In substance, decision trees could be built before repair actions through reliability
analysis or through reverse-engineering analysis. In this case, they are being used in the
context of a rule-based expert system. They also could be built-in to the call handling
process to construct actual decision trees based upon completed and closed out call data,
in the context of a self-organizing approach. Recent developments in the design and application
of self-organizing systems also have focused on parallel and sequential processing,
utilizing a combination of logic, hypertext, and parallel processing to achieve the same
solution as utilized in the post priori decision tree-approach described above.
A third major area of technology applied in diagnostics is to make use of newly developed
intelligent (data-) retrieval systems. These approaches attempt to use high-speed
computer technology to search existing files for case-based situations that are similar in
nature to the diagnostics being executed or to make use of neural networks or use very
high-speed hypertext search.
All three approaches attempt to make use of historically collected data to present the
correct and appropriate information to the repair provider, either centrally or in the field.
In summary, the current state of the art involves two general types of structure (i.e.,
retrieval and diagnostic advisory systems), and two types of explicit diagnostic systems
(i.e., expert systems and self-organizing systems).
Expert systems are the other category of advisory systems. Expert systems provide an
architecture for recording and using the experience of “expert” human service engineers:
● Rules-based expert systems are produced by recording and linking the heuristic decisions
that are made by the human mind when problem-solving or troubleshooting a unit or system’s
fault, based on “a priori” or before-the-failure analysis. Literally, this approach documents
the minute steps that a service technician must go through to remedy a fault.
Typically these systems are structured to work on a forward-chaining or a backward-chaining
logic depending upon the specific problem for which the system was designed to solve
● Model-based expert systems are also developed on an “a priori” basis from a database
of the system’s or unit’s calculated or anticipated failure structure and behavior. This
format makes the model-based approach better suited for the fault analysis of new
equipment that service personnel might not be familiar with; methods to add historical
information as it is developed are desirable in a model-based expert system.
Within this diagnostic state of the art, the basic process model for call resolution and fault
diagnostics and isolation, as shown in Figures 99-6 and 99-7, follow these procedural steps:
● Gather and collect information to determine the symptoms and situation in which the
problem appears
● Form a hypothesis to select a fault that is most likely to have caused the observed
symptoms
● Select a goal or set of goals to resolve a particular call derived from the hypothesis
that would have been previously developed this step usually takes the form of “determine whether element x is faulty”,” by selecting specific parts out of all
the parts to blame, making the process much more manageable and leading to a faster
resolution