Files in this item



application/pdfKUHNS-THESIS-2019.pdf (6MB)Restricted Access
(no description provided)PDF


Title:On-the-go soil physical properties characterization using acoustic emission detection
Author(s):Kuhns, Brendan Matthew
Advisor(s):Grift, Tony E.
Department / Program:Agricultural & Biological Engr
Discipline:Technical Systems Management
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):soil physical sensing
acoustic emission
soil texture
soil compaction
on-the-go sensor
frequency spectra
Abstract:Soil physical properties are a foundational classification of measurements with direct relation to the productivity of agricultural soils. Currently, measurements of these properties are either crudely taken in field or low volumes are meticulously characterized in laboratory settings at high costs. This work outlines the development of a novel system for on-the-go characterization of soil physical properties. The system utilizes a piezoelectric acoustic emission sensor with a voltage output, embedded in a wedge which measures the interaction at the soil-wedge interface. Experiments took place in the field and in indoor and outdoor soil bins. Effects of the speed of the implement, compaction, and texture were analyzed using voltage vs. time series and frequency spectra. A linear relationship was found between the speed of the implement and the sensor output with a coefficient of determination R2 of 0.79. Measurements taken in a high compaction soil were compared to those taken in a soil with low compaction. A difference in the population median signal energy was found at an 𝛼 of 0.01. Four soil textures were sampled and their frequency spectra analyzed to determine a correlation between the soil texture and its corresponding frequency spectrum. Analytical techniques included the Welch’s power spectral density estimate, wavelet analysis, and moving average Fourier transforms. Principal component analysis using the z-score normalization of the Welch distribution allowed for separation of the frequency spectra given the texture. High levels of self-similarity between replications were seen in sands and moderate levels in loam. An analysis of variance using the Welch correction was performed and subsequent post-hoc evaluation using the Games-Howell method was completed. The results show that at an α of 0.05 all textures are separable with respect to each other texture. Future work should investigate effects of other soil properties on the acoustic signature and include development of machine learning approaches to classify soils based on these data.
Issue Date:2019-04-22
Rights Information:© 2019 Brendan Matthew Kuhns
Date Available in IDEALS:2019-08-23
Date Deposited:2019-05

This item appears in the following Collection(s)

Item Statistics