Model-based Surface Defect Detection and Con dition Monitoring in Wire Ropes

Wire ropes are exposed to huge external powers every day. Unfortunately, this can lead to structural anomalies or even defects in the rope formation. A defective rope bears a high risk for human life. This motivates the strict rules summarized in the European norm [1], which instruct a regular inspection of wire ropes. Rec ent approaches to defect detection in surfaces or fabrics mostly are limited to texture analysis [2]. Usually not enough defective samples are available in a dvance. Therefore texture-based approaches are often combined with oneclass classification, also called novelty detection [3]. In [4] anomaly detection in wir e rope data was performed using linear prediction as feature extraction combined with a multi-channel one-class classification strategy. However, all these approaches are not able to detect structural changes in the rope structure. Thus, we propose a new approach for model-based detection of surface defects and structural deviations from the normal rope geometry. A wire rope is composed of strands which itself are composed out of single wires. Every wire can be described by a helix and a mathematical description of the rope geometry can be given by the space curve of the j-th wire in the i-th strand: @