Farming has not been untouched by the IoT revolution. The relationships among field conditions, crops, and strategies for planting, irrigating, and harvesting are complex, yet vital for the success of the enterprise. Agricultural machines such as tractors, combines, mowers, and balers are key to smooth operations. Intelligent, addressable sensors on the machines and in the fields, enable farm managers to keep track of operations that are not only widely dispersed but also in motion.
John Deere's (Moline, IL) sensor networks provide two important types of data: machine information and agronomic information. Machine information is used by farmers to schedule maintenance and by John Deere to gather ideas for future improvements. Agronomic information helps farmers decide how to maximize their productivity. Machine and agronomic data are combined on a display that can be mounted in the operator's cab or sent to a computer or mobile device. When that equipment and technology are used together, farmers can execute infield decisions in ways unimagined before now.
Machine data is focused on real-time performance. Typical data includes power loads, fuel usage, and fluid levels. Power data is broken down by function, for example, engine, transmission, hydraulic system, and power takeoff unit (PTO — which runs other machinery from a tractor's engine).
Fluid levels are generally measured with simple float sensors, either the swing arm version or vertical sensors with some type of measuring device affected by float movement, such as a resistance. Others are based on capacitance. The methods for sensing power depend upon the system being measured. The easiest way on a fuel injected engine is to sense fuel delivered to the cylinders. Using the rpm and torque curve programmed into the engine, you can infer power output. Most other mechanical shaft-delivered powers are sensed by using data from a torque sensor and an rpm sensor. Power used in hydraulic systems is usually calculated using pressure and flow measurements.
This type of machine-performance data can help determine whether there are any pre-failure indications. Also, knowing the different power levels that a machine uses helps Deere to better understand their customers’ needs.
Tracking machine performance is highly important for the farmer. If a machine has to be taken out of operation for maintenance during a critical time in the growing cycle, it can have serious economic consequences. Time scheduling is a crucial factor in running a farm. For example, with corn, there is a two-week window for getting the planting done if you want to have a successful crop. If the timing is not just right, the farm's output will be significantly reduced and that can make the difference between profit and loss.
Agronomic data includes factors like yield and soil moisture and makeup, as well as constituents of the grain or forage that's being cut. Grain yield is sensed on a combine by measuring the mass flow of grain being delivered into the grain tank. Although there are several ways to do this, most manufacturers impact the grain against a plate that measures the force of the impact. That force signal is used to calculate the total mass flow. The mass flow is then converted to yield.
When irrigating, you need the right amount of water. Too much slows down the growth process. If the roots become flooded, they start to pull back and plant processes, such as the way they make sugar, shut down and take a number of days to recover.
Moisture and other constituents can be sensed via NIR (Near Infrared Spectroscopy). That method is used on Deere's Self-Propelled Forage Harvesters. NIR looks at light wavelengths in the near-infrared region of the electromagnetic spectrum and can detect moisture and other constituents as crop or grain material passes by a sensing surface. However, grain moisture on a normal grain combine is typically measured with a lower-cost capacitance type sensor.
Information on moisture is also collected by soil moisture sensors. Additional data that affects soil moisture is gathered by sensors that measure air temperature, humidity, wind speed and direction, soil temperature, solar radiation levels, and rain and leaf wetness.
Field-installed soil moisture sensors from John Deere's FieldConnect™ system come in three different probe lengths, each containing multiple sensors to gauge moisture at different depths.
There are three main uses for agronomic data:
Operations. The data can be used to measure the distribution of moisture throughout a field and to determine depletion rates of soil moisture, for example. There are different irrigation methods to choose depending on expected rain amounts, water retention in the soil, water absorption rates of the crop, water source, and size and shape of the field. Fertilization choices can include whether to put fertilizer down in the spring before planting, and come in with a second pass after the crop is up. Or in some places, put down a small amount of starter fertilizer at planting — enough to boost the plants above the ground and then put down more fertilizer every two weeks.
Planning. The data can be used to correlate yield with critical variables such as soil type, moisture level, irrigation, and fertilization patterns. This correlation can help to answer the key question, “Given the amount of money I spent for a field or section of the field to get that crop growing, protected, and nurtured, what was my yield, and what was my bottom line?” The farmer could turn over his agronomic data to a specialist for analysis. Based on this information, the advisor could suggest changes to improve the yield.
Decision-making. Another use of agronomic data is data-driven decision making. For example, “Did my choice of hybrid seeds and planting rates do what I thought? Which fertilizer approach worked best, and most importantly, what do I want to do for my next crop based on what I've learned from current and past experience? Given the amount of money I spent on a crop and how many bushels an acre I received, what was the bottom line for that field.”
Deere uses a server and modem that are connected to a CAN bus — the main data highway for their machines. Data can be transmitted via 3G wireless radio to monitoring stations and via a web-based interface to a display in the operator's cab, computer, or mobile device. Information is sent out through the machine's telematics link, usually on a day-by-day basis. The user typically wants to know where his machines are — for example, there could be five combines operating over a 50-mile radius — so if the fuel level is low, the farmer can get to a machine before it runs out. Location information can be obtained from a standard GPS, which is accurate within 1 to 3 meters in North America or Deere's GPS system, called StarFire™, a worldwide network that locates a connected device accurately to about 3 cm.
The operational status of the whole fleet can be checked minute by minute. Agronomic data is usually sent hourly or on a full-field basis.
What's on the Horizon?
Terry Pickett, manager of advanced engineering at John Deere's Intelligent Solutions Group, sees a bright future for agricultural machine networking.
“In agriculture, you're really focused on getting the maximum yield for a particular farming operation. That means optimizing the overall cost and profitability,” Pickett said.
Increasingly, according to the John Deere manager, customers want to know more about the effects of any operation they perform — knowledge that could mean the difference between making money and not.
For example, if fertilizer is used after a plant's up, the farmer may want to know if that was the right choice. Was that really a better decision than to fertilize before? Should the farmer only fertilize a little bit? Four times through the season? Once?
Those choices vary, and data can help producers to make more informed decisions. Most of Deere's customers are small businessmen, Pickett said, who can be quick to experiment and make changes if needed.
“If I can get processes in place and figure out how to do them with the equipment I have, can I raise my profit per acre, perhaps by hundreds of dollars?’ That's a big deal for a customer — the difference between making a profit or not,” said Pickett.
“I would see sensor systems that optimize their ability to analyze things quickly and to automate execution in the field, as increasingly in demand.”
This article was written by Ed Brown, Associate Editor at Tech Briefs Media Group (New York, NY), based on an interview with Terry Pickett, Manager of Advanced Engineering at John Deere (Moline, IL). For more information, Click Here .