Introduction
Tor (2002) compared the use of L1 (minimum norm), L2 (least
squares), Kalman filter and time series analysis in deformation
analysis and found the Kalman filter a robust way of dealing
with noisy time-series data typically found in continuous
deformation measurement. Further work had since been done by
the author to provide the adaptive filtering and to incorporate the
near-real-time transmission of deformation results via GPRS
(General Packet Radio Service) data modem and issue of alarm
via SMS when “trigger” levels are reached. Comparison of the
Kalman-filtered results with that of the least squares is also made
in this paper.
Benchmark test of the application software developed by the
author was carried out at a test site in NTU. The test and the
achieved results as well as that of a simulated deformation test
are covered in the subsequent sections of this paper.

Authors:

Y. K. Tor (cyktor@ntu.edu.sg)


Download the Full Text Here