With the rapid development of the economy and the continuous improvement of consumers’ living standards, especially the development of construction, home appliances, automobiles and other industries, the overall market demand for steel is on an upward trend. Among them, cold-rolled steel plates, which are mainly used in areas close to daily life such as automobile manufacturing, household appliances, and electronic products, are important products in the steel market. While the market has increasing demand for the production quantity of cold-rolled steel plates, its quality requirements are also getting higher and higher. How to achieve double standards of quantity and quality of cold-rolled steel plates has become an important hub for enterprises to compete in the entire product.
Cold-rolled steel plate production process
Cold rolling is made by using hot-rolled coils as raw materials and rolling at room temperature and below the recrystallization temperature. It is generally understood as rolling using room-temperature rolling materials. Cold-rolled steel plate is a steel plate produced through the cold rolling process, referred to as a cold plate. It is a finishing product in steel products. Compared with a hot-rolled steel plate, its thickness is more precise (usually between 0.1-8.0mm) and the surface is more precise. Smooth and beautiful, it also has various superior mechanical properties (such as high stamping performance and low yield point, etc.).
Generally speaking, the production process of cold-rolled steel plates can be roughly divided into 5 steps: 1. Raw materials: Prepare hot-rolled coils as raw materials for the production of cold-rolled steel plates. 2. Pickling: Use a pickling unit to remove the iron oxide scale on the surface of the hot-rolled half coils. 3. Cold rolling: Under normal temperature and below the recrystallization temperature, the pickled raw materials are rolled in multiple passes to obtain products that meet the size, plate shape and surface quality requirements. 4. Annealing: Heating the cold-rolled product to a certain temperature, and then letting it slowly cool to the appropriate temperature, so that the steel plate can obtain corresponding properties. 5. Finishing: After smoothing the annealed product to obtain a uniform thickness and a smooth and smooth surface, cut it into the required size, package it, and then sell it.
Cold-rolled steel plates, production line problems that need to be solved urgently
It is easy to see from the above production process that the cold-rolled steel plate production process is long, has many processing steps, and has a long unit. Therefore, during the production process, various quality problems will inevitably occur on the surface of the steel plate, such as streaks, holes, pits, inclusions, etc. Surface cracks, etc., will not only affect the appearance of cold-rolled steel plates but also seriously affect their performance. Therefore, it is very important to detect defects on the surface of cold-rolled steel plates.
Traditional cold-rolled steel plate defect detection is mainly manual, and this method has obvious problems:
1. Low recognition accuracy: The human eye has limited viewing distance and low resolution, making it difficult to identify small flaws on the surface of the steel plate by manual visual inspection;
2. Slow detection speed: Human energy is limited and is greatly affected by emotional states, so the detection speed cannot match production needs;
3. Poor data integration ability: the defect samples collected are limited and effective analysis cannot be obtained for subsequent production adjustments;
4. High overall production cost: Manual inspection methods are highly subjective and prone to missed and incorrect inspections, resulting in manpower requirements and time consumption that still need to be further optimized.
Today, as the market’s dual requirements for quantity and quality continue to increase, facing the thorny issue of surface quality of cold-rolled steel plates, how to apply digital technology to achieve intelligent unmanned inspection has become the key to reducing costs and increasing efficiency for steel mills!
AI empowerment effectively solves defect detection problems
Over the years, companies committed to using AI technology to empower the intelligent upgrading of traditional industries have launched many solutions and intelligent hardware devices in the steel industry, aiming to use AI to empower every production link and help the steel industry realize the development of intelligent manufacturing. Intelligent transformation of the entire process.
This time, a team in China designed a “stainless steel strip defect detection system solution” to address the problem of defect detection in cold-rolled steel plates. It can complete defect detection of stainless steel strips at process speeds of 5-120m/min and export speeds up to 220m/min, thereby achieving a double improvement in quantity + quality of cold-rolled steel plates.
The “Stainless Steel Strip Defect Detection System Solution” uses a fixed 2D+3D optical measurement system to collect, analyze, integrate and classify defect data for finished cold-rolled steel plates: using the built-in probe of the “Strip Quality Inspection System”, collect original data (3D data) and 2D images related to stainless steel strip defects, and upload the data and images to the host computer through communication equipment, collect continuous strip steel three-dimensional and two-dimensional image data, and semi-automatically mark defects from the image data Category and defect location, and establish a “strip defect database”.
Then based on the “strip defect database”, a deep convolutional neural network is designed and trained to obtain a “defect discrimination model”, and the detected defects are classified according to the “defect map”. In addition to classification and storage according to the steel plate judgment scale (specification, type), it also automatically analyzes periodic defects in the steel plate during the inspection process; it has the ability to continuously analyze defects of the same type that do not exceed the standard; and it also provides quality prompts and alarms for the steel plate.
Human-machine collaboration, AI drives production automation
In addition to assisting in defect detection, the “stainless steel strip defect detection system solution” can also realize self-automation in many aspects: automatic temperature control, adjustment, and alarm functions of the detection unit; equipment status self-checking function; automatic calibration program to achieve accuracy self-correction; Detection data is connected to the secondary database for information exchange, etc.
Workers can use the control interface to achieve an overall overview, detail retrieval, identification analysis, and statistical analysis. For example, you only need to click on a single defect in the strip length determination bar to obtain a detailed graphic report, which contains a defect map analysis table and prompts the cause of the defect and its treatment method.
Article source: “China Machine Vision Network”