Advanced Signal Processing of Turbine Rotor Bore Waveforms

Whalen, M. F.; Mucciardi, Anthony N.

doi:10.1007/978-1-4684-4262-5_6

Cited by 2 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…UK computer-based methods for analysing ultrasoIlic data to distinguish between different defect types have been based on a variety of techniques such as adaptive learning [1]. artificial intelligence [2] and statistical pattern recognition [3].…”

Section: Introduction S F Burch Materials Physics and Metallurgy DImentioning

confidence: 99%

Objective Ultrasonic Characterization of Welding Defects Using Physically Based Pattern Recognition Techniques

Burch

1988

Review of Progress in Quantitative Nondestructive Evaluation

View full text Add to dashboard Cite

Section: Introduction S F Burch Materials Physics and Metallurgy DImentioning

confidence: 99%

Objective Ultrasonic Characterization of Welding Defects Using Physically Based Pattern Recognition Techniques

Burch

1988

Review of Progress in Quantitative Nondestructive Evaluation

View full text Add to dashboard Cite

“…This type of classification problem using ultrasonic waves is very suitable for employing the tools and techniques of artificial intelligence [1,2]. Adaptive learning methods, for example, have in the past been employed to train a flaw classification module so that it can distinguish between cracks and volumetric flaws [3]. Some of the limitations of this approach, however, have been due to the empirical nature of the features used for classification and the difficulty of understanding and adjusting the decision-making process when errors occur.…”

mentioning

confidence: 99%

Development of an Expert System for Ultrasonic Flaw Classification

Schmerr

Christensen

Nugen

1987

Review of Progress in Quantitative Nondestructive Evaluation

View full text Add to dashboard Cite

The complete characterization of a flaw requires information about the flaw type (crack, void, inclusion, etc.), flaw size, and orientation. Here we are only concerned with the determination of the flaw type so that the appropriate sizing algorithms can be chosen. This type of classification problem using ultrasonic waves is very suitable for employing the tools and techniques of artificial intelligence [1,2]. Adaptive learning methods, for example, have in the past been employed to train a flaw classification module so that it can distinguish between cracks and volumetric flaws [3]. Some of the limitations of this approach, however, have been due to the empirical nature of the features used for classification and the difficulty of understanding and adjusting the decision-making process when errors occur.In contrast, we have chosen to develop a classification scheme in the form of a rule-based expert system where the features used by the system for classification come from model-based fundamental knowledge, and where the rules are made explicit and modifiable. In this paper we will describe the nature of the expert flaw classification system we are building and demonstrate its use with some ultrasonic data. As currently constituted, the domain of knowledge of the system is highly constrained. The flaw classifier is concerned with distinguishing between single isolated volumetric and crack scatterers. The design of the system, however, is such that these constraints are not essential. SYSTEM OVERVIEWAs outlined in Fig. 1, the expert flaw classification system, FLEX (Flaw Expert), consists of essentially four major components: 1) a user-interface that allows the visual display and manipulation of ultrasonic data, 2) a set of tools that allow a user to manipulate and modify the rules of the system, 3) a module called FEAP (for FEAture Processing), and 4) a module called FLAP (for FLAw Processing). FEAP and FLAP are being designed as two separate, cooperating expert systems. 879Feature Processing (FEAP) It is the job of FEAP to take the preprocessed ultrasonic data from a given experiment, and determine confidence factors associated with each feature being used by the system. These confidence factors are numbers in the range [-1,1], where -1 indicates complete certainty that a feature is not present, 1 indicates complete certainty that a feature is present, and numbers in between indicate the degree of certainty or uncertainty (see Appendix). Both FEAP and FLAP manipulate these confidence factors according to the conventions and methods developed by Shortliffe and Buchanan for the MYCIN project [4]. FEAP also determines the percentage of the ultrasonic data sets, if there are more than one, in which there is positive evidence (positive confidence factors) for each feature. Currently, FEAP assumes that the data it uses has had non-flaw dependencies removed through the application of the measurement model of Thompson and Gray [5]. This preprocessing is done so that we can rationally evaluate features characteristic...

show abstract

Advanced Signal Processing of Turbine Rotor Bore Waveforms

Cited by 2 publications

References 0 publications

Objective Ultrasonic Characterization of Welding Defects Using Physically Based Pattern Recognition Techniques

Objective Ultrasonic Characterization of Welding Defects Using Physically Based Pattern Recognition Techniques

Development of an Expert System for Ultrasonic Flaw Classification

Contact Info

Product

Resources

About