LAUSR

Abstract Adaptivity has become a promising concept in civil engineering to improve the load-bearing behavior of buildings and to reduce material consumption. However, the integrated actuators and sensors increase the complexity and adversely affect the reliability making a fault diagnosis for actuator and sensor faults necessary. In this paper, the distributed fault diagnosis of an adaptive high-rise truss structure, which is characterized by a modular design, is investigated. Based on local models and the local measurement information of hydraulic actuators and strain gages, a distributed fault diagnosis scheme is proposed for the diagnosis of actuator and sensor faults. Since the local models do not have information about the interconnection to other modules, the model-based residual is uncertain and faults in the other module can affect the local residual. For this reason, an effective online estimation of the probability density function and the Kullback-Leibler divergence of the residual is presented for change detection considering the stochastic uncertainties. Moreover, the selected sensor layout of the adaptive structure allows the isolation of the investigated actuator faults such that fault propagation paths of the distributed system are analyzed and sensor faults are isolated by communicating the detected changes in the modules. The effectiveness of the approach is illustrated in a simulation study.