Sensor fault detection by sparsity optimization

Sensor faults in control systems could cause persistent damage to system components. Due to the severity of such occurrences, sensor fault detection is crucial in control systems. Its relevance and significance are deeply embedded across countless engineering fields. Sensor fault detection is one area that is immensely researched on before the sensor values can be relied upon for system configuration. In this paper, a statistical approach is proposed for automatic sensor fault detection. By assuming the sensor fault to be an additive term, the problem of sensor fault detection has been modeled as a least-squares optimization problem and an L1 penalty is introduced to control the number of biased sensors. As a result, the proposed method can accurately detect the biased sensors by tuning the amount of penalty and the problem is further addressed by selecting the proper regularization parameter in an automatic manner via the BINCO method. Overall, the experimental results have shown that the proposed method is indeed capable of detecting sensor faults despite the presence of random noise level.Bachelor of Engineerin