【作者】 何敏；

【导师】 王建国；

【作者基本信息】 合肥工业大学 ， 结构工程， 2008， 博士

【摘要】 本文以国际桥梁界地震激励振动控制的Emerson Memorial斜拉桥Benchmark模型为研究对象,尝试以电磁驱动主动质量阻尼器(Electromagnet-drived active mass damper,简称EMD)作为主动控制装置,运用模糊逻辑推理、人工神经网络、模糊神经网络、遗传算法等多种智能理论,建立EMD主动控制系统的智能模型;优化EMD主动控制装置的系统参数;对斜拉桥结构系统模型动力特性进行非参数辨识;设计若干控制指标,比较地震荷载作用下基于智能理论的斜拉桥结构振动控制效果;以及对EMD系统线圈输入电压的在线控制进行了研究。主要研究成果如下:1)利用模糊神经网络理论的非线性处理能力,以试验数据作为网络训练及验证数据,对电磁驱动AMD振动控制系统建模,建立更为宽泛的条件下系统输入输出变量之间的关系,从而有效改善现有力电关系模型的计算效果。2)利用遗传算法理论的非线性寻优能力,针对电磁驱动主动质量阻尼器的系统参数优化问题,以控制效果一定时作动器作功最小为目标函数,找出斜拉桥EMD控制系统在地震荷载作用下的最优控制器参数。3)利用人工神经网络理论的自学习、非线性逼近能力,建立桥梁结构非参数系统辨识模型,可以辨识桥梁结构的非线性特性,真实反映桥梁结构在无控及有控条件下的动力特性,通过对Emerson Memorial斜拉桥仿真实例验证了该方法的有效性。4)建立斜拉桥EMD系统的力电计算模型,根据大跨桥梁抗震设计特点,提出基于遗传-BP神经网络的斜拉桥EMD系统输入电压的在线控制方法,由传感器采集到的结构状态信号和外界地震激励信号瞬时做出系统输入电压选择,解决传统振动控制计算量繁重、运行慢导致时滞等问题,实现对桥梁结构在线较为精确的振动控制。5)利用模糊逻辑推理理论易于形成专家知识和经验,有较强鲁棒性,简单实用的特性,对斜拉桥EMD系统实施智能主动控制,使得桥梁结构能够在每一时刻根据结构状态和所受外荷载情况选择最优主动控制力,并通过EMD控制装置施加到结构中,达到减小结构动力响应的目的,通过对Emerson Memorial斜拉桥仿真实例验证了该方法的有效性。6)利用模糊神经网络理论的学习能力与聚类归纳能力,对斜拉桥EMD系统实施智能主动控制,使得桥梁结构能够在每一时刻能够根据结构状态和所受外荷载情况选择最优主动控制力,并通过EMD控制装置施加到结构中,达到减小结构动力响应的目的,对比基于模糊逻辑推理理论的桥梁结构智能主动控制,证明该方法减振效果更加明显有效。更多还原

【Abstract】 Benchmark bridge of Emerson Memorial was taken as the research object in the thesis. The electromagnet-drived active mass dampers (EMD) were selected as the active control devices. Intelligent theories such as fuzzy logic theory, artificial neural network, fuzzy neural network, genetic algorithm and so on, were applied to model the behavior of EMD system, find the optimal controller parameters, and identify the bridge structural dynamical characteristics. A set of evaluation criteria was designed to study the control effectiveness based on different intelligent theory. And online control method of EMD’s input voltage was studied. Main contents of the thesis are as follows.1. Using non-linearity handling ability of the fuzzy neural network, take the test data as the training and checking data of FNN, establishes the relations between the input variable and the output variable under the generalized condition, builds an intelligent model of EMD vibration control system, thus computation effect of the existing force electricity relations for electromagnet-drived AMD vibration control system can be improved effectively.2. A control parameter optimization method based on genetic algorithm was presented. To the electromagnet-drived active mass damper’s parameter design in bridge structural control, the optimal controller parameter under seismic excitation was found by the strong nonlinear optimization ability of genetic algorithm.3. A bridge structural system identification method based on artificial neural network was presented. Utilizing the strong learning and nonlinear approaches ability of the artificial neural network, the weakness of the traditional system identification such as bad weak tolerant ability and nonlinear identification was overcame. The method can identify the dynamic characteristics of the bridge under uncontrolled and controlled effectively. Through the numerical simulation example, validity and usability of the method was confirmed.4. The force electricity relation of electromagnet-drived active mass damper working for long span bridge was established based on the electromagnetic theory. According to the earthquake resistance design characteristic of long span bridge, online control method of EMD’s input voltage based on GA-BP network was presented. The input voltage can be instantly chosen as soon as the dynamic response and earthquake acceleration are gathered by the sensors. The big slow computation and time delay problem are both resolved. Thus, the online vibration control of the long span bridge can be realized precisely.5. Intelligent active vibration control online of the bridge structure based on fuzzy logic theory was presented. Utilizing the characteristics of fuzzy logic theory such as the easiness forming the expert knowledge and experience, strong robustness, simple but practical. The active force can be instantly chosen as soon as the dynamic response and earthquake acceleration are gathered by the sensors. The force can be exerted in the bridge structure by the EMD, and then, the dynamic response of the bridge can be reduced effectively.6. Intelligent active vibration control online of the bridge structure based on fuzzy neural network was presented. Utilizing the strong learning and clustering ability of the fuzzy neural network, the active force can be instantly chosen as soon as the dynamic response and earthquake acceleration are gathered by the sensors. The force can be exerted in the bridge structure by the EMD, and then, the dynamic response of the bridge can be reduced greatly.更多还原