|
With the increasing competition in the manufacturing industry, the quality and quantity of traditional manual manufacturing have gradually been unable to meet the needs of modern customers, and with the increase in labor wages and raw materials, it has become more and more difficult for enterprises that were originally mainly manual to operate and manage. Therefore, it is an inevitable trend for factories to step into automation. This article takes the processing of steel bars which in the woodworking clamps as an example. While meeting product quality and mass production requirements, it is necessary to complete the purpose of intelligent and automated production lines. Based on the DCOR design chain, this article discusses the design and development of an intelligent automated production line for carpentry clip support products, referring to the process management design of the DCOR design chain and actually participating in the gradual completion of the research and development of this intelligent automated production line.
The process of DCOR design chain is mainly divided into: production line concept design, production line engineering design, production line assembly and testing. The conceptual design of the production line is based on the process analysis and planning of the processed product, collecting the data required for the product process and analyze the process, and establishing a complete operation manual for the process planning. For example, after the steel bar in the woodworking clamp in cast, it needs to go through four processing steps: punching, chamfering, surface hardening and quenching, and tempering. The engineering design of the production line needs to complete the machine selection and clamping design of the single process processing method, and propose a suitable fixture design. Two machines continuous for punching and drill bit chamfering are selected for the processing of woodworking clamp products. In the surface hardening quenching and tempering process, a tunnel-type high-frequency furnace is selected for quenching and tempering heat treatment. In this paper, each unit process is divided into three module units: mechanical processing module line, tunnel quenching and tempering heat treatment module, and conveyor belt system, and the reasonable operating procedures of each unit are planned. In this paper, each unit process is divided into three module units: mechanical processing module line, tunnel quenching and tempering heat treatment module, and conveyor belt system, and the reasonable operating procedures of each unit are planned. At the same time, this paper proposes an artificial intelligence model to predict the quality of online products in real time (hardening depth and warping deformation), including neural network models and SVM support vector machine models, which can be used to predict equipment maintenance decisions of this automated product line , to decide whether to shut down the machine to check and replace the cooling water, in order to maintain the high yield rate of the product line.
After the production line was assembled and tested, it was found that if the steel bar is not shaped during the heat treatment, it will cause excessive deformation and make it bend, resulting in unqualified products. Therefore, it is necessary to install clamping devices that can assist in shaping before and after the quenching and tempering process, so that it remains straight after thermal processing. It was also found that the cooling step was tested and found that too much water pressure would result in too shallow hardening depth, and vice versa, the hardening would be too low. The water pressure must be maintained between 6.5 to 7 kg, and the water outlet must be adjusted to the shape of the workpiece, otherwise there will be a problem that only the surface in contact with water will be hardened and the other surfaces will not. The artificial intelligence model mainly captures the power of the high-frequency furnace, the speed of the conveyor belt, and the pressure value of the cooling water nozzle online, and predicts the quality change through the data of these process parameters. The program of the SVM model requires a large amount of data to make it complete and accurate, but it takes about 10 to 20 minutes for a random inspection of a steel bar from slicing to hardness testing. Therefore, the amount of data currently held is insufficient to completely simulate and compare the real production line situation.
|