基于融合智能算法斜拉桥振动控制Benchmark问题的混合控制策略研究

Many control strategies and devices have been developed and investigated to protect structures against natural hazard such as earthquake.The control of very flexible and large structures such as cable-stayed bridges is a unique and challenging problem.Benchmark control problem for seismic response of cable-stayed brdige(Emerson memorial bridge) has been presented in the Second International Conference on Vibration Control. Researcher must define the devices, algorithms to be used in the proposed control strategy. This Benchmark model is used as the research object in the project.The aim of the study is to investigate the application of the hybrid control strategy and the hybrid intelligent control algorithm, in reducing the displacement of the deck as well as base shear and moments at the base of the towers. In the first part of the project, the mechanical properties of the new Electromagnetic drive active mass damper (EMD) is identified by the method of the fuzzy neural network, and the smart power-electric relation is established. A hybrid control strategy combining passive and active control systems is proposed for the benchmark bridge problem. In this hybrid control strategy, passive rubber bearings are used as passive control devices.The passive control strategy has a great role for the effectiveness of control performance. A total of 24 electromagnetic drive active mass damper (EMD), which are used in the benchmark problem, are employed.The actuators have a capacity of 1000 kN. Based on artificial neural networks, fuzzy logic, genetic algorithms and other intelligent algorithms, hybrid intelligent control algorithm is adopted. A set of sixteen criteria have been developed to evaluate the capabilities of each control strategy. Three Benchmark historical earthquakes(El Centro earthquake,Mexico city earthquake,Gebze earthquake) are researched respectively. At last numerical simulations results of the hybrid control strategy are obtained. The performance of the proposed hybrid control design is compared to that of the passive control design and active control design. Finally, the robustness of the control system stiffness error caused by structural modeling is researched, one or a few Benchmark seismic waves are used to be example, the three test conditions, stiffness variation from -15%, 0% and 15% are calculated respectively.And the effectiveness,feasibility and robustness of the cable-stayed bridge with the EMD vibration control system is investigated.?

以国际桥梁界地震激励下进行振动控制研究的斜拉桥Benchmark模型（Emerson memorial bridge）为研究对象，利用模糊神经网络模拟新型电磁驱动主动质量阻尼器（EMD）的力学特性，建立其力电关系的智能模型,为工程中的计算仿真以及实时在线控制实施提供有效保障；综合运用EMD系统及合成橡胶支座，形成主动控制+被动控制的混合控制策略；设计基于人工神经网络、模糊推理、遗传算法等智能算法的融合智能算法，选取目标控制函数，设计16项控制效果指标，使桥梁结构在每一时刻根据结构状态和所受外荷载情况进行最优控制，及时消减结构的动力响应，对比分析采用混合控制策略和单一控制策略的有效性；最后针对结构建模中由于刚度误差引起的控制系统鲁棒性问题,以某个或某几个Benchmark地震波输入为例，设计3种试验工况，刚度变异从-15%、0%到15% ，分别仿真结构动力响应，评价该振动控制系统的鲁棒性。

目前我国正面临一个新的地震活跃期，如何对桥梁等生命线工程进行合理抗震设防，尽量减小地震过程中次生灾害的发生，加快救援速度，是工程研究人员急需解决的难题。1972 年美籍华裔J.T.P.Yao 提出结构振动控制概念以来，它已被理论和实践证明是结构抵抗地震破坏的有效手段。传统的桥梁结构控制振动对策是通过增强桥梁本身的抗力实现的，是消极被动的结构对策。结构振动控制是门新兴交叉学科，综合了控制论、计算机等多领域高新技术，在结构某些部位设置控制装置，当结构振动时，施加一组控制力或调整结构动力特性，减小或抑制结构振动反应。. 本课题以国际桥梁界地震激励下进行振动控制研究的斜拉桥Benchmark模型为研究对象，各国学者经过近十年科学研究，控制策略用到了被动控制（粘性、粘弹性阻尼器等）、半主动控制（MR阻尼器等）、主动控制（液压或伺服电机AMD）、混合控制（被动控制与半主动控制或主动控制组合）；控制算法运用了各种传统控制算法（LQG算法、滑移模态控制算法等），模糊推理算法。对该桥振动控制的Benchmark 问题，新的研究趋势为综合运用各种智能算法形成融合智能控制算法，采取混合控制策略，展开控制效果的数值仿真分析。主动质量阻尼器和磁流变阻尼器有成熟的数学模型，利用人工神经网络学习数学模型输入/输出数据的映射关系，建立控制装置的智能模型。为了改善智能模型的学习性能，引入遗传算法寻找人工神经网络最优连接权值。控制装置智能模型用于桥梁结构在线振动控制，节约计算时间，方便快捷，有效减少时间滞后问题带来的影响。. 利用模糊神经网络模拟新型电磁驱动主动质量阻尼器（EMD）的力学特性，建立其力电关系的智能模型,为工程中的计算仿真以及实时在线控制实施提供有效保障；综合运用EMD系统及合成橡胶支座，形成主动控制+被动控制的混合控制策略；设计基于人工神经网络、模糊推理、遗传算法等智能算法的融合智能算法，选取目标控制函数，设计16项控制效果指标，使桥梁结构在每一时刻根据结构状态和所受外荷载情况进行最优控制，及时消减结构的动力响应，对比分析采用混合控制策略和单一控制策略的有效性；最后针对结构建模中由于刚度误差引起的控制系统鲁棒性问题,以某个或某几个Benchmark地震波输入为例，设计3种试验工况，刚度变异从-15%、0%到15% ，分别结构动力响应，评价控制系统鲁棒性。