Sensor Networks

In the innovation project  InSensEPro , wireless communication within harvesters and the possible applications of artificial intelligence to sensor networks were investigated.

CLAAS KGaA mbH

Marvin Barther

Marvin Barther

Details

Content Blocks

Motivation

Crop conditions are subject to continuous change because fields are highly inhomogeneous. This means that in a field, plants can vary greatly in size and their ears can vary greatly in grain number and grain size due to soil influences, water spots, shading, or even different management of the field. Thus, the harvesting process must be continuously observed and evaluated, and the combine must be adapted to the respective conditions in the field.



Until now, the harvesting process in combines has only been recorded at very specific points, which makes observation and evaluation possible only to a limited extent. With a distributed sensor network, the process should be recorded in more detail. Since the processes can change not only in the direction of crop flow along the harvester, but also in width, the sensor network should be spanned two-dimensionally.



For this, however, sensors must also be able to be mounted at positions of moving elements. The strong mechanical influences and the severely limited installation space within the harvester do not allow for wear-free routing or relief of the cables for supply and communication with the sensors. For this reason, the sensors should be able to communicate wirelessly. 

Courage for a distributed sensor network

As part of InSensEPro, a distributed sensor network was introduced into the harvester. Due to the limited installation space, modifications were made to the harvester to enable optimal distribution of the sensors. The goal was to be able to scan the processes as spatially dense and multidimensional as possible.



During the evaluation of the data, different artificial intelligence methods were investigated with different architectures and data representations of the sensor network. A two-dimensional structuring of the data turned out to be particularly suitable. The high spatial resolution and two-dimensional data representation made it possible to analyze the processes in more detail and identify anomalies more precisely. From this, new insights into the harvesting process could be determined.



Thus, the insight can be gained from this project that for the analysis of material conveyances, crop flows, area loads or similar, a multidimensional data recording can contribute significantly to success. It is recommended to start with the density of the sampling initially as high as possible. 

BLE as a communication technology

In the context of InSensEPro, sensors should be able to communicate wirelessly. Through the use of a sensor network, communication with as many participants as possible should be possible. Furthermore, the transmission energy should be kept as low as possible, since the lack of wired connection options means that the sensors must be supplied either with energy storage devices such as batteries with the longest possible replacement periods or with the aid of energy harvesting. In terms of longevity, a standardized transmission technology should also be selected that is still supported in science and industry over what is expected to be a long period of time.



With these aspects in mind, Bluetooth Low Energy (BLE) prevailed for use in the harvester in preliminary tests. In both laboratory and field use, the communication technology in the harvester was investigated with different transmitting and receiving positions. An influence due to metallic materials used statically for stability or moving, often rotating, as tools could be measured, but only had a significant influence with very long transmission distances and large amounts of material. The movement of the metallic tools themselves had no significant influence. The greatest limitation of the transmission strength resulted from the moisture in the harvested material. Here it was shown that especially receivers in the installation space inside the harvester could guarantee a very good transmission quality with the transmitters.



Thus, the conclusion can also be drawn from this project that BLE is a suitable transmission technology. In order to achieve a high tolerance to moist crop material, care must be taken to place the receivers skillfully. Significant interference from metallic objects does not occur due to the nature of the installation space.  

Summary

In summary, a proven methodology for the deployment and networking of distributed sensor nodes in a challenging environment is available as a solution module. In addition, insights from the structuring of the distributed collected data and the evaluation by machine learning methods can be transferred. 

Comments