Abstract
The paper describes the tracking of the location of a moving load with an unknown, harmonically varying magnitude on a plate using the distributive sensing method. This method can lead to surface contact tracking systems of robust and simple mechanical construction. A mathematical identification model that relates the dynamics at strategically chosen sensing positions of the plate to the location of a force moving on a plate has been constructed using the distributive sensing method. This model involves numerical simulation of plate dynamics, genetic algorithms for sensor position optimisation, Karhunen-Loeve decomposition for data dimensionality reduction and neural networks for determining the contact locations of moving loads. Using this method, it is shown that the actual position of forces moving at various speeds can be determined to within 2% error of the value at speeds less than 3.2 m s-1. A technique explored involves a moving window frame of reference. The effects of moving window size and noise tolerance in the model are also investigated in the paper and show that the method is robust. Potential applications of the study are in tracking and identification of moving people, vehicles and other vibrating objects on mediums such as floors, bridges or industrial environments.
Original language | English |
---|---|
Pages (from-to) | 28-36 |
Number of pages | 9 |
Journal | Sensors and Actuators, A: Physical |
Volume | 138 |
Issue number | 1 |
DOIs | |
Publication status | Published - 20 Jul 2007 |
Keywords
- Distributive sensing
- Karhunen-Loeve decomposition
- Moving load
- Neural networks
- Object tracking
- Plate dynamics
ASJC Scopus subject areas
- Electronic, Optical and Magnetic Materials
- Instrumentation
- Condensed Matter Physics
- Surfaces, Coatings and Films
- Metals and Alloys
- Electrical and Electronic Engineering