Referencing acoustic monitoring of cutting knives sharpness in agricultural harvesting processes using image analysis

• Published in: Biosystems engineering Vol. 226; pp. 86 - 98
• Main Authors: Siebald, Hubertus, Pforte, Florian, Kulig, Boris, Schneider, Manuel, Wenzel, Andreas, Schweigel, Martin, Lorenz, Jonas, Kaufmann, Hans-Hermann, Huster, Jochen, Beneke, Frank, Hensel, Oliver
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2023
• Subjects: Acoustics; Forage chopper; Image analysis; Knive condition monitoring; Image analysis; Acoustics; Knive condition monitoring; Forage chopper
• ISSN: 1537-5110; 1537-5129
• DOI: 10.1016/j.biosystemseng.2022.12.007

In the past, attempts were made to determine the knives sharpness in agricultural harvesting processes online using various methods. One promising method is the recording and analysis of the structure-borne sound at the counter-blade of the chopper (Siebald, 2017; Siebald et al., 2017). First investigations show, further reference systems for monitoring the knife sharpness are needed to enable a calibration of the planned system. One such reference measurement that is readily available in practical operation is the assessment of the cut material, which has been used to date in the form of a qualitative visual inspection by the operator. In the present work, in addition to an analysis of acoustic data, the determination of the particles of the crop taken in parallel is carried out. This is done using computer image analysis, which is capable of reproducing the shape parameters of the chaff particles with high accuracy and complexity (Guth et al., 1995; Savoie et al., 2014). Computation of a predictive model for the cumulative throughput by means of projection of the manifest variables onto a subset of “latent factors” was performed with the software JMP using the NIPALS method. By projecting the mean values and standard deviations of the acoustic analysis parameters onto six latent variables, it was possible to generate predicted values for the cumulative chaff throughput whose correlation with the “actual” throughput values could be described by a regression fit with R2 = 0.813. In the image analysis of the chaff data, even with only 2 variables, a R2 = 0.912 was given. •Acoustic measurements, and taking chaff samples for image analysis.•Determination of chaff quality using computer image analysis of the harvested crop.•Investigation of the correlation of the chaff quality with the throughput.•Investigation of correlation between structure-borne sound and the total throughput.•Prediction of the throughput values based on acoustic and image analysis data.