Wind-induced vibration (WIV) of tall buildings is a major cause of occupant discomfort and potential fatigue damage. Catastrophic failure may also take place at wind speeds that are lower than the design values due to phenomena such as vortex shedding or flutter-induced instabilities. This paper presents a data-driven adaptive control strategy that continuously seeks to minimize WIV for a given average flow condition by independently adjusting the angular orientation of an active façade system composed of a set of plates. The controller utilizes Genetic Algorithm (GA) optimization to determine façade plate angles that minimize time-averaged WIV amplitudes by altering the aerodynamics of the building. The GA is assisted by two artificial neural networks (ANNs). A predictor ANN acts as a regression model that continuously estimates the system dynamics. An optimizer ANN allows the controller to quickly recall what plate angle combination to use for a given wind condition. A 2D fluid-structure-interaction (FSI) model is used to simulate the steady-state response of the building for given wind conditions and façade plate angle combinations. The 2D model is validated by comparing to published data. Initial results show that vibration amplitudes were reduced up to 94% upon enabling the proposed smart façade controller.

高层建筑的风致振动（WIV）是造成乘员不适和潜在疲劳损伤的主要原因。由于诸如涡旋脱落或颤振引起的不稳定性之类的现象，在风速低于设计值时也可能发生灾难性故障。本文提出了一种数据驱动的自适应控制策略，该策略通过独立地调整由一组板组成的主动立面系统的角度方向，不断寻求在给定的平均流量条件下将WIV降至最低。该控制器利用遗传算法（GA）优化来确定外墙板角度，以通过改变建筑物的空气动力学来最小化时间平均WIV幅度。遗传算法由两个人工神经网络（ANN）辅助。一个预测人工神经网络作为一种回归模型，可以连续估算系统动态。一个优化ANN允许控制器快速回顾一下板角组合使用对于给定的风力条件。在给定的风况和外墙板角度组合下，使用二维流体-结构-相互作用（FSI）模型来模拟建筑物的稳态响应。通过与发布的数据进行比较来验证2D模型。初步结果表明，启用建议的智能立面控制器后，振动幅度最多可降低94％。