An architectural building cluster morphology generation method to perceive, derive, and form based on cyborg-physical wind tunnel (CPWT)

Rapid deduction based on environmental data has become a key factor influencing the early stages of conceptual design decisions. Traditional wind tunnels and CFD computational tools are time-consuming and number-limited of assumptions in iterative design, hence both are more commonly used in post-evaluation. With the development of artificial intelligence, this paper presented a new cyborg-physical wind tunnel (CPWT) design platform. By combining a customized physical wind tunnel with an optimization algorithm, a large amount of physical data can be quickly derived, and evaluations and predictions can be made. This paper first introduces the concepts of wind environment performance-based design including the physical platform design and the sensing system. The design methodology is reviewed in three sections: deduction framework, optimization goal, and evaluation indicator. Then, the paper proposed a new wind tunnel device, the CPWT, which incorporates 120 dynamic lifting units equipped with sensor, Arduino, and servomotor to capture wind data and generate real-time responses. Artificial neural network is integrated in the experiment to train on data like the windward area index and staggering index for the morphology generation. The result demonstrated that the mean square error is able to quickly stabilize to the range of [-0.02783, 0.02892] and [-2.324, 2.845], which achieved a lower data error and a better result predictability. The enhanced sensibility and real-time feedback are able to provide a more rapid method for the wind environment performance-based design in this digital age.