Fault isolation in a winding-machine using RCE networks

The paper describes a mixed approach to the fault isolation on a winding machine. The first step is the generation of residuals, i.e. signals containing information about the faults. Based on a linear model of the winding machine, an observer based residual generation has been designed, however, measurements of the nonlinear system have been used for the calculation of the residuals. The second step is the residual evaluation and this is realised using Restricted Coloumb Energy Networks (RCE). The application of neural networks to this problems allows the isolation of different faults even if they are very small in comparison with the uncertainties. Note that the results achieved using the combined approach presented in this paper can not be achieved with known model based methods. It represents the modest contribution of this paper.