Frontiers in Built Environment (Sep 2020)
Optimal Static Load Compensation With Fault Tolerance in Nonlinear Adaptive Structures Under Input and State Constraints
Abstract
Adaptive structures are conventional truss structures that are equipped with sensors, actuators, and a control unit. This offers the opportunity of reacting and adapting to external loads but raises nontrivial issues. When actuators are placed optimally within a structure, they can be individually integrated either parallel to or in series with elements of the original passive structure. Additionally, some of the elements might be tension-only elements and thus have to be treated as nonlinear, as their stiffness depends on the stress within the element itself. Input constraints naturally arise for actuators, e. g., due to the maximum pressure limit of a hydraulic system and displacement limits of the actuators. We present modeling approaches for an add-on inclusion of these different types of actuators in an existing finite-element model of a passive structure. We place special focus on the ability of the model to reproduce the correct behavior in case of an actuator reaching its displacement constraint within a tension-only element. When such an adaptive structure is subject to static loads, e. g., wind loads, it is required to respond using its actuators to keep the structure within given safety and comfort limits. These limits can be expressed as state constraints. We present a method for optimally compensating these static loads under the given input and state constraints along with experimental results on a scale model of an actual high-rise building. An important aspect regarding adaptive structures is that of their behavior in case of actuator faults. An obvious result is that a structure's performance degrades, and the controller needs to recognize faults and deal with it properly. Assuming a diagnosed actuator fault, we present results illustrating the performance degradation. The designed controller can reconfigure and reinitialize itself. The performance with and without applied reconfiguration to the nominal case is compared.
Keywords
