Research Of AMD Robust Control System With Uncertainties For High Flexible Structures Based On Engineering Cases

Under a strong wind,due to the excessive vibration of highly flexible buildings,the performance indexes such as story drift and comfort often do not meet the requirements of the relevant codes.For this reason,additional dampers and energy dissipation materials are usually adopted to meet the expected vibration control effect and the comfort requirementsof highly flexible buildings.Among them,active mass damper/driver(AMD)is favored because of its good performance.However,uncertainties,including noise signals,the inaccuracy of structural parameters,time-varying delays and the fault signal of sensors,limit its application in vibration control of highly flexible buildings to some extent.Therefore,this paper focuses on the influence of the above uncertainties.Based on an improved reduced-order system and a state observation system,a corresponding robust control system was proposed.The main work is as follows:A new reduced-order model was constructed based on the improved balanced truncation(BT)method.During model-reduction process,the classical BT method ignored the information of high-order damping ratio and vibration modal information of the original structure,which leads to a lower accuracy of an AMD control system.In this paper,a new reduced-order model based on the improved BT method was designed to retain the abandoned high-order modal information.Transfer functions,retained-orders and control effects were regarded as the indexes to analyze the differences between the reduced-order model and the original model.Furthermore,the optimal gain of the reduced-order system was designed based on the guaranteed cost control algorithm.To verify its effectiveness,the proposed methodology was applied to a numerical model of a ten-story frame and an experiment model of a four-story steel frame.A state observer was constructed with a robust filter.First,for the problem that the full state vectors of AMD systems is difficult to measure directly,a state observer,which used wind-induced acceleration responses of structural stories as input,was designed.Secondly,focusing on the problem that the measurement noise are more easily to be introduced into the state observer system with the accelerometers,a Kalman filter and a linear matrix inequality(LMI)-based filter were proposed.Compared with the Kalman filter,the LMI-based filter considered the optimal design of the gain matrix to better guarantee the stability of AMD control parameters.A time-delay compensation system based on the regional pole-assignment(PA)algorithm was analyzed.The effect of time-varying delays on the performance and stability of single-degree-of-freedom(SDOF)and multi-degree-of-freedom(MDOF)systems were derived theoretically.In the calculation process of control gain,the regional PA algorithm was used to compensate for the two unfavorable factors such as uncertain structural parameters and time-varying delays.Both numerical and experimental results demonstrated that the proposed method enhanced the performance and stability of uncertain time-delay systems.In addition,the closed-loop poles of the time-delay system were distributed in an expected area after region PA compensation,and the PA range was more relaxed.The performance of the system was better than the compensation system based on the PA algorithm.A guaranteed cost control(GCC)system considering higher-order control-structure interaction(CSI)effect was optimized.The time-varying delay of AMD systems mainly results from the CSI effect.Therefore,three kinds of mathematical models of AMD systems were established to analyze the influence of the CSI effect.Then the influence mechanism was analyzed with the model considering high-order CSI effect.The main influencing factors included the input frequency of the control voltage,the structural parametric uncertainties,and the control gains.Finally,a new time-delay compensation controller based on the GCC algorithm was proposed to consider time-varying delays caused by higher-order CSI effect.The experimental structure was verified as an example.The results showed that the proposed GCC compensation controller reduced the influence of CSI effect effectively.An active fault-tolerant control(FTC)system with a state observer was analyzed.Since failures in accelerometers degrade the performance of AMD systems,a fault detection and isolation(FDI)filter design method was developed in this paper.The fault signal diagnosis and isolation problem was transformed into H2/H? control problem,and the dynamic optimal fault detectors were solved using LMI approach.Faults that existed in accelerometers were diagnosed and isolated.Furthermore,the state observer was used to estimate the state vectors of AMD systems using isolated signals,and then the control force was calculated to implement fault tolerant control(FTC).The results demonstrated that the FTC system not only ensured the stability of AMD parameters,but also controlled the dynamic response of the building well,and the control effects were basically equivalent to the system without fault signals.