Don Carr, Zapopan, Mexico

We have developed a system based on GNU Linux, Apache, PostgreSQL, end Framework, the REACT control engine, and various open source technologies to create a system for remote data collection, analysis, and control of silos. We also developed some printed circuit boards with micro-controllers to read the sensors and send the control outputs. The system measures ambient temperature and relative humidity in order to know when to ventilate the grain in order to arrive at the most optimum temperature and humidity of the grain, given the local climate conditions. The motors are turned on and off using two relays wired in parallel with the physical on/off buttons for the motor starters. We also measure columns of temperature inside of the grain with digital temperature sensors placed every 0.5 meters. The temperature of the grain is an important factor in grain conservation, and, the temperature readings can also be used to estimate how many tons of grain are in each silo. This estimation is possible since temperature inside of the grain is stable, but, makes wide swings above the grain. Thus, at every point there is a temperature sensor column, we can measure grain height ± 0.25 meters. The final step is to do a surface fit using the data from multiple columns.

The architecture of the system is as follows: There is a small GNU Linux based computer at each site that can be a small hardened Arm based computer, or, a standard x86 based computer, if conditions permit. The local computer runs the REACT control engine, that communicates with the local sensor systems, and a website, caching data locally in case that the Internet connection is down so that in can be synced when connections are restored. The local computer is connected to the sensor systems via 900 MHz radio, and to the Internet by cellular modem, or, local Internet connection depending on availability. The local GNU Linux computer communicates with the GNU Linux based server using http POST and GET, to send, and retrieve values respectively. For very large data sets transmitted such as one sample of all the temperatures in silo, we use the JSON format. The sample rates for the data stored on the server are slow enough that we can store it in a relational database (PostgreSQL), with samples for all columns stored in JSON to reduce complexity of the database structure. The user interface is based on the Zend PHP Framework. Experts in grain storage can remotely set control parameters such as the ranges of ambient temperature and relative humidity when the grain should be ventilated, hours of ventilation to avoid periods when electricity is more expensive, etc. They can also remotely turn off automatic ventilation when they have determined that the grain is in adequately conditioned.