Filtern
Dokumenttyp
- Beitrag zu einem Sammelband (3) (entfernen)
Sprache
- Englisch (3)
Referierte Publikation
- nein (3)
Schlagworte
The detection of defects in aircraft components or parts of a power plant or mines in the ground is connected with the task to distinguish between signals caused be the item searched for and noisy signals from other sources. The reliability of the system is the better the better the system is in detecting the wanted signal and avoiding false alarms caused by noise. It is usual to measure the reliability of diagnostic systems in terms of ROC-curves (Re-ceiver Operating Characteristics) and POD-curves (Probability of Detec-tion). The ROC curves describe the POD versus Probability of False Alarm with growing system sensitivity along the curve. POD curves describe the Probability of Detection for a fixed sensitivity for a variation of item sizes or depths. Reliability of the whole system is always a composition of the physics
and technique, application influences and the human factors. For the scien-tific understanding and possibilities of improvement, decomposition accord-ing to a modular model is helpful. The authors present their experiences with the reliability evaluation of metal detectors for humanitarian demining and an NDT system of mechanized ultrasonic phased array testing of copper welds. The special concern in demining is the influence of soil types, detec-tors and the experiences of human beings.
The paper will give an overview of new methodology for evaluating the
reliability of NDE systems accurately, reliably and efficiently in accordance with the
specific requirements of industrial application. After a review of the substantive
issues from the previous workshops, the go forward guidance from these
interactions is considered.
At the beginning of the reliability analysis, the actual safety demands have to
be defined in order to fit the investigation to the level of risk when the component
would fail. Next, all the essential influencing parameters need to be documented and
transferred to an appropriate design of experiments (DOE) to determine the
reliability in terms of a qualitative assessment for lower risk or in terms of a
quantitative probability of detection (POD) or ROC (Receiver Operating
Characteristics) curves for higher safety demands. A new paradigm is offered to
consider the POD or reliability of the system as a function of the configuration of
input variables and use it for optimisation rather than for a final judgement. An
advantage for the end user is also to sample all single PODs to an integral 'Volume
POD' of a part.
Among the influencing parameters, the human factor is the most important one.
A systematic psychological approach shall help to find out where the bottlenecks are
but most importantly to provide best possible working conditions for the human
inspectors.