Research On State Detection Method Of Copper Electrolytic Refining Process Based On Infrared Images

Copper is an important strategic material indispensable to national economic construction. In 2014 the national production of refined copper is 7.96 million tons. Copper electrolytic refining process draws much attention for enormous consumption of electricity. Current efficiency is the main factor affecting the copper electrolysis refining energy consumption. Let’s take 200,000 tons of electrolytic series for example. If current efficiency increases 0.1%, production will increase by more than 200 tons and annual sales income will increase by more than 10 million yuan. In the process of production, the main reasons for the decrease in electrolytic current efficiency are the short circuit between cathode and anode plate in the electrolytic cell, open circuit between the cathode conductive rods and bus bar, and uneven current distribution between the cathodes in the electrolytic cell. Therefore, the important content of electrolytic production management is to discover the short circuit and open circuit fault in the cell and process it in a timely manner to ensure uniform current distribution between cathodes. At present, the artificial inspection used to detect short circuit in our country requires big workload and high labor intensity. Achieving automatic inspection on the spot of manufacturing and finding abnormal short circuit/open circuit in time are of practical significance to improve the electrolytic current efficiency, increase yield, reduce energy consumption and product cost, reduce labor intensity and enhance the operating environment.Image processing methods are proposed in the paper, such as realizing automatic monitoring of electrolytic cell short circuit by cell FIR image enhancement, geometric correction, feature point positioning, edge detection, image segmentation and abnormal rods mark. Moreover, according to the mathematical model of relationship between temperature distribution on the conductive cathode bar and current, environment temperature, the infrared image of electrolytic cell is used to get temperature change, estimate the current distribution between cathodes in the cell and find the anomalies of short circuit and open circuit.Considering such problems exist in the collected FIR image as the contrast is poor and the detail is not sharp enough, the adaptive contrast enhancement method based on spatial domain is adopted. Besides, infrared image is enhanced by the modified particle swarm and the optimization mechanisms such as neural network algorithm, improving the computational efficiency and robustness. Geometric correction and angle correction are also applied to barrel distortion and deflection. Cell horizontal and vertical edges are accurately positioned by using feature points inherent in electrolytic cell pot sides and the improved edge detection algorithm, achieving the precise segmentation of the cell image.For the segmented cell image, difference image longitudinal integral method based on the floating threshold strategy is first proposed, making respective tests of ROIs for covered and un-covered electrolytic cell, with the detection rate of 97%. In order to further improve the detection rate, the feature vector of fault samples and non-fault samples is constructed, and a higher detection rate is obtained by using after-training SVM classifier. Batch image fault identification results verify the effectiveness of the proposed method.With more physical field simulation software COMSOL, a physical model of cell cathode bar temperature distribution is derived. Based on the sample data, with the method of PLS and SVM respectively, a mathematical model is built up between temperature distribution and current going through the cell cathode conductive rods, thereby making a detailed comparison between model accuracy and robustness. Moreover, with the help of the mathematical model of SVM regression, the cell cathode bar FIR image is processed. Consequently, the error between estimated value and actual value of current distribution in the cell cathode bar is small. The open-circuit faults between the cathode conductive rods and the bus are accurately detected. Meanwhile, non-contact monitoring of the current distribution in the copper cell is made possible, providing the enterprises with a strong support to optimize cell type and production technology according to current distribution.Finally, an inspection device with the function of positioning photo tour is designed and put into practice. This device can receive control instruction, send information via a wireless network, make reciprocating run according to the specified path, use radio frequency card to achieve cell grouping and positioning photographs, of which the FIR image taken is downloaded by WIFI network to acquisition end computer for futher image processing. The system has been applied to the 200,000-ton copper electrolytic refining plant of Jinchuan Group, obtaining good economic and social benefits and achieving desired effect.