Experimental Research on Intelligent Active Control of Building Structures by Fuzzy Optimal Control

本文对主动控制系统振动台试验进行了基础研究。本文采用了作者提出的一种智能主动控制方法。该方法具有以下特点：(1)用隶属函数描述目标响应和控制变量；(2)对地震输入进行实时预测和结构识别；(3)采用模糊最大化决策进行优化。采用输入减少法作为主动控制方法。被试体是一个单自由度系统，其顶部有主动质量驱动器（AMD）。AMD由直流伺服电机驱动。振动台试验系统由三台个人计算机组成。在第一台计算机上，启动振动台，观察试件的响应。在第二台计算机上，通过模糊最大化决策计算最优控制变量。在第三台计算机中，AMD按照给定的控制力进行控制。这些计算机与第一台计算机的AD定时器同步。首先，进行了一些基本测试，以检查测试系统的基本性能。下一步，对试件进行智能控制试验。在这些试验中，采用条件模糊集规则和分段线性响应方程作为地震输入和结构响应的预测方法。结果表明，在正弦波等谐波波的情况下，该测试系统可以减小试件的响应。然而，在地震等随机波输入情况下，主动控制的有效性变小。因此，为了开发出更有效的测试系统，需要对测试系统进行进一步的改进。通过本文的研究，我们可以得到建筑物模糊最优控制所采用的实验测试的大量基本数据和系统。

In this study, a fundamental research on shaking table tests of active control system is carried out. Here, an intelligent active control method proposed by the authors are employed. This method has the following features ; (1) Target responses and control variables are described with membership functions, (2) Real time prediction of earthquake input and structural identification are performed, and (3) Optimization is performed by fuzzy maximizing decision. An input reduction method is employed as an active control method. Test specimen is a single-degree-of -freedom system with active mass driver (AMD) at the top of it. AMD is activated by DC servomotor.A system of the shaking table test consists of three personal computers. In the first computer, the shaking table is activated, and responses of the specimen are observed. In the second computer, optimal control variables are calculated by fuzzy maximizing decision. In the third computer, AMD is controlled in accordance with given control forces. These computers are synchronized with the AD timer of the first computer.At first, some fundamental tests are carried out to examine a basic performance of the testing system. In the next step, intelligent control tests of the specimen are carried out. In these tests, a conditioned fuzzy set rules and piece-wise linear response equations are employed as the prediction methods of earthquake inputs and structural responses. These results show that in case of harmonized waves such as sine wave, this testing system can reduce the responses of the specimen. However, in case of random waves such as earthquake inputs, effectiveness of active control becomes small. So, further improvements of the testing system are necessary to develop the more effective testing system. Thorough this research, we can get a lot of basic data and systems on the employed experimental tests of fuzzy optimal control of buildings.