Methods for SHM System Capability Evaluation from Engineer's Perspective

摘要

Usually, the capability of Structural Health Monitoring (SHM) systems using Ultrasonic Guided Waves (UGW) is evaluated based on the Probability of Detection (PoD) curves. These determine the detectability of structural damage (e.g. a fatigue crack) of various size, just like in in other NDT testing methods. Such approach, although rigorous, provides fairly complex procedure to engineers in practice and is hard to follow. In this contribution, an approach more connected with engineering practice, is presented. It should be however noted, that although more practically focused, the presented method does not mean the design engineer would avoid using a sample structure to simulate the structural damage and transfer it, by technical similarity, on the real product. First, a traditional approach of PoD evaluation, is described and analyzed. An approach for evaluation of the PoD of individual damage size is presented- evaluation of a batch of repeated measurements enable the user to determine the variation of measured signal in terms of Difference index (DI) used (further referred to as the self-DI). Using the Central limit theorem enables then to determine not only the mean value of the self-DI, but also its variation regardless of its real distribution. Then, after imposing damage to the structure and performing the UGW measurement again, the user obtains the signal difference with respect to the first measurement with no damage (further referred to mutual-DI; additionally we obtain the information of the measurement system stability by comparing the first and second measurement self-DI mean and variation). The variations of the self-DIs and the mutual-DIs enable then to determine the probability of detecting the signal difference and thereby, the structural damage. Practical evaluation of repeated measurement on a carbon composite plate is presented and the relation of number of repeated measurements to the detection sensitivity is discussed. In the second part, a practically focused approach is presented. A new value, called Algorithm Success Rate (ASR), is presented. This value is directly linked to the particular algorithm, which the user selected for detection of the structural damage- in our case so called "Multipath" algorithm, which covers effects of both damage and environmental conditions. The ASR is obtained by comparing data from measurement on undamaged structure with the data on structure with damage. The ASR value, although not directly equal to POD, gives the user in engineering practice the measure of detection reliability, and therefore, in practice it is equivalent. Practical evaluation of ASR, measured on carbon composite plate, is presented for various temperatures and UGW measurement settings, like excitation voltage and frequency.