Fault detection of redundant systems based on B-spline neural network

The fault detection and isolation of redundant sensor systems based on B-spline neural networks is presented. The network is trained using an algorithm with an adaptive learning rate. To further save computation time, the residual vector is transformed from a multivariate B-spline function to a univariate B-spline function. The detection of abrupt and drifting faults using the proposed method is discusses. The performance of the proposed method is illustrated by an example involving a redundant system consisting of six sensors