Maize precision seeding scheme based on multi-sensor information fusion

Abstract

Seeding plays a crucial role in agricultural production. The traditional mechanized seeding suffers from inefficiencies, low precision, and lack of control, which makes it inadequate for the high demands of the modern precision agriculture, such as the high speed, high precision, and real-time control. Therefore, this study proposes a precision seeding scheme based on multi-sensor information fusion. The system uses a Controller Area Network bus to collect and analyze data from multiple sensors for accurately controlling the seeding and fertilization mechanisms and real-time monitoring the operational conditions. In addition, the structural design, functional development, and field testing of the proposed seeding scheme are analyzed. A dual-speed measurement method, which employs an encoder and a Global Navigation Satellite System receiver, is then used to develop the motor drive model. The test results show that the maximum average error in motor speed does not exceed 1.5 %. The system can accurately alarm for seeding and fertilization faults reaching a 100 % success rate, with no missed or false alarms. The incorporated novel features include a field headland switch and a one-click pre-seeding function. During the lifting and lowering of the seeder, the motor stop and start success rate also reach 100 %, with a system response time <0.7 s. The pre-seeding time can be arbitrarily set, which allows to avoid the issue of no seeds falling at the start of the seeder. Moreover, the wind pressure measurement of the system has an average relative error of 0.83 %. The long-term operation tests show no faults, and all the functions remain normal. Furthermore, the field test results show an average qualified seeding rate of 94.81 % and an average seed spacing variation coefficient of 14.1 %, which demonstrates the high accuracy and stability of the system.