Shape Feature and RBF Network Based Intelligent Fault Identification of Rotating Machinery

The shape of shaft orbit reflects the working state of rotating machinery. It plays an important role in the fault identification of water turbine generator set. This paper is mainly focused on using chain codes technique and Radial basis function (RBF) network to perform intelligent identification of different shaft orbit generated by different fault. Chain code is a contour-based representation for shaft orbit. It has properties of simple calculation, low storage requirement and translation invariant. And it is used as the feature of shaft orbit. In succession, the features are input to RBF network to identify the shaft orbit for fault identification. The experimental result indicates the proposed approach is very effective and has satisfactory accuracy.