasibility and practicability.