University of Tokyo – Department of Architecture

Miyata Lab.

Exploring new value in architecture by developing fundamental and applied smart technologies

Research Theme: Smart Building System
In addition to the development of IoT and deep learning, coupled with social backgrounds such as carbon neutrality and GX, there is a growing expectation for higher functionality during operation phase in the building sector. Smart buildings regard architecture as an interface for services to people, and aim to build models and accumulate data in the digital world to constantly improve their value through various applications.

In order to realize smart buildings, Miyata Lab will conduct research activities focusing on the development of models (emulators of building services systems) in the digital world and data-driven control algorithms.

*Miyata Lab is a new laboratory established by the “Smart Building System Research Initiative,” a Social Corporation Program inaugurated on November 1, 2023. Miyata Lab will work jointly with the Akashi Lab, which is the organizing laboratory of the Social Corporation Program.

smartbuildingarchitecture3

The following projects are for reference only and are research projects that Dr. Miyata has been working on.

Simulation development of HVAC systems

A simulation that can represent the control behavior of an Heating, Ventilation and Air-Conditioning (HVAC) system has been developed originally. This simulation is used to generate data when problems occur in the system (fault datasets) and to study optimal control.

Automated fault detection and diagnosis

Using high quality datasets generated using the simulation, the causes of faults can be directly diagnosed using deep learning. This is an issue that requires further advancement and industrialization of the methodology.

Optimal control

Optimal control algorithms are being developed for various scales, such as control based on power system conditions (demand response) and control that neutralizes room pressure to prevent drafts. As with fault detection and diagnosis, various methods such as deep learning are attempted to be applied.

neutralization

Semantic data model utilization

We are investigating the possibility of describing and utilizing the semantic structure of a system in a program. This theme has recently entered the building equipment field from the field of computer science, and is a key issue for future technological development.

brickmodel

Contribution to campus GX and sustainable campus

As a member of the university, we are also involved in campus GX and sustainability activities, using our expertise in building data analytics and system architecture.

greenut

For those looking for opportunities (Message from Shohei Miyata):

I myself had no programming experience until I entered the master’s program. Currently, my research focuses on simulation, deep learning, semantics, and other research that requires programming. Therefore, I think that those with no experience may feel uneasy, but I hope that you will grow a lot through your research activities.

buildinggx.com