Published papers

This paper proposes an optimal trading algorithm for a decentralized energy trading system among multiple building energy management systems (BEMS). Each building generates solar power and has a battery. As a decentralized energy trading algorithm, we first solve the proposed optimization problem to determine the optimal trading volume that minimizes the total purchase price of electricity while satisfying the battery capacity constraint under the consideration of the solar power generation error and the time-varying electricity price. To take into account the foregoing, model predictive control and a scenario-based robust optimization are applied. After the information on the desired amount of electricity transactions is communicated between buildings, the electricity sales price is updated. These steps are iterated until the termination condition is satisfied, and the final trading volume is determined. The convergence of the proposed algorithm is analyzed and the sufficient conditions for the convergence of the price update parameters are shown. Finally, the total price cost and computational cost are compared with the centralized one, and the effectiveness and the superiority of the decentralized control system are shown by several numerical simulations.

本論文では厳密な線形化を用いて磁気軸受の広域安定化を行った．まず磁気軸受に対して非線形モデルを導出し，その非線形モデルが非線形な座標変換と入力変換により厳密線形化可能であることを証明した．そして厳密線形化手法を用いて非線形情報を保存した線形モデルを導出し，線形モデルに対して最適制御により状態フィードバックを設計した．最後に制御実験により，動作点移動，ロータの回転速度変化に対する制御性能・安定性の有効性検証を行った．

本稿ではリニア直流モータをアクチュエータとして用いるパンタグラフ基礎実験装置において, 種々の力外乱に対しても架線-パンタグラフ間の接触圧を一定にするレギュレータ問題を扱った．本研究では時間応答に対する制御性能を考慮したH∞制御系設計を行い，特に被制御量に重みをかけることにより，時間領域でのレギュレーションと応答特性の改善を図った．パンタグラフ基礎実験装置での力制御実験を行い，有効性を確認した.

This paper deals with load frequency control problem of power system. In recent years, due to the large introduction of generators using renewable energy, the relative load fluctuation increases, making it difficult to maintain power balance. Therefore, this paper proposes frequency control method using Moving Horizon Estimation and iterative gradient method for connected power system. In this paper, since system handles frequency control problems in real time, the system manager needs to grasp all states of power system every moment, and estimates the state of power system by using Moving Horizon Estimation. Next, control inputs are determined such that each area minimizes evaluation function including barrier function considering constraint conditions for generator and battery by using iterative gradient method. At this time, since the control target is connencted power system, it is represented as a model including the state of adjacent area, so that each area shares information with each other and performs cooperative control between areas. Finally, simulation of load frequency control using Moving Horizon Estimation and iterative gradient method is carried out and the effectiveness is confirmed.

This paper presents an analysis of H∞ Filter(HF) for Robotics Mapping and Localization with unknown noise statistics. HF which is also known as the minimax filter is proposed in this paper to estimate the robot and landmarks location while robot moves through an unknown environment. Some of the conditions are proposed to ensure that the state covariance in HF is converging to a steady state value. Furthermore, the analysis of HF convergence for a robot observing landmarks are presented to examine its behavior through the observations. From the experimental results, HF gives a sufficient estimation about the environment. Subsequently, such a result can provide other available estimation methods with the capability to ensure and improved estimation in robotic mapping and localization problem.

This paper deals with the problem of secure state estimation in an adversarial environment with the presence of bounded noises. The problem is given as min-max optimization, that is, the system operator seeks an optimal estimate which minimizes the worst-case estimation error due to the manipulation by the attacker. To derive the optimal estimate, taking the reach set of the system into account, we first show that the feasible set of the state can be represented as a union of polytopes, and the optimal estimate is given as the Chebyshev center of the union. Then, for calculating the optimal state estimate, we provide a convex optimization problem that utilizes the vertices of the union. On the proposed estimator, the estimation error is bounded even if the adversary corrupts any subset of sensors. For the sake of reducing the calculation complexity, we further provide another estimator which resorts to the interval hull approximation of the reach set and properties of zonotopes. This approximated estimator is able to reduce the complexity without degrading the estimation accuracy sorely. Numerical comparisons and examples finally illustrate the effectiveness of the proposed esti

Recent rapid developments in reinforcement learning algorithms have been giving us novel possibilities in many fields. However, due to their exploring property, we have to take the risk into consideration when we apply those algorithms to safety-critical problems especially in real environments. In this study, we deal with a safe exploration problem in reinforcement learning under the existence of disturbance. We define the safety during learning as satisfaction of the constraint conditions explicitly defined in terms of the state and propose a safe exploration method that uses partial prior knowledge of a controlled object and disturbance. The proposed method assures the satisfaction of the explicit state constraints with a pre-specified probability even if the controlled object is exposed to a stochastic disturbance following a normal distribution. As theoretical results, we introduce sufficient conditions to construct conservative inputs not containing an exploring aspect used in the proposed method and prove that the safety in the above explained sense is guaranteed with the proposed method. Furthermore, we illustrate the validity and effectiveness of the proposed method through numerical simulations of an inverted pendulum and a four-bar parallel link robot manipulator.

In this paper, we investigate a predictive control problem with information structured constraints motivated by control of micro grid. A system with information structures is defined as a system in which each subsystem collects spatio-temporally different information. For the system, we consider a predictive control law and the finite time constrained optimization problem to be solved online is reduced to a deterministic convex programming problem. Then, we reduce a control problem for a simple micro grid system to the framework and the effectiveness of the proposed control and estimation law is demonstrated through a numerical simulation.

In this paper, we analyze the vulnerabilities due to integrity cyber attacks named zero-stealthy attacks in cyber-physical systems, which are modeled as a stochastic linear time invariant (LTI) system equipped with a Kalman filter, an LQG controller, and a χ2 failure detector. The attacks are designed by a sophisticated attacker so that the measurement residual of the compromised system coincides with the healthy one, and thus it is impossible to detect the attacks. First, we characterize and analyze an existence condition of the attacks from an attacker's standpoint. Then, we extend the attacks into an attacker's goal: The scenario when the adversary wishes to manipulate the systems to an objective designed by him/her. Our results provide that the attacker can manipulate the compromised system to the objective without accessing the networks of real-time sensor or actuator data. Finally, we verify the dangerousness of the attacks through a simple numerical example.

This paper deals with a regional demand response method based on dynamic electricity pricing in a multi-period energy market. In the proposed method, first, the power supply and demand in each area are determined considering power flow using the retail and wholesale electricity prices in a day-ahead market. Next, in a real-time market, the proposed method adjusts to power deviations caused by errors in power generation with power demand reduction by consumers and increased power supply from balancing generators. In addition, in order to guarantee the non-deficiency in the real-time market trading, the incentive and penalty price design methods are discussed in this paper. This paper also shows a distributed algorithm for obtaining the optimal values of each market player in both the day-ahead and real-time markets, and finally, numerical simulation results demonstrate the effectiveness of the proposed method.

This paper deals with a demand adjustment problem of each consumer having appliances by an aggregator based on optimal pricing in a day-ahead electricity market. In this paper, we model consumers, generators, and aggregators and design a market mechanism, where they act to maximize their own profit based on the power price and decide the electricity supply and demand. The dual decomposition is applied to the market problem to maximize the social welfare, where the proposed algorithm decides an electricity price based on the exchange of information among market participants to solve each distributed problem for improving convergence. In addition, the convergence of this proposed method is proven using Lyapunov's stability theorem. Finally, simulation results show that matching of supply and demand is achieved, while satisfying constraints on the consumption amount and the use time of each appliance, and the price determined using proposed price update algorithm converges to a certain price.

This paper deals with passivity-based stability analysis of electricity market trading with dynamic pricing. In the deregulated electricity market, power consumers and generators will participate in market trading as market players. For such a new kind of the market trading system, the dynamic pricing procedure has to be taken into account of the intermittent participation of power consumers and generators. Then, this paper discusses the stability of the electricity market trading system including power flow using passivity analysis. This paper also shows that the optimal power demand, supply and electricity prices are asymptotically stable and these values are derived in a distributed manner through market trading.

This paper addresses the crossing order problem for connected and automated vehicles at an intersection where the problem can lead to higher traffic congestion, especially in high traffic density. First, we formulated the problem with mixed integer linear programming where the solution yields optimal crossing time for each vehicle to cross the intersection. Then, we present an optimization framework to drive each vehicle towards the assigned crossing time while conserving the fuel consumption. Finally, we simulate the intersection scenario with the proposed solution, where it is shown that both congestion and fuel consumption can be reduced significantly.

In this paper, we deal with the charging scheduling optimization problem of electric vehicle in highway via Stackelberg game and matching theory. At first, we propose an algorithm for electric vehicle allocation to charging station which aims to uniformize waiting time of each charging station. In addition, we present energy demand and price decision problem and solve it using Stackelberg game. We compare Stackelberg equilibrium point and Nash equilibrium point and confirm Stackelberg game provide bigger benefit to charging station. Finally, we show validation of proposed algorithm via numerical simulations.

This paper deals with the problem of gyroscopic effect and unbalance vibration of the Magnetic Bearing system. Using a gain scheduled Hm control with a free parameter phi, we design a control system which attenuates the unbalance vibration, and guarantees the stability against the gyroscopic effect in specified rotational speed. Further we implement the controller and evaluate the effectiveness of the proposed approach by experiments. First, our experimental setup is explained, a mathematical model of the magnetic bearing is derived. Then, we introduce the gain scheduled H infinity control with free parameters to a magnetic bearing control, in order to reject the disturbances caused by unbalance of the rotor, and guarantee the stability against gyroscopic effect, even if rotational speed of the rotor changes. At last, several experimental results show the effectiveness of this proposed method.

The authors consider the simultaneous localization and mapping (SLAM) problem with an H∞ filter and with an observation of a landmark that is known a priori. With this observation, the system satisfies observability, and the estimated error is suppressed and the determinant of its covariance matrix becomes small compared with that of the original H∞ filter. As a result, our proposed method avoids finite escape time, the divergence of the error covariance matrix that occurs in the estimation when using the original H∞ filter. We prove the convergence of the error covariance matrix. In addition, with simulations and experimental results, we confirm that finite escape time is avoided, that the derived theorems for the convergence are correct, and that we can accurately estimate the state of the robot and the environment.

This paper deals with a dynamic pricing using the H∞ control with uncertain behavior in electricity market trading. The dynamic pricing is a decision procedure of an electricity price based on power supply and demand balance in power grids. However, in a future deregulated electricity market, both power consumers and generators determine their own power demand and supply selfishly according to the price information. Moreover, uncertainties are also included in their behavior. For this problem, we propose a novel price decision method using the locational price-updating H∞ controllers. This paper shows a design process of this price-updating H∞ controller to evaluate the following performance to reduce supply-demand imbalances in power grids and the robustness against uncertainties in electricity market participants' behavior, respectively. In addition, this paper also shows the effectiveness of our proposed price decision method using the designed H∞ controller through numerical simulation results.

In this paper, we propose a novel parameter initial value estimation method of Lithium-ion battery for state of charge (SOC) estimation by using Local Regression Modeling (LRM) and Vector Fitting (VF). To estimate SOC accurately using nonlinear Extended Kalman Filter(EKF), an adequate set of initial parameters of the battery model are required. Therefore, we apply LRM and VF method to derive the initial parameters and the battery model from a constructed database of frequency response data. In addition, we demonstrate the SOC estimation method using EKF by the determined initial value and model. Finally, the effectiveness of the proposed method is shown via several experimental results.