Handbook of air conditioning and refrigeration / Shan K

give a printout. A friendly dialog between the KBS and the user and a simple question-and-answer
format with a provided menu of possible answers are expected. An intelligent operating system
which accesses various application tools is recommended. Modular new application tools can be
plugged into the operating system as required. A user interface with built-in intelligence, graphically
based high-resolution display with standardized menus and format may be the preferable answer.
4. Knowledge acquisition. A knowledge acquisition module provides strategies to capture the
experts’ knowledge to develop a KBS. Sometimes, it also checks for consistency and completeness.
Knowledge acquisitions are usually accomplished through personal interviews with experts and review
of the application literature. Most HVAC&R-related knowledge systems have long-term plans
to continue the process of knowledge acquisition. A new trend is machine learning; i.e., the computer
learns from the experience how to capture and manipulate new knowledge.
Development of KBSs. Many KBSs are developed by using commercially available development
tools called shells. A shell consists of mainly a rule editor, a knowledge base, an inference engine,
and user interfaces. Usually, the inference engine and user interface are fully developed within a
shell. Thus, the user can focus on the collection and the input of the knowledge. He or she can
change the existing knowledge base and does not need to change the entire system. KBS is expected
to operate on a PC.
The performance of a KBS and the resources needed during development are highly affected by
its knowledge engineer. The knowledge base is built through the cooperation of the knowledge engineer
and experts in a specific domain. The knowledge engineer has the responsibility to choose an
appropriate inference strategy, and a suitable shell and to ensure compliance of the system with the
task.
The development cycle of a KBS is an iterative and incremental process. It begins with the initial
prototype. The next is an improved, and expanded, one. The process is repeated and may take years.
How the KBS Works. For example, if the space air is too humid and the space cooling load is
only one-half of the design load, it is required to find the general cause of these symptoms. The
problem is a diagnosis problem. Brothers and Cooney (1989) stated that if-then rules are composed
of parameters (symptom, set point) and values (too hot, too humid, correct). In the inference
engine, the computer program of the KBS will begin to search for an if-then rule through the
knowledge base that will give a value (too humid) for a general cause. If the values do not match
those required by the rule, the computer program will search for the next rule, until the following
if-then rule has been found:
IF symptom is too humid, and
cold supply air temperature set point is correct, and
cold supply air relative humidity is O.K., and
sensible cooling load is 50 percent of design load
THEN general cause is size of supply airflow rate CF 90
ENERGY MANAGEMENT AND CONTROL SYSTEMS 5.49
The cause of a too humid space air is that the volume flow of supply air is too great. The term CF is
the abbreviation of certainty factor. That means the confidence in the answer to the problem is 90
percent.
Testing, Verification, and Validation. In Jafar et al. (1991), testing often detects logical errors related
to the knowledge base, syntactic errors, and missing knowledge. Testing only shows errors
and does not explain their cause. Verification and validation should be performed at each stage of
development, to check the knowledge base for internal inconsistencies, mismatches, etc.
Applications. There are three primary areas of HVAC&R-related applications of KBS, as reported
in the paper by Hall and Deringer:
● Monitoring—interpretation of measured data in comparison to expected behavior