Methodologies of Damage Identification Using Non-Linear Data-Driven Modelling

Manufacturers are required to design and construct safe, ecological and reliable structures, which are still cost effective. In order to guarantee these requirements, especially in industries where the component reliability is crucial, regular inspection intervals must be defined. Common techniques are visual inspections, magnetic particle testing, dye penetrant testing, eddy current inspection, radiography, infrared/thermal testing and standard ultrasonics. As traditional inspection techniques can be very expensive in terms of both man hours and structure down-time, the development of suitable automatic and reliable monitoring methods, which can be used on demand, would be very valuable. This requires techniques which can monitor the given structure either continuously or in fixed intervals and can provide suitable early warning before a propagating damage reaches the limits of criticality. Here is the place where structural health monitoring systems enter into play. To perform these tasks, a monitoring system should decide autonomously whether the host structure is damaged or not. On that account, this article describes the theoretical background and methodology of a novel data driven approach based on scale-frequency analysis, multiway hierarchical nonlinear principal component analysis (h-NLPCA), squared prediction error statistic (SPE) and self-organizing maps (SOM) for the detection and classification of damage in structures. The application of this approach is described in detail in chapter “Damage Identification using Non-linear Data-Driven Modelling – Application.” This article contains the theory of the proposed method and is organized as follows. First, the theoretical background introducing the basic concepts of structural health monitoring and the evaluated signal processing techniques are presented. Afterwards the proposed methodology is described in detail including the application of the introduced signal processing techniques. Finally, concluding remarks are given in the last section. Miguel Angel Torres Arredondo MAN Diesel and Turbo SE, Germany