Tampere University of Technology

TUTCRIS Research Portal

Anomaly Detection and Diagnostics of a Wheel Loader Using Dynamic Mathematical Model and Joint Probability Distributions

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Details

Original languageEnglish
Title of host publicationProceedings of the Fourteenth Scandinavian International Conference on Fluid Power, SICFP15. May 20-22, 2015. Tampere, Finland
PublisherTampere University of Technology. Department of Intelligent Hydraulics and Automation
Number of pages14
ISBN (Electronic)978-952-15-3530-7
Publication statusPublished - 2015
Publication typeA4 Article in a conference publication
EventScandinavian International Conference on Fluid Power -
Duration: 1 Jan 1900 → …

Publication series

NameThe Scandinavian International Conference on Fluid Power
ISSN (Electronic)2342-2726

Conference

ConferenceScandinavian International Conference on Fluid Power
Period1/01/00 → …

Abstract

In this paper, we present anomaly detection and diagnostics for articulated frame steered hydraulic wheel loader. The presented methodology is based on the analysis and comparison of the responses of a dynamic mathematical model and a real wheel loader using a joint probability distribution of correlation coefficients of multiple variables. The behaviour of an undamaged machine is modelled by probability density functions of the correlation coefficients using histograms and test how well the future behaviour fits the model. First, the
time series data of multiple variables are segmented into segments of the same length. Correlation coefficients are then calculated for each segment and the distributions of the correlation coefficients are estimated by computing probability density functions using histograms. Finally, the joint probabilities that the correlations in the data segments of the time series data are observed are calculated using the already computed histograms. The diagnostics is based on the combination of static threshold and threshold based on mean value of joint probabilities. The dynamic mathematical model of the wheel loader is presented with verification results. A jammed flushing valve of the hydrostatic transmission was used as an anomaly to study the changes in the joint probability values. Finally, the efficiency of the presented method is presented
with good results regarding detection of anomalies and diagnostics of the wheel loader.

Keywords

  • Diagnostics, Time series, Anomaly detection, Joint probability, Correlation coefficients, Simulation, Dynamic mathematical model, Wheel loader, Hydraulics

Publication forum classification

Downloads statistics

No data available