Adaptive, Intelligent Methods for Real TimeStructural Control and Health Monitoring

[摘要] By framing the structural health monitoring and control problem as being oneof enhancing structural system intelligence, novel solutions can be achieved throughapplications of computational strategies that mimic human learning and attemptto replicate human response to sensory feedback. This thesis proposes several newmethods which promote adaptive, intelligent decision making by structural systemsrelying on sensory feedback and actuator compensation. Four significant contributionscan be found in this thesis study. The first method employs an adaptable subclass ofArtificial Neural Networks (ANNs), called Radial Basis Function Networks (RBFNs)for robust control in the presence of sensory failure. The second method exploitsthis computationally efficient network to detect and isolate system faults in real time.The third algorithm utilizes an RBFN to effectively linearize the nonlinear actuatordynamics of a Magnetorheological (MR) damper, thereby improving control of thesemiactive device. Lastly, an open loop observer is implemented experimentally toboth detect damage and act as a trigger for control of the newly developed AdaptiveLength Pendulum-Smart Tuned Mass Damper (ALP-STMD).Some limitation of existing algorithms in the field of real time structural healthmonitoring and control are that they rely heavily on fixed parameter methods, assumestandard linear time invariant assumptions, or mandate accurate modeling of systemdynamics. By embedding the proposed reasoning and decision making algorithms intothe feedback methodology and design, greater generalization and system adaptivityis possible. Specifically, the proposed methods develop novel solutions for adaptiveneural control, fault (sensor failure) tolerant control, real time damage detection,adaptive dynamic inversion, and control applications for STMDs.The neural network adaptive control formulation is successful in rejecting firstmode disturbances despite online sensor failure. It is also capable of improving theperformance of a baseline Hoc controller in the presence of sensor failure and earthquakeground motion. The proposed fault tolerant controller is validated on a twodegree of freedom shear frame subjected to six earthquake records. Furthermore, thisapplication involves the use of piezoelectric patches as sensors and actuators.The RBFN algorithm in combination with an open loop observer is capable of bothdetecting and isolating stiffness degradation and recovery in multi-degree of freedomsystems in real time. The method is validated on experimental data taken from onlinedamage tests using the Semi-Active Independent Variable Stiffness (SAIVS) device.Other validations involve simulations on a two degree of freedom system and a tendegree of freedom system with both independent and coupled damage case scenarios.In all scenarios, the RBFN is capable of identifying the length of time and degree offreedom in which stiffness variation occurred.A neural network formulation is developed to perform dynamic inversion for semiactivecontrol of an MR damper. The MR damper acts as a base isolator in a scaledtwo story building. Both the building and damper models were based on tests performedat Rice University. The control performance of the adaptive RBFN dynamicinversion method is compared to both passive-off and passive-on methods of semiactivecontrol for MR dampers.The last contribution serves to combine both real time structural health monitoringand control in a proof of concept experimental study. An open loop observer isused to trigger an ALP -STMD device in the presence of base excitation and stiffnessdamage. The stiffness damage is generated from strategically regulating the currentapplied to Shape Memory Alloy (SMA) braces in a two degree of freedom shear frame.Once damage exceeds a predefined threshold, the ALP-STMD uses a another SMAto adjust its pendulum length to tune in real time to the dominant pulse present inthe base excitation.

[发布机构] Rice University

[效力级别] engineering