The high-growth trend of my country's photovoltaic industry is significant, and it plays an important role in the international market. It has achieved end-to-end independent control, and is expected to be the first strategic emerging industry that becomes a model of high-quality development. It is also an important engine to promote my country's energy transformation.
Hikvision Robot actively participates in the wave of intelligent manufacturing upgrades in the photovoltaic industry, conducts in-depth research on industry needs, and combines the advantages of machine vision software and hardware products to continuously optimize system capabilities and provide visual solutions such as serial EL appearance inspection, typesetting positioning, and junction box post-welding inspection. Photovoltaic industry partners jointly promote industrial intelligent upgrading.
Quality assurance of photovoltaic module production
Photovoltaic modules use the photovoltaic effect to directly convert light energy into electricity to generate electricity. The specific process of photovoltaic module preparation can be divided into: string welding→stitch welding→lamination→EL testing→framing→junction box installation→IV testing→finished products Inspection→packaging, etc., in which stringing is to weld solar cells in series through ribbons, and enter the lap welding link after the battery strings are laid out: use the confluence tape to correctly connect the ends of the battery strings that have been welded into strings, and use other auxiliary The materials are laminated to finally complete the connection of the battery string to the battery string.
The quality of the lap welding process is directly related to the electrical conductivity of the components, but some defects may be introduced during the lap welding process, such as the offset of the welding strip, so it is particularly important to inspect the welding quality of the bus strip. At present, the commonly used detection method is manual visual inspection. Due to the large number of detection positions and detection items, and the influence of the speed of the assembly line, human inspection is often insufficient. Hikvision robotic photovoltaic module confluence belt visual inspection solution uses vision to inspect the confluence belt produced by the stitch welding machine, which can greatly improve the detection efficiency.
Challenges: The busbar texture and lead-out wires interfere with the inspection; the EVA film arches and so on will make the solder features appear random; the middle busbar solder strips show various forms such as crossing, overlapping, and single virtual soldering.
Line scan camera high-quality imaging + deep learning algorithm
hardware solution
This solution adopts Hikrobot’s self-developed high-performance color line scan camera (MV-CL042-91GC). The optical imaging has been tested and verified in multiple bases and has become a standard solution. At the same time, it is also equipped with a self-developed image acquisition signal control module. The camera, lens, industrial computer , light source, wire and other hardware of the hardware system are all provided by Hikvision Robotics.
software solution
■ Extract different channel images, each detection item is processed on the corresponding channel image;
■ The traditional algorithm locates the position of the bus bar and the lead-out line, and realizes the missing "0" false detection of the bus bar;
■ Determine the position of the cell based on the position of the lead wire, and accurately measure the string spacing, creepage distance, etc.
■ AI image segmentation algorithm is used to separate the welding strip from the confluence strip, so as to realize the positioning of the welding strip and complete feature extraction;
■ By comprehensively processing the area, length, axial ratio and other characteristics of each welding strip, it is judged whether there are defects such as excess welding strip, empty welding, and shrinkage of the welding strip.
