Automatic seedling avoidance system for paddy field weeder based on deep learning and adrc control

Abstract

The invention discloses an automatic seedling avoidance system of a paddy field weeder based on deep learning and ADRC control. The system comprises an image collection module M0, an SSD algorithm-based rice seedling detection module M1, a seedling line fitting module M2, and an ADRC algorithm-based control module M3, wherein the image collection module collects rice seedling images, the rice seedling images are input into the rice seedling detection module M1, and detection and identification are performed by using an SSD target detection algorithm model so as to obtain an identification result; a seedling column line is fit according to the identification result, the position of a weeding wheel is optimized according to the input seedling column line information by using a maximum and minimum spacing optimization algorithm, so as to obtain a target position y<r> of the weeding wheel; and with the target position y<r> as a control target, the ADRC controller realizes active disturbance rejection control in weeding process, so as to minimize seedling damage ratio. The system improves control accuracy and reduces manufacture cost of the weeder without adding the sensor, and solves akey technological problem of intelligent, efficient, low-damage mechanism-based weeding in a complex paddy field operation environment.

Description

technical field [0001] The invention relates to the technical field of measuring devices, in particular to a light-sensitive intelligent body-shaped dimension ruler. Background technique [0002] In farmland ecosystems, weeds are one of the main reasons for the decline in crop yield and quality, so the task of weeding becomes particularly important. There are two main methods of weeding in farmland: chemical weeding and mechanical weeding. For chemical weeding, the application of chemical herbicides is a simple and efficient way to control weeds in paddy fields, but the extensive application of herbicides has also brought many negative problems such as crop phytotoxicity, weed resistance and environmental pollution. As an environmentally friendly and green weeding method, mechanical weeding is in line with the development direction of the agricultural industry proposed by the state to promote agriculture with quality and green. [0003] At present, the research on farmland...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the present invention provides an automatic seedling avoidance method for paddy field weeders based on deep learning and ADRC control, and comprehensively uses theories and methods such as deep learning, target detection, and automatic control to solve the problem of intelligent, high-efficiency, and low-loss problems in the complex operating environment of paddy fields. Key Technical Problems of Mechanical Weeding

Images