Research And Application Of Machine Vision-based Cutting Target Localization Method For Cutting Machine

The cutting machine is a processing equipment that uses a forming knife mold to complete the cutting action to obtain the required sheet or semi-finished product.It can be widely used in light industry such as leather goods and packaging.With the rapid development of the Chinese economy and the improvement of people’s living standards,the accuracy requirements for product cutting processing are increasing.However,most of the cutting machines produced by domestic cutting machine manufacturers still use traditional manual positioning technology,which results in long positioning time,low positioning accuracy,inability to effectively monitor the cutting process,and other defects.They can no longer meet the high-standard die-cutting requirements of current multi-variety and small-batch products.To address the above shortcomings,this article proposes a machine vision-based cutting machine target positioning method to solve the pain points of long traditional cutting positioning time and low positioning accuracy.On the other hand,a digital twindriven cutting target positioning monitoring system was designed and developed to locate and monitor the cutting target,achieving digital monitoring of the cutting process.The main research work of this article is as follows:(1)Introduce the working principle of the cutting machine and analyze the defects of manual positioning technology,clarify the design goals of the positioning monitoring system,build the overall system framework,and complete the design of the software positioning process.The hardware equipment parameters design and selection were completed according to the requirements of positioning accuracy.(2)Research and propose a machine vision-based cutting target positioning method.For the problem of image edge blur and poor pulse noise suppression effect of traditional Gaussian filtering algorithm in image data preprocessing,an improved mixed filtering algorithm was proposed.By adding a value convolution kernel to improve Gaussian filtering and combining it with median filtering,the mixed filtering processing of the image was achieved,solving the problem of edge information loss and difficult pulse noise suppression,and improving the quality of image preprocessing.On the other hand,due to the poor image filtering effect and the small number of gradient calculation directions of the traditional edge detection algorithm in the cutting target edge extraction process,an improved edge extraction algorithm was proposed.The image filtering and gradient calculation steps in the Canny edge detection operator were improved using the improved mixed filtering and 8-direction Sobel operator,respectively,improving the detection accuracy.The feasibility and reliability of the machine vision-based cutting target recognition and positioning method proposed in this article were further verified by using the cutting target template matching of the cutting machine.(3)Research and design a digital twin-driven cutting target monitoring system.Firstly,based on the physical properties/characteristics of the cutting target,a cutting system digital twin model was established on the Unity3 D platform.Secondly,the pose information of the cutting tool,the pose information and quantity information of the cutting target were classified and processed using OPC UA technology to achieve data interaction between physical entities and digital twin models.Thirdly,using the cutting process data collected by sensors,the digital monitoring of the cutting process was realized through OPC UA,and the difference between the physical entity and the preset value was obtained by comparing with the preset processing data of the digital twin model.Finally,Unity3 D data driving was used to realize the visual display of the cutting process data and the cutting quality evaluation.(4)An experimental testing platform has been established to conduct tests on the system’s functionality.The experimental results indicate that the visual recognition and positioning of the system designed and developed in this paper achieves the cutting target,with a visual positioning time of less than 100 ms and a positioning accuracy of±1 mm.The digital twin-driven cutting target positioning monitoring system is capable of dynamically monitoring the cutting and processing process,including real-time monitoring of the tool position,real-time monitoring of the cutting target position,material offset alarm,automatic tool correction,and other functions.The study proposes a machine vision-based method for positioning cutting targets on a cutting machine.This method employs visual positioning technology instead of manual positioning,simplifying the targeting process and significantly reducing the positioning time,thereby enhancing work efficiency.Moreover,the designed cutting target positioning monitoring system utilizes OPC UA technology to achieve real-time data exchange between physical and digital spaces,boasting high positioning accuracy and reliable monitoring capabilities,and holds certain application value.