Fault Detection Based on Mean-Shift Clustering and Immune Danger Theory
Open AccessModern control systems rely on a complex network of interacting sub-systems. Because a failure in any of these sub-systems could have catastrophic consequences, it is necessary to detect and isolate faults in control system components, i.e. actuators or sensors before the fault is allowed to cause a system failure. Early detection and isolation could enable timely system reconfiguration and increase safety of operations. In the literature, there are numerous fault detection and isolation techniques presented. The central challenge with fault detection is determining the difference between normal and potentially harmful activities in the system. This thesis demonstrates the ability of one technique, based on mean-shift clustering and immune danger theory, to detect and isolate faults.The simulations and experiments for multiple sensor systems have confirmed the ability of the new approach for online fusing and fault detection. The hybrid system provides fault tolerance by handling different problems such as noisy sensor signals and multiple faulty sensors. This makes the hybrid approach attractive for solving such fusion and fault detection problems during real-time operations.The contributions of this thesis are summarized as follows:• The thesis proved that the Mean-Shift (MS) algorithm with isolated stationary points generates a convergent sequence. This thesis also studied special one- dimensional case and showed that in this case the MS algorithm generates a monotone and convergent sequence with both analytic and non-analytic kernels. Furthermore, a slightly modified version of the MS algorithm was proposed in order to improve the performance of the generated mode estimate sequence.• The effectiveness of the modified MS algorithm was examined on three data sets - closed, sparse and two cluster.• A fusion algorithm using MS and Fuzzy-C means (FCM) was presented.• A fault detection algorithm was generated from the immune danger theory. • Combining the aforementioned algorithms produced an efficient, hybrid approach for multiple sensor data fusion and fault detection.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.