Noise source evaluation of a real-time soil sensor, part II: Dynamic noisesources

Citation
Pw. Elliott et al., Noise source evaluation of a real-time soil sensor, part II: Dynamic noisesources, J AGR ENG R, 77(4), 2000, pp. 365-377
Citations number
12
Categorie Soggetti
Agriculture/Agronomy
Journal title
JOURNAL OF AGRICULTURAL ENGINEERING RESEARCH
ISSN journal
00218634 → ACNP
Volume
77
Issue
4
Year of publication
2000
Pages
365 - 377
Database
ISI
SICI code
0021-8634(200012)77:4<365:NSEOAR>2.0.ZU;2-#
Abstract
Research was conducted which examined the effects of dynamic elevation chan ges on a real-time soil organic matter sensor. Such soil sensors are intend ed to provide quick and economical methods of measuring soil properties. Th ese soil properties are then useful in determining the proper chemical and fertilizer application rates for small sections of land. During the mapping process for organic matter content, it was found that the signal from the soil organic matter sensor contained several high-frequency components. A s tudy was conducted to examine the relationship between dynamic elevation ch anges and the high-frequency components of the sensor signal. A finite element model of the sensor housing was created to obtain estimate s of the first three natural frequencies of the system. These frequency val ues were then used to design laboratory experiments and to set data acquisi tion speeds for field tests. Controlled tests were performed at Herrick Lab oratories, Purdue University, where the only noise source to the sensor sys tem was dynamic changes in the sensor elevation relative to the soil surfac e. It was found that the level of error produced in the laboratory was sign ificant enough to warrant study during field use. Field experiments were co nducted in which acceleration inputs to the sensor assemble and the signal from the sensor were recorded by a portable data acquisition system. The re lationship between the acceleration inputs and the sensor signal were exami ned using both single and multiple coherence analysis methods. The results of the coherence analysis indicated a stronger relationship between the acc eleration inputs near the natural frequency of the sensor housing than at o ther frequencies. This suggests that great care should be taken to determin e the resonant frequency of any such sensor housing and that components of the signal above this frequency should be removed via filters prior to usin g the data for analysis. For this sensor housing, no frequency component ab ove 10 Hz should be used for the analysis of soil organic matter content. I f higher sampling rates are desired, subsequent sensor housings should be d esigned such that the housing's natural frequency is raised. (C) 2000 Silso e Research Institute.