| Weed is one of the main causes for the decline of yield and quality of rice.Especially intra-row weeds,which are always close to plants,have a greater impact on the rice yields.Using chemical herbicides is an efficient and simple method to remove inter-row and intra-row weeds simultaneously,but it also brings some negative problems such as pesticide damage to the crop,development of weed resistance,and environmental pollution.Mechanical weeding is an environmentally-friendly weeding method,but it is difficult to kill intra-row weeds without injuring the rice plants,due to the complexity of paddy fields and the high plant densities in rice production.In this work,a systematic research was carried out from rice plant identification and positioning,and intra-row mechanical weeding,to discrete element simulations of cutter-soil interaction and optimization of machine operation parameters.The main research contents and conclusions are summarized as follows.(1)A method of positioning rice seedlings using rice canopy images was proposed.The rice canopy image acquisition system was constructed based on the characteristics of paddy field environment.The method of collecting images under shading conditions was developed by analyzing the influence of reflections and bright spots of natural light on the recognition accuracy of rice plants.Grayscale images were converted from RGB images based on the super-green feature algorithm.After filtering by Point Filter,the image of the rice plant was segmented by the Ostu automatic threshold method.Then a complete rice plant was identified by delineating the seedling examination area in a binary image,calculating the number of connected areas to extract the complete rice plant.In the end,a dynamic positioning method of a rice plant was proposed by determining the rice plant center based on the connected areas.The experiments were also carried out both in a soil bin and a field.Results from the soil bin experiment showed that the positioning accuracywas 8 mm for 50 samples captured with a water layer thickness of 1.5 cm,and 9.1 mm for50 samples with an anhydrous condition.In the field experiment(12 days after mechanical transplanting),the positioning accuracy was 8.6 mm for 50 samples and the correctness was90% in the protected areas delineated by the rice plant stem base.The experimental results showed that the proposed recognition and location method of rice seedling center met the requirements of technical accuracy in early mechanical intra-row weeding(approximately10 days after mechanical transplanting).(2)A rice plant positioning method based on the edge fitting of the stem base region was developed.Camera installation parameters were obtained for laterally capturing the characteristics of stem at the base of the rice plant in a paddy field.The rice plant image was processed and segmented by ultra green characteristic transforms,automatic thresholds,and morphological operations.The fitting range was divided into subranges in the binary image,and the edge fitting and extracting the midpoint of the stem base were performed in each subrange.The middle line of the stem base was fitted by the least squares method,and the accurate positioning of the rice plant was obtained by the midpoint of the middle line that was defined as the positioning center.A field comparative experiment was also carried out to test the accuracy of the identification and location.Results showed that within 10?20days after transplanting,the positioning error based on the stem base was kept below 7.0mm.The positioning errors at 10,15 and 20 days were 6.9,5.9,and 6.3 mm(the water condition)respectively,and the positioning errors at 10,15,and 20 days were 6.8,5.8,6.5mm(the anhydrous condition)respectively.The positioning error based on rice canopy at10,15,and 20 days were 8.2,11.3,15.4 mm(the water condition)respectively,and the positioning errors at 10,15,and 20 days were 9,10.7,and 16.1 mm(the anhydrous condition)respectively.The positioning accuracy based on the positioning method of the stem base of rice plants exceeded 94% during the weeding period.The experimental results showed that the rice plant positioning method based on the edge fitting of the stem base region was suitable for the weeding period(10?20 days after rice transplanting),and met the requirements of technical accuracy in mechanical intra-row weeding.(3)A fast-responding pneumatic swing type intra-row weeding device for rice fields was designed.The design parameters were optimized by analyses of the motion trajectoryof the mechanism.The length of the connecting rod was 35 mm,the length of the swinging rod was 72.24 mm,and the horizontal and vertical distances of the weeding tool to the center of rotation were 84 and 191 mm respectively.The telescopic speed of the cylinder to dodge rice plant should be greater than 0.25 m/s,determined from kinematic simulations by the motion analysis module of Pro/E,which were done based on the forward speed of the weeding device and the expansion speed of the cylinder.The greater the expansion speed of the cylinder,the lower the probability of plant injury was.Considering the plant injury rate and the stability of the system,the reasonable cylinder expansion speed was in the range of0.25?0.45 m/s.A control system working mode was established using a PLC as the core of the control system.By measuring the delay time and avoidance time of the system,the PLC parameters were set to alternate the two states of weeding and evading rice plant.(4)The structure of a weeding tool was optimized by modeling the interaction process between the weeding tool and paddy soil using the discrete element method(DEM).The mud-water mixed particles of paddy soil and the action of weeding tool in paddy soil were modeled by a DEM software,EDEM.Using the model,the resistance of weeding tool and soil disturbance were analyzed under different working parameters.The weeding tool was designed to have different structures of blade claw,and the optimal balde claw structure was chosen.Simulation results showed that when the blade claw interacted with the paddy soil,bending balde claw had less soil resistance than straight balde claw,and the bending balde claw also could generate a larger amount of soil disturbance.The better structure was the bending balde claw in 10°,whose force was 3.12 N when contact with soil,and the soil was disturbed to a greatest extent at this angle.The simulation results were confirmed by the soil bin test which had the same measured values.The resistances of weeding tool and soil disturbance were analyzed in different working parameters with the optimized bending balde claw.The result showed that the resistance of weeding tool and soil disturbance were increased with higher forward speed and the greater weeding depth(cutting depth).The weeding depth had more significant effects on the results than the forward speed.(5)The experimental design optimization method was used to test the intra-row weeding effects under different working parameters in a paddy field.The intra-rowmechanical weeding test platform consisted of a paddy field transplanter carrier,an integrated visual positioning system,a intra-row weeding device,a hydraulic rectification system,a speed measuring system,and a control system.The field experiment was designed as a three-factor,five-level,two-rotation orthogonal test.Cylinder expansion speed,tool forward speed,and weeding depth were selected as three test factors,and weeding rate and plant injury rate were selected as two test indexes.Results showed that the factor significantly affecting the weeding rate was the weeding depth,followed by the forward speed of the machine.The cylinder expansion speed had little effect on the weeding rate.The best combination of the working parameters obtained from the experimental optimization were the cylinder expansion speed of 0.45 m/s,the machine forward speed of 0.25 m/s,and the weeding depth of 2.5 cm.Multi-site field experiments on intra-row mechanical weeding performance were carried out in the Comprehensive Experimental Station of National Rice Industry Technology System in Jiangmen and Zhaoqing cities.Field test results indicated that the intra-row weeding technology and the weeding device developed in this study could meet the requirements of inter-plant weeding rates more than 80% and plant injury rates less than 4% during the specific weeding period(10-20 days after mechanical transplantation). |
PDF Full Text Request