Exploring the limits of the Truncated SPRT method for vibration-response-only damage diagnosis in a lab-scale wind turbine jacket foundation structure

Abstract

The performance limits of the truncated Sequential Probability Ratio Test (SPRT) method for damage diagnosis via vibration-response-only signals are explored. This is accomplished by considering damage detection and identification (classification) in a laboratory-scale wind turbine jacket foundation structure subject to damage scenarios ranging from minor to mild in terms of their effects on the dynamics. Low and limited frequency bandwidth (~ 3.5—200 Hz) signals are employed in order to better reflect actual excitation conditions. As an additional challenge, only a single vibration signal, measured via laser vibrometer or accelerometer, is employed. The results indicate moderate damage diagnosis performance for the minor damage scenarios, but good for mild ones. Expectedly, the overall performance is similar to that of the residual variance based method, but inferior to that of the model parameter based method.