Contribution to Incomplete and Noisy Information Problem Solving by Artificial Intelligence Principles Applying

Engineering or manufacturing domain knowledge is rarely in the form required by these systems. Furthermore, we have to solve a lack of information, presence of incomplete and noisy knowledge about solved complex diagnostic system from reliability, optimal and predictive manufacturing point of view. Decision support purposes require for the knowledge provider to know about primary "cause-effect" relationships but not be in a position to assert that other relationships are nonexistent. The use of maximum entropy inference in reasoning with uncertain information is commonly justified by an information-theoretic argument. This contribution deals with a possible objection to this information-theoretic justification, presents a probabilistic reasoning methodology and a maximum entropy application, which can estimate missing information by making some sort of global assumption and provide advice based on the knowledge available. Some diagnostic problems could be operated in a chaotic fashion. Achieved results showing how the fractal dimension and entropy describing the chaotic motion depend on the operating characteristics of the device. Diagnostic frequency waves could also lead to chaotic fluctuations in the time evolution of the transmitted intensities.