Design And Research On Sugarcane Multi-cutter Cutting Equipment Based On Machine Vision

Sugarcane is an important cash crop in our country.It is an inevitable trend of industrial development to realize the mechanization and automation of the whole process of sugarcane production.The existing sugarcane planting machines include real-time seed cutting planter and pre-cutting seed planter.Among them,the pre-cut planter has the characteristics of high sowing efficiency,less seed usage and miniaturization.It is a new type of planter born under the development of the sugarcane industry,which has gradually been popularized and applied.However,the sugarcane seeds used for sowing in the pre-seed planting machine are mainly artificially cut.The damage rate of sugarcane buds will increase significantly during manual cutting,with high labor intensity and low efficiency.Therefore,to meet the development needs of various pre-cutting planters,it is urgent to develop an automatic sugarcane seed cutting equipment.By analyzing the research results of seed cutting equipment and stem node recognition at home and abroad,and combining automatic control,image processing and deep learning technologies,this project designs a sugarcane multi-cutter cutting equipment based on machine vision.The main research contents are as follows:1.According to the agronomic requirements of sugarcane planting and the physical properties of sugarcane,a new scheme of sugarcane transportation,image acquisition and cutting is proposed,followed by the design of overall structure and working principle of the cutting equipment.On this basis,the key components of the seed cutting equipment are designed and selected,including the conveying mechanism,the image acquisition mechanism and the cutting mechanism.The three-dimensional model of seed cutting equipment is established by using Solidworks software.Through motion simulation,it is ensured that the motion mechanism will not interfere with each other.2.Taking Siemens PLC as the controller,the seed cutting equipment control system is constructed by analyzing the composition of the control system.The control principles of conveying system,acquisition system and cutting system are described in detail.The communication instructions between image identification software and PLC are designed.The principle of the control system is: the camera collects the sugarcane image and locates the stem node by the image identification software,the cutting point is obtained by artificially setting the offset from the stem node and sent to the PLC.PLC controls the lateral movement and positioning of multiple cutters,and completes synchronous seed cutting.3.With the purpose of stem node recognition,a new algorithm combining machine vision and YOLOv3 is designed.Aiming at the problem of sugarcane attitude tilt,an image correction method based on affine transformation is proposed.In order to eliminate the interference of complex background,a method using rotation matrix to obtain sugarcane region of interest is designed.In order to improve the robustness of the algorithm,a stem node recognition model based on improved YOLOv3 network is designed to preliminarily locate the stem node position.Aiming at two kinds of errors in the recognition results of YOLOv3 network model,the improved edge extraction algorithm and stem node location algorithm are proposed to accurately identify and locate sugarcane stem nodes.The off-line test shows that the average recognition rate is 97.3%,and the recognition time is 415 ms.4.The physical prototype of sugarcane seed cutting equipment is made,and the cutter positioning accuracy and seed cutting performance of the cutting equipment are verified by experiment.The results show that under the working conditions of conveying speed of 200mm/s and moving speed of cutter of 150 mm/s,the cutter positioning accuracy is far less than5 mm and the repeated positioning accuracy is less than 1mm.The average error of cutting point of seed cutting equipment is 3.8mm.