Measurement for monitoring, probing and recording evidence - Tunnel Engineering: Basics and Additional Services for Design and Construction - page 144

In order to evaluate the process, important information about the face and ground condi-
tions is required in addition to the measured values of support pressure, total thrust force,
pressing force on the cutting wheel and temperature. It should be noted for example that
in mixed face conditions (50 % rock; 50 % loose ground), only half of the discs are loaded,
although these may be loaded with twice the force under some conditions. Since reliable
advance probing systems are still in development at the moment, expert knowledge from
the contract documents in the form of standardised check lists is entered manually into the
DBMS for the description of the decisive ground parameters.
Seven relevant data information streams reach the DBMS. In the DBMS, the input data is
filtered, subjected to preliminary analysis and processed to the characteristic parameters of
boring class, temperature and cutter press force (averaged). The determination of a boring
class for the specific process requires an independent subordinate fuzzy logic system to
analyse the ground parameter data. Proven methods are already available in the literature
[261], which can be adapted for the specific conditions of the process and the project. In
order to determine the average cutter press force, the action and reaction relationships
have to be evaluated by the DBMS based on known deterministic relationships.
The fuzzy logic system finally performs a problem-oriented data analysis to determine the
ideal values for cutting wheel revolutions and oil volume flow difference. The characteris-
tic procedure always includes the processes of fuzzification, inference and defuzzification.
Fuzzification can be understood as transforming the given sharp values (in this case the
parameters temperature, cutter press force, boring class) and the ideal values (in this case
cutting wheel revolutions and oil volume change) into unsharp values. The degree of be-
longing of the individual values to the relevant unsharp quantity is described through
linguistic variables. These form a so-called fuzzy set for each parameter of a data set.
Inference. Extensive expert knowledge was already necessary to create the fuzzy set,
which now to be processed by inference into processing rules. According to the principle
of rule-based reasoning, one step is taken after another according to the principle IF ...
AND ... THEN, similar to the human use of symbols in thinking. The evaluation of the IF
... AND part is described as aggregation and the evaluation of the THEN part is described
as composition.
This rule matrix provides the core of the following data analysis. All subsequently incom-
ing data is evaluated solely on the basis of these rules, so that a high degree of care and
expert knowledge is demanded here.
A simple possible rule would be, for example:
Aggregation: IF cutter press force = low AND boring class = good rock
Composition: THEN oil volume flow = increase.
Precondition: temperature = low
The aggregation represents the combination of the individual conditions into a rule. Many
hundreds of fuzzy operators are now available, although however the Minimum/Maxi-
mum operator [6] is mostly used in practical applications. The Minimum operator in this
case corresponds to a simple AND, the Maximum operator a simple OR.