Abstract
〈Vol.13 No.5(2020.9)〉

 

Titles

[Contributed Papers]

 

■ Load Frequency Control and Real-Time Pricing with Stochastic Model Predictive Control

Kumamoto University・Taro YANAGIYA,
Joetsu University of Education・Yusuke OKAJIMA,
Osaka Institute of Technology・Tomoaki HASHIMOTO,
and Kyoto University・Toshiyuki OHTSUKA

We developed a load frequency control system with stochastic  model predictive control (SMPC) for power systems where the market  penetration of wind power generation is high. The controller adjusts the electricity price for heat pump water heaters while at the same time  controlling thermal power plants and batteries in order to maintain the frequency in the designated range. We propose an approach for solving SMPC problems on Hammerstein models including affine disturbance feedback parametrization. Simulation results show that SMPC with affine disturbance feedback parametrization outperforms both SMPC without  parametrization and deterministic model predictive controlin terms of the stage-cost and constraint violation.

 

■ Relative Position Estimation for Formation Control with the Fusion of Predicted Future Information and Measurement Data

Tottori University・Tsuyoshi OGAWA,Kyoto University・Kazunori SAKURAMA,
Tottori University・Shintaro NAKATANI,and Shin-Ichiro NISHIDA

This paper addresses a relative position estimation problem for formation control of multiple robots. In the authors' previous paper, a relative position estimation method has been proposed, which fuses information from distance sensors and wireless communication. In this method, it is assumed that the robots communicate with others by wireless devices at every control sampling time. Therefore, depending on the performance of the wireless devices, the control sampling time should be set to a large value, which can degrade control performance. In this paper, we propose a new relative position estimation method, which is effective even if the communication sampling time is longer than the control sampling time. The idea  in this method is to use predicted information on the time-series of the control input from detected robots. We develop a method to generate the time-series of the predicted control input for successful estimation. Finally, we verify the effectiveness of the proposed method by simulations and an experiment.

 

■ Stability Optimization of Positive Semi-Markov Jump Linear Systems via Convex Optimization

Nara Institute of Science and Technology・Chengyan ZHAO,
Osaka・University・Masaki OGURA,
and Nara Institute of Science and Technology・ Kenji SUGIMOTO

In this paper, we study the problem of optimizing the stability of positive semi-Markov jump linear systems. We specifically consider the problems of tuning the coefficients of system matrices for maximizing the exponential decay rate of the system under a budget-constraint and minimizing the parameter tuning cost under the decay rate constraint. By using a result from the matrix theory on the log-log convexity of the spectral radius of nonnegative matrices, we show that the stability optimization problems are reduced to convex optimization problems under certain regularity conditions on the system matrices and the cost function. We illustrate the validity and effectiveness of the proposed results by using an example from the population biology.

 

■ A Performance Evaluation of Periodic Signal Analysis by ARS Compared with Frequency Analysis by FFT

Aichi Prefecutural University・Ryota TAKAO,and Yukihiro KAMIYA

Accumulation for real-time serial-to-parallel converter (ARS) has been proposed as a computationally-efficient method for signal analysis. Since it consists of additions and a few divisions only, it was shown that the computational load is drastically reduced compared with the fast Fourier transform (FFT). In this paper, we clarify that ARS achieves a higher resolution in low-frequency bands comparing with FFT. In addition, selection criteria between ARS and FFT are clarified theoretically in terms of the resolution. Through the performance analysis of ARS in low-frequency bands, it is expected that ARS is a powerful tool for signal analysis in the Internet of Things realizing the edge computing.

 

■ Synthesis of Memory Gain-Scheduled Controllers for Discrete-Time LPV Systems

Kobe University・Izumi MASUBUCHI,and Yuta YABUKI

This paper considers synthesis of discrete-time gain-scheduled controllers for linear parameter varying systems based on linear matrix inequalities (LMIs). Unlike most of the previous results, gain-scheduled controllers that depend on memory of the scheduling parameters are  investigated in this paper. Through the method of change-of-variables, parameter-dependent LMIs are obtained for synthesis of gain-scheduled controllers from extended LMIs for H∞ and H2 performances. Numerical examples are provided to illustrate the proposed synthesis methods.