The idea is then that deviations from the expected behavior can be used to detect
malfunctions in these aircraft components in real time of operation. This principle
can easily be applied to other systems such as a computing system. In this case, the
malfunctions are not detected in aircraft components, but hardware components.
Applied to a software system, the “components” are tasks, which communicate
with each other and thus depend on each other and also depend on the hardware
they use. In fact, MASS and PASS could be considered as an extension of
system-level generalization of GAFT.
The questions that arise are the following:
– How can the component that is responsible for a deviation be found?
– How can the fault type (permanent fault or malfunction) be determined?
True, the location of faults and faulty element(s) from the set of their manifestations
is one of the fundamental problems of complex technical systems [2].
The complexity of the problem can be illustrated by the fact that literature often
uses simplifying assumptions that eases the analysis of the problem and allows to
introduce solutions which fit the assumptions but unfortunately not the real world.
Solutions to the fault localization problem, for example, are in literature based on
very strong assumptions about the system and a priori information about the set of
possible faults and the fault modes [3]. Often faults are considered as so-called
simple faults, i.e., single non-repeatable ones or mutually independent ones.
However, the miniaturization and imperfection in current technologies and
manufacturing processes force us to face the situation where multiple faults of
elements must be considered if we are willing to make practically useful solutions,
based on type of fault detection and further malfunction analysis and toleration.
The interdependence between impacts is reflected in the matrix of mutual
dependence, the so-called Dependency Matrix that uses a directed graph to represent this information.
The alternative way to react to the object’s condition, composed of the condition
of all its elements, is defined in the Recovery Matrix. When a particular fault
occurs, the Recovery Matrix allows to analyze and describe “what the system is
going to do.” There is a close relationship between the nodes of the Dependency
Matrix and those of the Recovery Matrix.
Figure 5.1 illustrates how the three features behave in relation to each other.
Each particular condition of the node might in some way be related to the collected
parameters. In turn, a trend can be detected if the data of several nodes are taken
together [1].
5.3 Dependency Matrix
The Dependency Matrix D defines dependencies between elements of the system.
A node of the dependency graph corresponding to its dependency matrix represents
every element.
50
5 GAFT Generalization: A Principle and Model of Active System…
malfunctions in these aircraft components in real time of operation. This principle
can easily be applied to other systems such as a computing system. In this case, the
malfunctions are not detected in aircraft components, but hardware components.
Applied to a software system, the “components” are tasks, which communicate
with each other and thus depend on each other and also depend on the hardware
they use. In fact, MASS and PASS could be considered as an extension of
system-level generalization of GAFT.
The questions that arise are the following:
– How can the component that is responsible for a deviation be found?
– How can the fault type (permanent fault or malfunction) be determined?
True, the location of faults and faulty element(s) from the set of their manifestations
is one of the fundamental problems of complex technical systems [2].
The complexity of the problem can be illustrated by the fact that literature often
uses simplifying assumptions that eases the analysis of the problem and allows to
introduce solutions which fit the assumptions but unfortunately not the real world.
Solutions to the fault localization problem, for example, are in literature based on
very strong assumptions about the system and a priori information about the set of
possible faults and the fault modes [3]. Often faults are considered as so-called
simple faults, i.e., single non-repeatable ones or mutually independent ones.
However, the miniaturization and imperfection in current technologies and
manufacturing processes force us to face the situation where multiple faults of
elements must be considered if we are willing to make practically useful solutions,
based on type of fault detection and further malfunction analysis and toleration.
The interdependence between impacts is reflected in the matrix of mutual
dependence, the so-called Dependency Matrix that uses a directed graph to represent this information.
The alternative way to react to the object’s condition, composed of the condition
of all its elements, is defined in the Recovery Matrix. When a particular fault
occurs, the Recovery Matrix allows to analyze and describe “what the system is
going to do.” There is a close relationship between the nodes of the Dependency
Matrix and those of the Recovery Matrix.
Figure 5.1 illustrates how the three features behave in relation to each other.
Each particular condition of the node might in some way be related to the collected
parameters. In turn, a trend can be detected if the data of several nodes are taken
together [1].
5.3 Dependency Matrix
The Dependency Matrix D defines dependencies between elements of the system.
A node of the dependency graph corresponding to its dependency matrix represents
every element.
50
5 GAFT Generalization: A Principle and Model of Active System…