Aspects of diagnostic rules for manufacturing systems: generation, generalization and reduction

The diagnostic step of the error recovery process in manufacturing systems is considered, and theories for evaluating the redundancy of failure knowledge in such systems are developed. The work is based on a diagnostic knowledge structure called the action-error-feature data set. Previously derived theories and a simple computational operation, called the covering analysis, for determining and testing N-multiple-error diagnosability of the data set are further developed to improve the efficiency of the diagnostic steps. Methods for evaluating and deleting redundant features in the data set are obtained, resulting in a qualitative measure for the embedded diagnostic knowledge in terms of the ability to diagnose the number of multiple simultaneous errors and a quantitative measure in terms of the redundant features in the system. Diagnostic rules for a specific error are generated, generalized, and encoded soon after the error is diagnosed at the first occurrence. The generalized rules can be applied in similar cases for a specific error and improve efficiency