The second block must contain algorithms for the construction and analysis of all control
charts that are incorporated into the first block of the expert system. Some of them (classical
Shewhart control charts, classical EWMA charts, CUSUM charts, Hotelling chart) are
contained in many statistical software packages – some such statistical program could be a
part of the operational system of the proposed expert system. But the algorithms for the
construction and analysis of other control charts mentioned in Table 4. must be created and
incorporated into the expert system.
Another part of the knowledge base for the second block of the proposed expert system are
the tests of the process stability. It is suitable to divide this testing into two parts:
1. Testing in classical Shewhart control charts for sample mean and range
- for testing stability using these control charts the expert system should contain all
developed tests of non-random patterns .
2. Testing of all other control charts incorporated into the expert system
- for all other charts we recommend to use only basic test (identification of the point
behind the control limits) because of the lack of sufficient theoretical background.
Very important part of the knowledge base that the identification block needs for its
operating is the part which contains special knowledge and expert experience with the
controlled process:
a) Results of the team analysis of the process variation (using for instance Ishikawa chart,
Pareto analysis) that give an initial summary of possible general and root assignable
causes of the process variation and possible probabilities of their occurring.
b) Results of the history of the process – identified non-random patterns in the old control
charts, records about all changes of the process, records of all really identified assignable
causes, records of all realised actions.
The second block must contain algorithms for the construction and analysis of all controlcharts that are incorporated into the first block of the expert system. Some of them (classicalShewhart control charts, classical EWMA charts, CUSUM charts, Hotelling chart) arecontained in many statistical software packages – some such statistical program could be apart of the operational system of the proposed expert system. But the algorithms for theconstruction and analysis of other control charts mentioned in Table 4. must be created andincorporated into the expert system.Another part of the knowledge base for the second block of the proposed expert system arethe tests of the process stability. It is suitable to divide this testing into two parts:1. Testing in classical Shewhart control charts for sample mean and range- for testing stability using these control charts the expert system should contain alldeveloped tests of non-random patterns .2. Testing of all other control charts incorporated into the expert system- for all other charts we recommend to use only basic test (identification of the pointbehind the control limits) because of the lack of sufficient theoretical background.Very important part of the knowledge base that the identification block needs for itsoperating is the part which contains special knowledge and expert experience with thecontrolled process:a) Results of the team analysis of the process variation (using for instance Ishikawa chart,Pareto analysis) that give an initial summary of possible general and root assignable
causes of the process variation and possible probabilities of their occurring.
b) Results of the history of the process – identified non-random patterns in the old control
charts, records about all changes of the process, records of all really identified assignable
causes, records of all realised actions.
การแปล กรุณารอสักครู่..
