Development and implementation of advanced control methods for hybrid simulation.

摘要

Hybrid simulation is an effective way of testing structures that combines the benefits of a computational analysis and experimental testing techniques. Innovative structures consists of state-ofthe-art components and assemblages whose function as a system needs to be tested experimentally. Often times, these components and assemblages push the controller and other testing equipment to its limits. Performing hybrid simulation with the controller in displacement control mode does not always suffice. Force control, switch control and mixed control methods in hybrid simulation are explored in order to overcome these limitations and provide robust ways of performing hybrid simulation. Force control hybrid simulation is a type of hybrid simulation where the control system is in force control mode. Switch control hybrid simulation is another type of hybrid simulation where the control system switches between displacement control and force control modes. Switch control hybrid simulation is applicable with setups that have only one control degree-of-freedom. Mixed control hybrid simulation is an extension of switch control hybrid simulation where multiple control degrees-of-freedom are switching between control modes independently of each other. Force control, switch control and mixed control hybrid simulation methods are developed, tested and verified. The motivation for the development of these methods is discussed. New methods are presented and explained. New OpenFresco classes and Simulink/Stateflow models are developed to implement these methods. Then these three alternative control methods are tested using the mu-NEES experimental setup at the nees berkeley structural engineering lab using two different configurations. The first configuration is setup with one control degree-of-freedom which contains a relatively stiff specimen. The second configuration extends the first one with two control degrees-of-freedom. The performance of various time integration schemes with these methods are studied. These tests verify and validate the three alternative control methods as well as the new OpenFresco classes and Simulink/Stateflow models. The test results are evaluated to assess the performance of each control method by comparing them first to the numerical results, then to the displacement control results and finally to each other. The force control methods, in general, provides better results than the displacement control method for the second mu-NEES configuration. Switch control hybrid simulation methods provide better results than their force control counterpart for the one degree-of-freedom configuration but not better than the displacement control results. Mixed control hybrid simulation methods do not provide better results than the force control methods because of the interaction between the two actuator control modes. This dissertation concludes with a presentation on the direction of future research. Further validation of these methods is required with a very stiff experimental setup and an in-depth error analysis. New parameters for switching need to be explored. Continual development of mixed control methods is suggested. Simulink/Stateflow models using high order polynomials should be tested. Explicit and predictor-corrector time integration schemes should be implemented for the equilibrium force control method.