Sensor Fault Detection Approach to Tensegrity Structures Using Markov Parameters

This paper presents a sensor fault detection method based on output error covariance and demonstrates its efficacy on tensegrity structures. An approximation model of the fault system is developed first using input and output signals. Subsequently, this fault system is compared with a reference system, and their output covariance is analyzed using the Markov parameters of both systems. In addition, an algorithm is presented to identify the fault sensor channels from the output error covariance. An examination of a tensegrity double prism tower, assuming fault sensors producing zero-mean Gaussian white noise, is conducted. The result validates the effectiveness of this approach in pinpointing the malfunctioning sensor channels. This proposed approach is adaptable to other structural applications of fault sensor identification.