A Novel Fault Detection Method Under Weighted Try-Once-Discard Scheduling Over Sensor Networks

Abstract

In this article, a distributed method based on set-membership filtering is developed to solve the fault detection problem over sensor networks, where each sensor consists of multiple sensing units. In order to prevent data collisions and reduce the energy cost, the weighted try-once-discard (WTOD) scheduling, which allows only a sensor unit to send its measurement data at each transmission instant, is considered to govern the disseminated information between the sensor units and a filter. The aim of the presented problem is to propose a novel fault detection method to determine whether there exist faults in discrete time-varying networked systems. To this end, in the presence of the WTOD protocol, two ellipsoids are obtained, which include all possible prediction states and measurement updated states. Then, the sensor fault occurs when there is no intersection between the prediction set and the update set. Furthermore, two recursive optimization methods are scheduled to look for minimal ellipsoids in order to improve fault detection performance. At last, the effectiveness and the applicability of our proposed fault detection method are demonstrated via a practical example.