Short-circuit Detection Method Of Zinc Electrolytic Electrode Plate Based On Multi-view Infrared Thermal Imaging And Its Application

In the process of zinc electrolysis,the short-circuit of anode and cathode plates is an important factor affecting the current efficiency and the quality of cathode zinc precipitation.The electrode plate short-circuit detection based on infrared thermal imaging has the advantages of wide temperature measurement area and strong real-time performance.However,the multi-view infrared thermal imaging occurred under the ball machine brings some challenges to the classification and location of electrode plate short-circuit.Therefore,this paper carrried out the research on the method and application of zinc electrolytic electrode plate shortcircuit detection based on multi-view infrared thermal imaging.The main research work and innovative results are as follows:(1)Aiming at the influence of multi-view perspective distortion on the classification and location of electrode plate short-circuit under the ball machine imaging,a perspective distortion correction method based on line group screening and vanishing point extraction is proposed to correct perspective distortions automatically.Aiming at the distortion of image gray features after perspective distortion correction,a cubic convolution interpolation method based on local gradient fusion is proposed to restore image gray features effectively.(2)In view of the difficulties in electrode plate short-circuit classification and location caused by single short-circuit features,small class differences,fuzzy thermal images and low resolution,an electrode plate short-circuit detection method based on improved YOLOv5 is proposed.Firstly,aiming at the plate scale variations caused by multi-view perspective distortion,a multi-scale channel attention(MSCA)mechanism is introduced to improve the detection accuracy of small-scale short-circuit plates.Furthermore,CIo U Loss and Focal Loss are proposed to improve the positioning accuracy and difficult sample learning ability.Experimental results show that the improved yolov5 network significantly improves the overall accuracy of electrode plate short-circuit detection.(3)Combined with the demand of zinc electrolysis production,an online detection system of zinc electrolysis electrode plate short circuit based on thermal imaging ball machine is developed.The position of the slot area is calibrated to establish a mapping library before the detection.The system automatically completes the collection and processing of the slot area infrared thermal image in the PTZ preset track inspection,and then detect the slot number,plate number and short-circuit type of the short-circuit electrode plate,thus improves the production efficiency and automation level.