Shoji Kawasaki

Insolation Forecasting Method Using NN with EPSO in Consideration of Correlation between Other Regions

Recently, there is a movement that aims to realize a distributed energy system in Japan. The PV (photovoltaic) generation is especially expected, and it is anticipated that interconnection number of PVs will increase more and more in the future. However, an amount of insolation is easily affected by the weather, the output of PV and the supply of electricity becomes unstable. And it causes various problems in a distribution system. Therefore, it is important to forecast the amount of insolation. In previous study, MLPNN (multilayer perceptron neural network) is a general method to forecast the amount of insolation. However, it is in danger of falling into a local solution by only MLPNN. In this study, the authors propose a forecasting method of amount of insolation using MLPNN and EPSO (evolutionary particle swarm optimization). The authors use EPSO in addition to MLPNN to solve the problem. The authors also propose a forecasting method of amount of insolation using other regions weather data for the accuracy improvement.

A Decision Method of Cogeneration Operation Schedule by Combining Use of SOFCs and PEFCs

In this paper, the authors propose a cogeneration system by combining two kinds of FCs (fuel cells) for a collective housing. The good points which each FC has are applied to the cogeneration operation schedule. In this study, some rooms interchange electric power and heat with each other for high efficiency and reduction of energy loss. The authors determine an operation schedule of FCs by multi-evaluation from viewpoints of energy cost and CO2 emissions.

Influence analyses of harmonics on distribution system in consideration of non-linear loads and estimation of harmonic source

AbstractRecently, non-linear loads have diversified and increased. It results in deterioration of power quality on a distribution system, since the equipment, which includes the non-linear loads, generates harmonics. The harmonics cause not only deterioration of power quality but also further various problems. Nowadays, the state of harmonics is becoming evident by the recent researches. However, since the harmonic analysis by using a detailed model of distribution system is too difficult, the analysis is carried out by a simplified model. In this paper, the authors construct an analytical distribution system model in consideration of the non-linear loads with the harmonic sources based on the practical data. Moreover, the influence analyses of harmonics are carried out by using this model. In addition, the authors estimate the harmonic sources in the simplified distribution system.

Examination for Experimental Verification of Improvement of Power Quality in Distribution System by PCS

In the future, the power quality will decrease by the introduction of a lot of renewable energy sources. The topic of this research is a new method of operation of PCS (power conditioning systems) in the future distribution system. The purpose of this research is development of PCS with a function of improvement of the distribution system. Therefore, the authors propose a method of the power quality improvement of the distribution system by PCS. In addition, the authors construct the control logic to use in PCS. The control logic suggests adding harmonic restraint function to conventional control. These were verified by simulation and an experiment. As the results, we confirmed that basic operation of PCS being carried out, harmonics were restrained, and power quality had improved.

FRT Control Method Including Islanding Detection in PCS Model for Photovoltaic Generation

In this study, we made experiments by the simulated PV system for understanding the cause of disconnection when instantaneous voltage sag occurs. Moreover, it is understood that the PCS with the voltage phase jump detection method which is one of the main causes of the disconnection. We modeled it in the simulation. When fault occurs, even though the residual voltage is more than 20 % and the voltage sag time is within 1.0 sec, the PCS disconnected by the voltage phase jump detection was confirmed in the simulation results. If a large number of PV systems are interconnected in the power system, we verified through the simulation that the interaction caused by the disconnection which generated by the voltage phase jump detection. We proposed threshold limit value of islanding detection. The validity of proposed method was verified through the simulation in this study.

Building Mathematical Model of Campus through Measurement Data

In this paper, we proposed the mathematical model of campus by measurement data and theoretical approach. We measure dynamic characteristic and static characteristic of load in a system. Power capacitors in the university are assumed to be electric power customers load. The load model is presumed by combining them and the linear model calculated theoretically. The line voltage and the line current are measured, and we calculate the line load capacity in distribution system of campus. The load fluctuation of distribution system during a day is observed. The load will be modelized by evaluating a lot of data. We can forecast the demand of electric power by few information that is the load model and the measurement data of receiving end.