Corn population measuring system
Date
2022-12
Authors
Zhao, Chenyu
Major Professor
Advisor
Birrell, Stuart
Tang, Lie
Steward, Brian
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Precision agriculture refers to the use of modern information technology for intensive farming. The goal of precision agriculture research is use information and decision support systems (DSS) for farm and crop management, and to optimize sustainable production, while protecting the environment and improve economic returns.
Shafi (2019) stated that, “Precision Agriculture (PA) is comprised of near and remote sensing techniques using IoT sensors, which help to monitor crop states at multiple growth levels.” There are many factors that can affect crop yield, including weather, soil texture and nutrient status, and crop germination, which can significantly increase variability in plant population and plant spacing.
Over the last three decades, there have been significant advances in crop yield sensors. The yield monitors are based either mass or volume flow measurements, as the grain is conveyed into the grain tank. Therefore, there are significant time delays between when the crop is harvested and the yield sensor. These delays and co-mingling of the grain through the combine head and threshing system, makes it difficult to determine the instantaneous yield over short distances. This is particularly true, during step yield changes at the start and end of rows, and small areas within the field that have significantly lower or higher yields. A real-time corn population sensor could provide information on plant spacing, average plants, number of weak plants, and skips. This population information would provide another data layer to improve interpretation of yield sensor data and allow more accurate spatial resolution in yield maps, and also provide an important data layer for Decision Supports Systems to determine the cause and effect of yield variability.
Over the past two decades, various types of population measuring system been developed and tested, including mechanical systems, optical sensor systems, and advanced machine vison systems. All these systems have different advantages and disadvantages. The mechanical systems are contact type sensors, and in general non-contact type sensors are preferred. The infrared and optical sensors are binary, with the sensor blocked or not block, therefore it is very difficult to distinguish whether the response if from the desired target (stalks) or non-desirable targets (leaves, weeds). The machine vision type systems are very capable, but can be significantly downgraded in dirty and high ambient light conditions, and require significant processing time.
This research is to evaluate the possibility of using ultrasonic sensor technology to measure corn plant population during the harvest. The ultra-sonic sensors are utilized in attenuation mode with a transmitter on one side of the row and the receiver on the other side of the row. The rational is that the level of attenuation of the target (stalks), and non-target (leaves, weeds) will allow an attenuation threshold determined to differentiate the desired target from non-target species. In attenuation mode, the ultrasonic sensors would not be operating in a binary mode, as is the case for infrared sensors. Testing of the concept was conducted under a laboratory setting. The ultrasonic transmitter and receiver were mounted opposite of each other with the corn stalks passing between the transducers. The system tested at three different speed modes (2,4,6 mph) which spans the typical range of combine operational speeds. The normal spacing between targets was 6 inches, which would be equivalent to a plant population of 35,000 plants per acre. In addition, some targets were spaced 2 inches apart to represent double plants in the field, and some targets were spaced 5 and 7” apart to represent inherent variability in plant spacing due to seed bounce during planting. The total number of targets for one cycle was 20 targets. After signal processing, the sensor identified locations of the targets was compared to the known target positions.
The sensing system was successful in identify both artificial targets (wooden rods) and corn stalks. The number of total objects in each test was 20 objects. The systems were able to identify the location of the artificial rod targets with a 100% accuracy. In corn stalk testing, the system error was 0%, 5%, and 5%, at speeds of 4.5, 5.0, 6.0 mph respectively.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
thesis