ChenYang-ISM works also in the area of digital signal processing
in order to improve the reliability and measuring accuracy of measuring and testing systems.
It develops digital signal processing methods and algorithms for its partners. The research
and development works are focused on:
A self-calibrated measuring method is based on a self-calibration with the use of internal
reference elements and quantities. After the self-calibration the measuring errors are automatically
corrected by digital signal processing algorithms, so that the measuring accuracy of the resulted
measuring system can be improved in comparison with that of the original measuring system.
- Self-calibration measuring methods for precise measurement of electrical, geometric and mechanical
- Self-correction algorithms for accuracy improvement of digital signal processing
- Adaptive filtering technique for noise suppression of measuring systems and telecommunication systems
- Precise Fourier analysis of electrical, acoustic, seismic and mechanical signals
- Precise parameter determination of damped oscillation signals.
For measuring systems with a linear input-output relation, two reference elements are used for
the self-calibration. The measuring result is determined by a linear interpolation using the measuring
and reference data of the self-calibration.
For measuring systems with a nonlinear input-output relation, the self-calibration needs three
reference elements. The measuring result is determined by a quadratic interpolation.
The measuring errors are compensated by the interpolation. Therefore, the measuring accuracy of
a self-calibrated measuring system depends only on the tolerance of the reference elements,
normally better than 0.1%.
Different methods e.g. analog and digital filter, averaging, smoothing and lock-in amplifier are
used for noise reduction in order to improve the signal noise ratio of a measuring system. These
methods, however, are only suitable for the reduction of noise, the spectrum of which is different from
the signal spectrum. The problem is the reduction of noise parts, whose spectrum superimposes with
the signal spectrum. This problem can be solved by using a frequency selective-adaptive filtering.
In the frequency selective-adaptive filtering the signal frequence components are determined by means of an autocorrelation
function and a Fourier-analysis. The noise parts are then filtered in the frequence domain by a spectral
analysis. The output signal is resulted from a signal reconstruction by means of a Fourier-series.
Therefore, uncorrelated noise parts are reduced by the adaptive filtering without deforming the signal