Design And Experiment On Intelligent Intra-row Mechanical Weeding Device Of The Ripple Maize

As one of the main food crops in our country,corn has an important economic status,so ensuring corn production is of great significance to my country’s economic development.Relevant studies have shown that weeds in the field are one of the main factors leading to the decline in corn yield,and severe weeds can make corn yield loss rate up to 50%.At present,the main weeding methods at home and abroad include chemical weeding and mechanical weeding.Chemical weeding is highly efficient and easy to operate,but it also produces many negative effects,such as environmental pollution,weed resistance,and pesticide residues that threaten people’s health.In recent years,with the increase of people’s awareness of environmental protection and the continuous expansion of demand for green food,mechanical weeding methods have become more and more popular.This method has high weeding efficiency and no pollution to the environment.The development of mechanical weeding equipment between rows has been relatively mature,and the weeding rate can reach more than 90%.Because of the risk of damaging crop plants in the process of avoiding seedlings,mechanical weeding between rows has always attracted everyone’s attention.Therefore,research on mechanical weeding between rows has Significance.Based on the analysis of the existing mechanical weeding mode between plants,this paper aims at the problem that some domestic and foreign mechanical weeding devices have the risk of damaging the crop roots during the avoidance of seedlings.Combining the target detection technology based on deep learning algorithms,the optimization design method is adopted.An intelligent inter-plant weeding device based on protection of the corn root system.Through theoretical analysis,virtual prototype design,training and optimization models,analysis and simulation of key components,and indoor and outdoor tests,the designed weeding performance is optimized.The main contents and conclusions of the research are as follows:(1)Design and working principle of intelligent intra-row weeding robotThe intelligent inter-crow weeding robot includes a robot mobile platform and an intelligent inter-crow weeding unit.The target detection method of the intelligent inter-plant weeding unit uses a detection model based on deep learning technology.The working method is selective weeding,that is,only weeding is performed when weeds are detected between the plants.In addition,the device mechanism design of the weeding unit can satisfy the three-dimensional opening and closing movement of the weeding shovel,and the weeding shovel can realize the mode of avoiding seedlings and weeding on the soil,reducing the damage to the root system of the corn seedlings.(2)Design and analysis of key components of intelligent intra-row deviceThe maize seedlings at the 3～5 leaf stage of ridge cultivation and their planting environment were measured,and the measurement results were used to guide the design of the key components of the intelligent inter-plant weeding device.The key components include the frame,the profiling mechanism,the transmission mechanism and the weeding shovel.The transmission mechanism of the intelligent inter-plant weeding device controls the weeding shovel to complete the spatial threedimensional opening and closing movement.The crank and rocker mechanism with no quick return feature indirectly controls the horizontal opening and closing movement of the weeding shovel,and the linear slide module controls the vertical direction of the weeding shovel.Lifting movement.Through theoretical analysis and calculation,it is determined that the lengths of the crank,connecting rod,rocker and frame of the four-bar mechanism are 33 mm,270mm,128 mm and 297 mm,respectively,and the length of the connecting frame of the profiling mechanism is 235 mm.According to the motor parameters,the time period of the opening and closing movement of the weeding shovel is in the range of 0.45～1s,and the time period of the lifting movement is in the range of 0.38～1s,which all meet the working conditions at different vehicle speeds.Using ADAMS software to simulate the trajectory kinematics of the model,the obtained trajectory meets the expected requirements.The ANSYS Workbench software was used to perform finite element static analysis on the frame and the shovel,and the connection between the shovel and the shovel handle was reinforced according to the simulation results in order to reduce the stress concentration.(3)Establish a corn seedling and weed detection model based on deep learning technologyIn order to establish a corn seedling and weed detection model based on deep learning technology,two parts of work were carried out.1.Collected 3～5 leaf stage corn seedling field images,performed data preprocessing,data enhancement and labeling,and made a data set containing 8000 images.2.Select the YOLOv4 detection network and use the above data set for model training.After 20,000 iterations of learning,the highest accuracy,recall,F1 value and m AP of the model are 96.07%,96.59%,96.27% and 95.17%,respectively.The model Performance is good.(4)Construction of weeding strategy and control system of weeding shovelA weeding strategy is formulated for the intelligent weeding unit: After the seedling and grass detection system detects the corn seedlings and weeds,it calculates and judges the distance between the targets.Only when the position of the weeds between the plants meets the weeding requirements,the detection system will Send information such as weeding instructions to the control system.The intelligent inter-plant weeding unit is a selective weeding method,so the discontinuous weeding action during weeding will cause the servo motor to start and stop frequently.At the same time,the screw motor that controls the lifting of the weeding shovel also needs high-frequency change of direction,resulting in reduced motor control accuracy.In order to prevent the sudden change of the motor acceleration and reduce the impact,this research adopts the S-shaped acceleration and deceleration speed curve to control,so that the stepping motor runs quickly and stably.(5)Experimental research on intelligent intra-row weeding robotIn order to study the performance of the intelligent inter-plant weeding robot,prototypes were trial-produced and the indoor soil tank test and field test were carried out in sequence.The indoor soil tank test debugs the linkage of each system in the intelligent weeding unit and conducts a preweeding experiment.Single-factor,multi-factor,and comparative experiments were carried out in the field.The machine advancement speed and the diameter of the corn seedling protection area were used as the experimental factors,and the performance indexes of weed removal rate,seedling damage rate and root damage rate were studied.The test results showed that the detection rate of corn seedlings ranged from 92.86% to 97.11%,and the detection rate of weeds ranged from 90.21%to 94.44%.The data processing software Design-Expert 8.0.6 was used to analyze and optimize the obtained test results,and the primary and secondary sequence of the impact of the intelligent weeding unit’s advancement speed and the diameter of the corn seedling protection area on the operation performance and the optimal parameter combination of the weeding unit were obtained.When the advance speed of the weeding unit is 0.46m/s and the diameter of the corn seedling protection zone is 70 mm,the weeding performance is the best,and the weeding rate,seedling damage rate and root damage rate are 82.75%,3.08% and 5.96%,respectively,which meets the requirements of corn field weeding.