Subproject A2 Multiscale Geometry Measurement

The results of a first diagnosis for worn turbine components is the basis for decisions in the regeneration process and therefore crucial for the process itself. In order to repair such components the degree of damaging has to be estimated. For that, the current geometrical state as well as microscopic structures and the surface roughness must be known in detail.

Currently, the decision making of the previously mentioned regeneration processes are based on human visual inspections and experiences. In order to fully automate the regeneration process, a suitable, automated measurement routine has to be developed. Since conventional single-sensor-systems do not offer the possibility to cover the spectrum of required scales, the main objective of sub project A2 was to develop a robot based, non-destructive multiscale sensor system. For an accurate positioning, the measurement system was attached to an industrial robot with six degrees of freedom.

After a successful measurement, the acquired data from all used sensors can be fused into a single dataset, providing a holistic model of the measurement object. That model can then be examined for features like abrasions and damages which are furthermore classified to make a valid assertion about the components current state.

Gathered data is used by other sub-projects to run simulations or plan tool paths for further processing steps.

The pose of the robot endeffector with the attached multi sensor system is determined by a high accurate laser tracker, which later connects the individual measurements with each other. Each sensor and its corresponding measurement principle is shown in the following order: Fringeprojection and the resulting stitched 3D-Mesh of a worn turbine blade, the edge sensor with poly-view optics, the BRDF-sensor for an approximation of the bidirectional reflectance distribution function and a measurement of the surface topographies using the low coherence interferometer with an extension arm measuring a turbine blisk.

Current research is focused on the fusion of measurement data of individual modularities. Based on the acquired data, sets of damages are to be detected and classified. The extracted features are then combined to a single data base which will be provided to following sub projects in the regeneration path.

In order to automate the measurement process, another topic of interest is an adaptive, intelligent path planning which leads to the possibility to measure prior unknown geometries. Furthermore intelligent measurement poses are supposed to reduce measurement uncertainties which can occur during the process.