1. at an advisory level, and
2. at a level capable of interpretation of data.

Types of uncertainty that can be present in a body of knowledge could be large. They can be classified as follows: Variation in the degree of [a] belief, [b] likelihood, [c] possibility, [d] precision, [e] truth of a proposition, [f] provability of a proposition, [g] compulsion, [h] relevance, etc. In the knowledge pertaining to the natural world, such as in NDT&E, the sources of uncertainty could be unreliable sources of data and information, abundance of irrelevant data, lack of understanding, faulty sensory equipment, conflicting sources of facts, hidden variables producing apparent randomness, etc.

A number of ways to manage uncertainty have been proposed. Among the many methodologies to manage uncertainty, the choice of fuzzy logic, seems appropriate in NDT&E, due to the following reasons: [a] It has been established that in nearly every practical application in knowledge engineering, the approach of fuzzy logic holds promise to be the appropriate one. [b] It can be shown that fuzzy measures incorporate and generalise, various other theories of uncertainty. [c] Fuzzy logic is good at handling uncertainties, that cannot be expressed analytically, (e. g. "45 angle probe is a better choice for this NDT&E application"), i. e., linguistic uncertainties. From the nature of NDT&E knowledge, it is known that linguistic quantifies are abundant.

In this paper, the use of fuzzy logic in a knowledge-based system for ultrasonic testing of austenitic stainless steels is described. Parameters that may contain uncertain values are identified. Methodologies to handle uncertainty in these parameters using fuzzy logic are detailed. The overall improvement in the performance of the knowledge-based system after incorporating fuzzy logic is discussed. Extension of this methodology for other KBS for NDT&E is highlighted.