An Intelligent Approach to Earthmoving Analysis
Ian J. Griffiths* BSc (Hons), MSc., PhD
David J. Edwards** BSc (Hons), PhD, FFB
Norman E. Gough*, BSc (Hons), MSc., PhD, CEng
Multimedia and Intelligent System The Built Environment Research Unit
Technology Research Group School of Engineering and the Built
School of Computing and IT Environment
University of Wolverhampton University of Wolverhampton
35 – 49 Lichfield Street Wulfruna Street
Wolverhampton Wolverhampton
West Midlands, WV1 1EL West Midlands, WV1 1SB
E-mail: ex1131@wlv.ac.uk

FREE ABSTRACT (to read the full article, please log in below).

The accurate prediction of tracked hydraulic excavator cycle time and productivity output is notoriously difficult for the UK construction practitioner, not least because such data is only freely available from a limited number of plant manufacturers. This paper presents the development of a feed-forward artificial neural network (ANN) for predicting tracked hydraulic excavator cycle time where the machine may operate under a myriad of environmental conditions. The ANN was trained using a sample of 86 observations for 43 machines (supplied by four collaborating plant manufacturers). The ANN met error criteria after only 70 training epochs and had a sum squared error of 0.0194. With a normal distribution of residuals, the parametric 'mean absolute deviation' performance measure, valued at 1.47 seconds, reveals the ANN model is a robust and reliable predictor of machine cycle time. The paper concludes by illustrating the benefits of the research to the construction practitioner.


To read the full article,
please log in:
If you haven't registered already,
you can do so for free:

(c) Copyright 2001 Professor F.R. Hall - University of Wolverhampton.