Yasuaki Wasa

Yasuaki Wasa, Kenji Hirata, Kenko Uchida

We discuss an incentivizing market and model-based approach to design the energy management and control systems which realize high-quality ancillary services in dynamic power grids. Under the electricity liberalization, such incentivizing market should secure a high speed market-clearing by using the market players' private information well. Inspired by contract theory in microeconomics field, we propose a novel design method of such incentivizing market based on the integration of the economic models and the dynamic grid model. The conventional contract problems are analyzed for static systems or dynamical systems with control inputs directly operated by the principal. The analysis is, however, in discord with the incentivizing market. The main challenge of our approach is to reformulate the contract problems adapted to the market from the system and control perspective. We first establish the fundamental formulas for optimal design and clarify the basic properties of the designed market. We also discuss possibilities, limitation and some challenges in the direction of our approach and general market-based approaches.

Takeshi Hatanaka, Yasuaki Wasa, Masayuki Fujita

This paper investigates cooperative environmental monitoring for Pan-Tilt-Zoom (PTZ) visual sensor networks. We first present a novel formulation of the optimal environmental monitoring problem, whose objective function is intertwined with the uncertain state of the environment. In addition, due to the large volume of vision data, it is desired for each sensor to execute processing through local computation and communication. To address the issues, we present a distributed solution to the problem based on game theoretic cooperative control and payoff-based learning. At the first stage, a utility function is designed so that the resulting game constitutes a potential game with potential function equal to the group objective function, where the designed utility is shown to be computable through local image processing and communication. Then, we present a payoff-based learning algorithm so that the sensors are led to the global objective function maximizers without using any prior information on the environmental state. Finally, we run experiments to demonstrate the effectiveness of the present approach.

Junsei Ito, Yasuaki Wasa

This article proposes a data-driven PID controller design based on the principle of adaptive gain optimization, leveraging Physics-Informed Neural Networks (PINNs) generated for predictive modeling purposes. The proposed control design method utilizes gradients of the PID gain optimization, achieved through the automatic differentiation of PINNs, to apply model predictive control using a cost function based on tracking error and control inputs. By optimizing PINNs-based PID gains, the method achieves adaptive gain tuning that ensures stability while accounting for system nonlinearities. The proposed method features a systematic framework for integrating PINNs-based models of dynamical control systems into closed-loop control systems, enabling direct application to PID control design. A series of numerical experiments is conducted to demonstrate the effectiveness of the proposed method from the control perspectives based on both time and frequency domains.