【作者】 曹志峰；

【导师】 孙玉萍； 杜永峰；

【作者基本信息】 兰州理工大学 ， 结构工程， 2004， 硕士

【摘要】 过去三十年间主动控制技术取得了很大的发展，逐渐使其成为保证结构安全、抵抗自然灾害的一种可行的技术手段。结构主动控制的一个重要特征就是能够根据环境干扰的特点，通过作动器按照预先设定的控制算法规则，提供实时控制力来保证结构反应被控制在安全的范围内。因此关于结构控制算法的研究一直贯穿着这一领域研究的主线。为了能对各种算法有一个公认的评价标准，美国Notre Dame大学Spencer研究组通过对主动质量驱动(AMD)控制的三层框架模型实验的数据分析，建立了AMD控制Benchmark模型。最优控制算法由于能根据控制目标函数中权重矩阵的选取综合考虑控制力和控制响应的因素确定最优控制力，符合土木工程结构振动控制问题的需要，因此在实际中得到了广泛的应用。目前为止结构振动控制领域已经发展了三种具有代表性的最优控制算法——经典最优控制近似算法、瞬时最优控制算法、序列最优控制算法。本文主要是结合AMD控制Benchmark模型对三种最优控制算法的控制效果进行了仿真分析比较。主要作了如下几个方面的工作： 1、利用AMD控制Benchmark模型的物理参数及实验数据建立了基于Simulink的仿真模型，该模型考虑了系统时滞和传感器测量误差的影响，可以真实地反映研究对象的动力特性。 2、针对序列最优控制算法具有时变增益矩阵的特点，提出了该算法在Simulink仿真环境中的两种实现方法：S-函数法、数据存取模块法。仿真分析表明两种实现方法是完全等效的，其中S-函数法具有灵活方便的特点，而数据存取模块法则具有仿真速度快捷的特点。 3、以AMD控制Benchmark仿真模型为平台，对经典最优控制近似算法、瞬时最优控制算法和序列最优控制算法在相同控制约束、控制能量相当情况下，进行了仿真分析。仿真结果显示，在基于加速度反馈的AMD控制结构中，运用序列最优控制算法所求得的控制力在跟踪结构峰值响应，以及削减结构峰值响应方面都是三种最优控制算法中表现最出色的。更多还原

【Abstract】 Tremendous progress has been made over the last three decade on the structure vibration control. The main feature of structure vibration control is that it can give the real time control force with the actuator by the control rule of pre-design to guarantee the safe of the structure. So the control algorithm is main line of the research in the field of structure vibration control. For to evaluate the various control algorithm, Spencer, et al. present a AMD benchmark model. Optimal control algorithm’s control objective function includes the effect of control force and control response. So it is very suitable for the civil engineering structure vibration control. At present, three effective optimal control algorithm has been developed in the field of structure vibration control. They are classical optimal control approximate algorithm, instantaneous optimal control algorithm and sequential optimal control algorithm. In this paper, The simulation of optimal control algorithm was implemented base on the AMD benchmark model, and mainly focuses on the following things:1. This paper sets up a simulation model of simulink base on the AMD benchmark model. This model includes the effect of the system delay and the measurement error, So it can accurately represents the behavior of the real experimental model.2. One of the significant features is that the control gain matrix of sequential optimal control algorithm is time-varying, so this paper suggests two methods for realizing sequential optimal control algorithm using Simulink. The two methods are S-function method and data store and take method. Simulation result shows that the two methods are equivalent. The feature of S-function method is convenient, and the feature of data store and take method is faster on the speed of simulation than S-function method.3. Base on the AMD benchmark model, the simulation of optimal control algorithm was implemented with the same control constrain and the same control energy, the sequential optimal control algorithm is advantageous over the instantaneous optimal control algorithm and the classical optimal control algorithm in terms of the ability of reducing the peak response and the control efficiency.更多还原