Condition monitoring of rotating machinery exploiting encoder and video measurements: development of novel encoder-based gear fault indicators by exploiting direct fault measurements as a ground truth

Condition monitoring of rotating machinery gains importance in a number of industrial applications including wind turbines and automotive, in order to optimally schedule maintenance and to guarantee operation safety and production efficiency. Diagnostic indicators serve as tools to evaluate the condition of a rotating component and can be subdivided into two types. Indirect diagnostic indicators are based on signal processing and machine learning techniques applied on indirect measurements such as acceleration, sound or rotational speed. Direct diagnostic indicators can be defined as based on direct damage measurements by using images. Online monitoring systems exist using indirect diagnostic indicators, but these are rarely compared with direct damage measurements. Therefore, this proposal aims to develop a standalone direct monitoring (vision) system which is able to detect defects in rotating machinery at operational conditions, in order to serve as a ground truth for novel indirect monitoring techniques, with a focus on encoder measurements and pitting on gears. Firstly, a design for an online vision system is developed which should be robust against the harsh conditions in a gearbox complicating the visual access to the gear teeth. Secondly, novel encoder-based diagnostic indicators for gear monitoring are investigated and developed. The third step consists of integrating the two systems and methods on in-house test rigs with industrially relevant gearboxes.