V. Conclusion and Concluding Remarks
(1) Experiments prove that a posteriori compensation of
systematic errors is a method easy to use and with better
compensation effect, when signal-to-noise ratio 2, more
than 60% systematic errors can be generally eliminated by
using this method, the accuracy could be improved by 1.7
times above.
(2) From experiments of effect on a posteriori compensa-
tion of various systematic errors and those of different in-
terpolation methods, some conclusions can be drawn, especially
the relationship between benefit of a posteriori compensation
and signal-to-noise ratio demonstrates that during the pro-
ductive process random error ought: to be reduced as little as
possible and signal-to-noise ratio should be raised .
(3) During the process of a posteriori compensation, it
would be of great importance to assess the effect of inter-
nolation method on compensation of systematic errors. In our :
experiments,it is evident that effect on compensation of poly-
nomial interpolation is better than that of weighted mean
value method.
The disadvantage of a posteriori compensation is: during the
computation process it is, in general, necessary to add 2-5
iterations for the whole adjustment, thus much computation
time is needed. :
Table 3. Statistical table of signal-to-noise ratio and
average benefit of a posteriori compensation
random |signal-to- : (1) fined (3)
refined prog. refine rog.(2
error noise ratio pres prog
4m ^ j Gol 1 dep. Yigy Oz CoL Ÿ coop I c» 1] o:
MD Cor C, Cs 7 / / / " ” v d
« Out Top Op Cz CoL Cop P Cz
15 1,68 | 0.635 | 3.08 | 0.80 | 1.81 1.10 | 2.60 1.16 1.86 1.33 | 2,85 (41.2;
* 10 2.70 | 1.22! 5.89 | 1.88 | 2.55 1.88 | 4.38 1.12. | 2.5 1.42 | 4.66 | 1.87
5 5.33 | 2.69 iio 4.35 | 3.88 1.95 | 5.64 | 2.39 4.19 | 2.04 | 6.82 | 2.08
15 0.76 | 0.30 | 0.78 | 0.42 | 5.83 | 0.42. | O.85 | 0.56
** 10 „40:81 | 0.63 | 0.84 { 0.60 | 0.82] 0.53. ! -0.86 10.71
119 0.82 | 0.60 | 0.84 | 0.62 | 0.84 | 0.64 | 0.58 | 0.72
* improvement of rate of multiple of mse.
** percentage of eleminated systematic errors
-.5010*
