Measurement for monitoring, probing and recording evidence - Tunnel Engineering: Basics and Additional Services for Design and Construction - page 142

Systematic data evaluation. Due to the complex time-dependent interaction between
machine, support medium, annular gap grouting, segment lining, ground and the incon-
sistent recording and saving of the data, there is still no standard procedure for the evalu-
ation and analysis of the tunnelling data. The enormous amount of data therefore makes it
very likely that essential parameters and relationships will fail to be recognised and much
information will be lost.
The aim of systematic data evaluation must be to analyse the arriving data as the tunnel
advances in order to increase the quality and quantity of the information gained from the
data and make it available to all parties in real time. The specialists on site should be sup-
ported by computer-assisted automatic evaluations, with the displayed results being avail-
able to all project parties. This can include statements about machine condition concerning
wear and damage and the effects on the surroundings with guidelines for the adjustment of
the machine to suit the prevailing geology and hydrology. This enables the proportion of
avoidable settlement to be minimised and the performance and availability of the machine
to be optimised.
Working from the already described structured recording and archiving of data in a data-
base system, there are various processes for computer-based evaluation, which are now
briefly described. For a detailed examination of this subject, reference is made to [151].
Classic statistics. For the purpose of statistical evaluation of tunnelling data, the known
processes of multi-dimensional regression and correlation analysis should be mentioned
first. Correlation analysis serves to obtain a quantitative magnitude for a relationship, in
other words to determine the degree of relationship. Regression analysis is a process to
detect relationships between various variables. The mathematical derivation is not given
here and reference should be made to the relevant specialist literature (for example [88])
for further details. In the following, only the possible applications of data analysis for a
shield drive will be presented.
Unusual deviations or fluctuations, which could easily be missed in the quantity of data,
can be detected immediately from trend analyses and forecasts of characteristic initial
values, like the relationship of specific energy consumption to penetration rate [97] or to
the settlements produced [165], and can be used to trigger automatic alarms (example: col-
lapse warning [145a]). Presuming that the system behaviour follows reproducible rules,
information about the effect of adjusting individual parameters can also be used to opti-
mise the process in the next time step. For example, the relationships between advance and
settlements could be analysed in sections and the settings on the shield machine adjusted
so that the configured ideal values are not exceeded [151].
Fuzzy logic makes it possible to take into account both engineering know-how (expert
knowledge) and also intuition in data analysis and process control. The excellent compat-
ibility and integration capabilities in standardised control and SPS or PLC (Programmable
Logic Controller) systems have proved particularly useful for shield tunnelling applications.
In contrast to conventional methods, fuzzy logic is not based on the processing of sharply
or discretely defined values, but on the theory of unsharp or fuzzy quantities, a recognised
discipline in Mathematics. The basic principle is the description of a condition or a situa-
tion through linguistic variables instead of numbers and formulae. Rules are not described
by elaborate mathematical models but by simple rule-based conclusions related to human