As one of the important types of steel materials, wide and thick plates are widely used in aerospace and military industry, energy and power, marine ships and other fields, and play a decisive role in national defence construction and national economic development. Taking the high-end equipment manufacturing industry represented by marine engineering equipment as an example, the technical requirements for the outer dimensions of wide and thick plates are significantly higher than the current national standards. The width and size fluctuations are required to be controlled within a few millimetres, and the unevenness is required to be ≤2mm/ m etc. Strict quality requirements pose great challenges to the precise control level and quality assurance capabilities of the production process.
The rapid development of the new generation of information technology is driving the manufacturing industry to transform into intelligent production methods. my country’s wide and thick plate manufacturing industry generally has a good automation foundation and has the richest application scenarios in the industrial manufacturing field, represented by digital twins and CPS cyber-physical systems. The emerging technologies show strong advantages in sensing, computing, communication, control, etc., and can realize the functions of on-demand response, rapid iteration, and dynamic optimization of each control unit in the system. Therefore, it is of great significance to use intelligent means to further improve the level of precise control, promote the further development of my country’s wide and thick plate manufacturing technology, and maintain my country’s leading position in steel manufacturing.
Currently, most of my country’s wide and thick plate production lines face the following long-term problems in terms of perception of key process quality parameters and multi-process coordination and optimization:
The automation of rolling, shearing and plate forming processes has reached a high level, but there is no linkage between systems, and dynamic coordination and optimization control cannot be carried out between processes, which affects the further improvement of quality and efficiency;
There is a general lack of online detection equipment for contour and flatness. Key quality parameters such as the profile and flatness of rolled steel plates are still measured manually offline and cannot participate in the optimal control between processes;
The problem of low plate assembly order matching and shearing control accuracy that relies on manual experience is obvious. The optimal plate assembly and shearing strategy cannot be determined based on the real-time profile information of the steel plate, resulting in too much remaining material in the assembly and large cuts at the beginning and end, resulting in Yield loss;
Apart from basic product information exchange, there is no other process quality data exchange between the three processes of “plate assembly-rolling-shearing”. There is an urgent need to develop an optimal control strategy oriented to multi-objective constraints to reduce cutting losses and improve order matching.
In response to the above problems and related technical bottlenecks, Shandong Iron and Steel established a joint R&D team with Northeastern University and other units in 2010 to further promote industry-university-research cooperation and carry out continuous scientific research around key common technologies of the project.
The team focused on the three key processes of “plate assembly-rolling-shearing”, and for the first time in the industry clarified the multi-process coordination and optimization mechanism and the scientific issues that need to be solved. The developed contour and plate shape detection instrumentation device realizes online perception of key quality data in the hot rolling process, solving many problems such as the construction of the CPS information physics system, the “data + mechanism” model dual-driven appearance size control model, and multi-process coordination and optimization decision-making. Technical difficulties have resulted in the formation of original technology for efficient coordination of multiple processes in the manufacturing process of wide and thick plates. The ability to accurately control the appearance dimensions of wide and thick plates has been significantly enhanced, and the level of coordinated optimization control of multiple processes in the production line has been greatly improved.
The main innovative achievements achieved by the project are as follows:
(1) Take the lead in developing two major instruments and devices for online inspection of wide and thick plate contours and plate shapes based on machine vision, and carry out the first stabilization engineering application in my country’s representative 4300mm wide and thick plate production line.
In view of harsh working conditions such as uneven lighting and reflection on site, random distribution of descaling residue and dust on the surface of steel plates, and vibration and slippage during movement, an image acquisition mechanism based on the theory of human binocular vision was proposed, and a multi-type mixed noise high-efficiency filter was developed. It implements multiple algorithm libraries and model libraries such as removal, adaptive image segmentation, and warpage feature extraction to realize online detection of steel plate contours and plate shapes.
The width detection accuracy of the detection device reaches ±2mm, the length detection error is less than 5‰, the side bend detection accuracy reaches ±5mm, the detection accuracy of the irregular deformation area of the head and tail reaches ±2mm, and the flatness detection accuracy is controlled within ±1mm/m. . The successful development of dynamic steel plate profile and plate shape online detection instruments has completely replaced the traditional production methods of manual estimation and offline measurement. It has filled the gap in the industry in the development of large-scale detection instruments and devices and provided a basis for intelligent control of steel plate outer dimensions. Data support.
(2) Based on the flat shape online detection information, a flat shape dynamic digital twin model system with joint analysis of multi-source data and multi-models was constructed, and a flat shape CPS with features such as flat shape online identification, accurate calculation, closed-loop control and dynamic optimization was developed. The system provides new intelligent technical support for precise control of plane shape in the rolling process.
Analyze the cause mechanism of slab formation during the billet rolling process, and focus on the dynamic optimization of the control parameters of the rolling process to develop the loaded roll gap shape, elastic deformation of the roll system, controllable points of the planar shape of the rolled piece, and deformation areas in the finishing rolling forming process. Micro-tracking and other mechanism prediction models, the application of the model has greatly improved the controllability of the planar shape in the rolling process.
A data model was established using multi-source information such as existing process performance, model settings, incoming material size and composition, and a plate-shaped digital twin was constructed based on the integration of geometric modelling and “data + mechanism” models. Based on online sensing flatness data, a flatness CPS system with dynamic feedback of flatness information, closed-loop control of the rolling process, and dynamic model optimization was developed, breaking through the process bottleneck of lagging flatness control and forming an intelligent and precise control of flatness. New process technology.
(3) Developed digital twin-driven controllable point plane shape intelligent prediction technology, innovatively proposed a shear optimization strategy based on multi-objective constraints and a dynamic plate grouping strategy based on visual feedback, and successfully constructed a CPS system for wide and thick plate contours, achieving ” Coordination and optimization of the three key processes of plate assembly-rolling-shearing.
Relying on the contour data feedback and plane shape prediction model, the daughter plate arrangement strategy based on process constraints and the motherboard plate assembly model was developed based on the multi-knapsack problem, realizing scientific and dynamic plate assembly of continuous casting billets, residual material slabs, and residual material steel plates. , effectively reducing the board loss. Based on the contour information and current contract orders, the steel plate shearing strategy is calculated online to determine the optimal shearing line position to achieve intelligent shearing.
Research and development of cold/hot shape dimension mapping model systems for thousands of varieties and specifications. A multi-process coordinated optimization CPS system with precise control of steel plate profile and shape as the core is built around the contour detection data, plate grouping planning data, and rolling process planar shape prediction models among the key processes of wide and thick plates, achieving scientific plate grouping and appearance. Technical system for intelligent and precise size control.
Application and effects
The technology development and successful application of this project is a successful exploration and practice of emerging technologies such as digital twins and CPS cyber-physical systems from concept popularization to innovative applications. It plays a leading role in the “no man’s land” of intelligent manufacturing in the main production process of wide and thick plates.
After the wide and thick plate profile and plate shape CPS system based on machine vision were applied in Laiwu Iron and Steel’s 4300mm wide and thick plate production line in 2017, it effectively improved the collaboration efficiency between key processes, reduced the labour intensity of operators, and improved the production line’s overall yield rate and production The control level of key indicators such as efficiency and process cost has been significantly improved. For example, the production line production efficiency increased from 147 tons/hour to 172 tons/hour, an increase of 17%; the finished product rate increased by 91.65% from 91.21% before the project was implemented, an increase of 0.44%; the success rate of remaining wood panel assembly was 50%, with The yield rate decreased from 1.38% to 0.31%, a decrease of 1.07%; the power consumption per ton of steel decreased from 105.37kWh to 91.11kWh, a decrease of 14.26kWh; the shearing time of a single steel plate was reduced from the original 1.43min to 1.2min, and the shearing efficiency increased by 13.3 %. In the past three years, direct economic benefits have been created reaching 287 million yuan.
Some of the relevant technologies in this project have been promoted and applied in medium and thick plate production lines such as Sanshan Iron and Steel Co., Ltd. and Tangshan Iron and Steel Co., Ltd., which have improved the control accuracy of plate shape and plane shape and created good economic and social benefits.
Source of article: R&D and application of CPS intelligent manufacturing technology for wide and thick plate contours and plate shapes based on machine vision. China Iron and Steel Industry Association