Search

This article addresses a control problem on air-conditioning systems in a room that is regarded as a control practice of nonlinear distributed parameter systems. Specifically, we consider designing a controller to suppress an oscillatory response of the in-room temperature field. The main idea of this article is to apply the emergent theory of the Koopman operator and Koopman mode decomposition (KMD) for nonlinear dynamical systems and to formulate a technique of damping assignment to a nonlinear Koopman mode in a fully data-driven manner. Its effectiveness is examined by numerical simulations of a simple 2-D problem guided by the measurement of practical room space. A finite-element commercial package is also used for validating the technique in a practical 3-D problem.

Measurement of power system oscillations is the first step for monitoring and control of interconnected power systems. Koopman Mode Decomposition (KMD) is a novel technique of nonlinear time-series analysis guided by the spectral theory of Koopman operators for nonlinear dynamical systems. This paper presents an online algorithm of KMD that is suitable to streaming data derived from synchrophasors. Its effectiveness is established by applying it to extract an inter-area mode of the Japan's western power system directly from synchrophasor data at three locations.

We develop an algorithm for synthesizing a spatial pattern of charging/discharging operations of in-vehicle batteries for provision of ancillary service (AS) in power distribution grids. The algorithm is based on the ordinary differential equation (ODE) model of the distribution voltage that has been recently introduced. In this paper, first, we derive analytical solutions of the ODE model for a single straight-line feeder through a partial linearization, thereby providing a physical insight to the impact of spatial electric vehicle (EV) charging/discharging to the distribution voltage. Second, based on the analytical solutions, we propose an algorithm for determining the values of charging/discharging power (active and reactive) by in-vehicle batteries in the single feeder grid, so that the power demanded as AS (e.g., a regulation signal to the distribution system operator for the primary frequency control reserve) is provided by EVs, and the deviation of the distribution voltage from a nominal value is reduced in the grid. Effectiveness of the algorithm is established with numerical simulations on the single feeder grid and on a realistic feeder grid with multiple bifurcations.

The concept of microgrid has been proposed as an effective method for utilizing distributed multiple generation units including renewable energy. A microgrid connected to the commercial grid has a potential to contribute to not only short-term output levelling by maintaining the output power [kW] at reference but also long-term balancing of electricity supply and demand by regulating the actual amount of output energy [kWh]. In this paper, we propose a new approach to optimal supply planning of a microgrid considering the imbalance (i.e. the difference between planned and actual amounts of electric energy). By evaluating the approach numerically, we show that the minimization of imbalance contributes to not only the supply-demand balancing but also its economically-efficient operation in terms of the sum of the running and imbalance costs.

Recently, power flow patterns in distribution network has considerably modified due to a large amount of PV installation and increase of consumers behavior patterns. Hence, a system operator has become difficult to control voltage by the conventional method. As a countermeasure of this problem, the reactive power control method with the PV inverters has been reported. In this paper, we proposed a new voltage control method with reactive control of PV inverters using reinforcement learning. In particular, the purpose of proposed method is to bring out the cooperative reactive power control without the information between PV inverters. We verify the effectiveness of proposed method through computational simulation using distribution network model.

ABSTRACTRenewable energy such as wind generation is a countermeasure against global warming and fossil-energy resource depletion. However, wind-farm (WF) output fluctuates because of unstable wind conditions and the occurrence of long-term fluctuations, which are called ramp events. WF power generation operator needs to observe the grid code (GC) by controlling wind generators themselves or adjusting a WF output with the energy-storage system (ESS). Accordingly, scheduling the operation of WF with ESS and WF output forecast is necessary. Scheduled operation has two parts: scheduling of generation plans (SGP) and operating WF (OWF) outputs with the charge/discharge of ESS. This paper proposes the scheduling method, which can control ramp events and satisfy GC formulation such as mixed integer liner programming in the SGP part, and the balancing method, which can reduce the imbalance energy due to control delay of ESS in the OWF part. We also evaluate these methods by annual numerical simulation. We further confirm the validity and effectiveness of the proposed method, as well as approximate its battery capacity requirement.