Hideaki Ohtake

On the Use of Maximum Likelihood and Input Data Similarity to Obtain Prediction Intervals for Forecasts of Photovoltaic Power Generation

The objective of this study is to propose a method to calculate prediction intervals for oneday-ahead hourly forecasts of photovoltaic power generation and to evaluate its performance. One year of data of two systems, representing contrasting examples of forecast’ accuracy, were used. The method is based on the maximum likelihood estimation, the similarity between the input data of future and past forecasts of photovoltaic power, and on an assumption about the distribution of the error of the forecasts. Two assumptions for the forecast error distribution were evaluated, a Laplacian and a Gaussian distribution assumption. The results show that the proposed method models well the photovoltaic power forecast error when the Laplacian distribution is used. For both systems and intervals calculated with 4 confidence levels, the intervals contained the true photovoltaic power generation in the amount near to the expected one.

Evaluation of unit commitment based on intraday few-hours-ahead photovoltaic generation forecasts to reduce the supply-demand imbalance

Japanese power systems are expected to increase the installation of photovoltaic (PV) generation in the future. We have developed operation and control methods for future power systems based on PV generation forecasting. In our previous study, we proposed a supply-demand operation method based on the intraday few-hours-ahead forecasts of the PV power output and concluded that the unit commitment (UC) of thermal power generators needs to be modified to reduce the supply-demand imbalance. In this study, we developed a new supply-demand operation method in which the UC is repeatedly modified based on intraday few-hours-ahead forecasts. We evaluated the effectiveness of the proposed method by long-term (2 months) numerical simulations of a large power system model for Kanto Japan. Actual hourly data on the load demand and solar irradiation were used in the simulations.

Economic impact of photovoltaic power forecast error on power system operation in Japan

In this study, we analyzed the operational situation and cost for Tokyo Electric Power Companys service area by 2030, when a high penetration of variable renewable energy (VRE) sources will have been attained. Day-ahead unit commitment and real-time economic dispatch simulation were carried out by taking into account photovoltaic power yield forecasts and errors. The analysis allowed us to evaluate the economic impact of forecasting errors from VRE sources on the annual supply and demand balance within the target area. The analysis also showed that the use of day-ahead forecasting reduced the amount of unserved energy and shrank photovoltaic power yield curtailment in real-time operations.

Fast parallel calculation for optimal power demand control in multi-layer smart grids

This paper presents an optimal method for controlling power demand probabilities in a negawatt trading between aggregators in a multi-layer smart grid. The calculation algorithm can be parallelized by using the multi-layer structure; therefore, we can avoid computational effort explosions. In the power demand control, according to a cost distribution representing maximum probabilities at assumed total power demands of aggregators in a lower-layer, an independent system operator in the top layer can optimally allocate some desired variation of the distribution to each aggregator. The allocated distributions may be offered to aggregators as peak shifts and/or power restrictions with their benefits.