Cable yarding remains an important option in steep terrain timber harvesting, a reason for which new or improved operational efficiency models are required to support science and practice. Developed traditionally, these models are known to require many resources, a reason for which new approaches to the problem were researched lately, mainly by the use of Global Navigation Satellite System (GNSS) data, spatial and statistical inference systems. This study evaluates the possibility of using GNSS data and machine learning techniques to classify important cable yarding events in the time domain. Three classes were assumed by the study as being relevant for cable yarding operational setup, namely carriage moving in the uphill (MU) and downhill (MD) directions, as well as carriage stopped (S). Data collected by a consumer-grade GNSS unit was processed to extract some differential parameters which were coupled with GNSS motorial and geometric features to feed a Multi-Layer Perceptron Neural Network with Back propagation (MLPNNB) in a pre-evaluation phase which aimed at mining the data structure as a strategy to develop the best MLPNNB configuration for training and testing. Leg distance, difference in elevation, speed of the carriage, and difference in heading were used together and interchangeably in this phase, based on logical assumptions. As a result of pre-evaluation, a MLPNNB using all these datasets was found to be the best scenario. Based on this outcome, the data was split into a training (70%) and a testing (30%) subset, then the MLPNNB was used to learn and generalize on these subsets. The main results indicate that the MLPNNB had an excellent performance, with a classification accuracy of 98.7, 98.4, and 98.8% in the pre-evaluation, training, and testing phases, respectively. Log-loss errors were also found to be very low (5, 5.9 and 4.1%, respectively), indicating a high generalization capability of the MLPNNB model. Based on the results, the main conclusion of the study is that original and derived GNSS data coupled with machine learning techniques could prove to be an important tool for operational monitoring and cable yarding efficiency model development, mainly due to the possibility of working with large amounts of data.

Classyfing operational events in cable yarding by a machine learning application to GNSS-collected data: a case study on gravity-assisted downhill yarding

Proto A. R.
2022-01-01

Abstract

Cable yarding remains an important option in steep terrain timber harvesting, a reason for which new or improved operational efficiency models are required to support science and practice. Developed traditionally, these models are known to require many resources, a reason for which new approaches to the problem were researched lately, mainly by the use of Global Navigation Satellite System (GNSS) data, spatial and statistical inference systems. This study evaluates the possibility of using GNSS data and machine learning techniques to classify important cable yarding events in the time domain. Three classes were assumed by the study as being relevant for cable yarding operational setup, namely carriage moving in the uphill (MU) and downhill (MD) directions, as well as carriage stopped (S). Data collected by a consumer-grade GNSS unit was processed to extract some differential parameters which were coupled with GNSS motorial and geometric features to feed a Multi-Layer Perceptron Neural Network with Back propagation (MLPNNB) in a pre-evaluation phase which aimed at mining the data structure as a strategy to develop the best MLPNNB configuration for training and testing. Leg distance, difference in elevation, speed of the carriage, and difference in heading were used together and interchangeably in this phase, based on logical assumptions. As a result of pre-evaluation, a MLPNNB using all these datasets was found to be the best scenario. Based on this outcome, the data was split into a training (70%) and a testing (30%) subset, then the MLPNNB was used to learn and generalize on these subsets. The main results indicate that the MLPNNB had an excellent performance, with a classification accuracy of 98.7, 98.4, and 98.8% in the pre-evaluation, training, and testing phases, respectively. Log-loss errors were also found to be very low (5, 5.9 and 4.1%, respectively), indicating a high generalization capability of the MLPNNB model. Based on the results, the main conclusion of the study is that original and derived GNSS data coupled with machine learning techniques could prove to be an important tool for operational monitoring and cable yarding efficiency model development, mainly due to the possibility of working with large amounts of data.
2022
Artificial intelligence
Automation
Big data
Classification
Efficiency
Forestry 4.0
Machine learning
Steep terrain
Time study
File in questo prodotto:
File Dimensione Formato  
Borz_2022_Bulletin_Classyfing_editor.pdf

accesso aperto

Descrizione: Free Access sul sito dell'Editore https://webbut.unitbv.ro/index.php/Series_II/article/view/1836
Tipologia: Versione Editoriale (PDF)
Licenza: Creative commons
Dimensione 755.02 kB
Formato Adobe PDF
755.02 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12318/128006
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 8
  • ???jsp.display-item.citation.isi??? ND
social impact