“…The former, known as a machine learning algorithm, follows the load by changing various weights relative to the inputs data such as temperature and time. However, NN is easy to be influenced by the end of the data and is likely to enlarge the forecasting error in the case forecasting ultra-long tenn of more than one year [4]. The latter, which is a high-dimensional analysis algorithm, is not a suitable model for applying to multi-region because it needs many input data [5].…”
With spread of photovoltaic (PV) system in power systems, it will become more difficult to keep supply-demand balance of the power in mUlti-region by control of only the system side. Cooperation between the centralized energy management and distributed energy management is important for improving the supply-demand balance. For this purpose, it is necessary to accurately forecast the load and the PV power generation in a region. However, forecasting of residential load is very difficult because of steep fluctuations. In addition, we cannot obtain every residential load data. So we need to develop a basic model that forecasts the load of mUlti-regions without using actual measured data in all of the regions.In this paper, we analyzed the main component of the load to create a load estimation model. The load response associated with temperature fluctuations was modeled per time resolution of one hour by the main component analysis. We also estimated aggregated residential loads of multi-region. On the other hand, PV power generation was chosen from a similar day based on irradiance and temperature of multi-region. As a result, a data set for large-scale power system control simulation was made, and a stable power supply from the system side will be possible by demand response (DR) using demand-side storage batteries.
“…The former, known as a machine learning algorithm, follows the load by changing various weights relative to the inputs data such as temperature and time. However, NN is easy to be influenced by the end of the data and is likely to enlarge the forecasting error in the case forecasting ultra-long tenn of more than one year [4]. The latter, which is a high-dimensional analysis algorithm, is not a suitable model for applying to multi-region because it needs many input data [5].…”
With spread of photovoltaic (PV) system in power systems, it will become more difficult to keep supply-demand balance of the power in mUlti-region by control of only the system side. Cooperation between the centralized energy management and distributed energy management is important for improving the supply-demand balance. For this purpose, it is necessary to accurately forecast the load and the PV power generation in a region. However, forecasting of residential load is very difficult because of steep fluctuations. In addition, we cannot obtain every residential load data. So we need to develop a basic model that forecasts the load of mUlti-regions without using actual measured data in all of the regions.In this paper, we analyzed the main component of the load to create a load estimation model. The load response associated with temperature fluctuations was modeled per time resolution of one hour by the main component analysis. We also estimated aggregated residential loads of multi-region. On the other hand, PV power generation was chosen from a similar day based on irradiance and temperature of multi-region. As a result, a data set for large-scale power system control simulation was made, and a stable power supply from the system side will be possible by demand response (DR) using demand-side storage batteries.
“…These research works mostly concern the total demand forecast of EMS operation planning [7][8][9]. There is no practical local demand forecasting method.…”
Section: Case 1: Study On a Basic Nn Model For Local Demand Forecastmentioning
“…As described in Section 3, prediction of smoothed values of electric power will be effective for reduction of the storage capacity. Up to now, although the neural network technique has been widely used for prediction processing [5], accuracy of this method is likely to be greatly affected by learning conditions, and the whole procedure becomes rather complicated. And, as far as the authors know, this method has not yet been applied to dispersed power supplies using new energies.…”
SUMMARYIn this paper, the modified Euler type Moving Average Prediction (EMAP) model is proposed in order to operate a dispersed power supply system using new energy in autonomous mode. Furthermore, the EMAP model is used to operate a new type of dispersed power supply system consisting of a large-scale photovoltaic system (PV), a fuel cell (FC), and a small-scale superconducting magnetic energy storage system (SMES). This distributed power supply system can meet the multiple-quality electric power requirements of customers, and ensures voltage stability and UPS (Uninterruptible Power Supply) functions as well. Each subsystem of this distributed power supply contributes to the above-mentioned system performance with its own excellent characteristics. Moreover, the response characteristics of this system are confirmed by simulation with PSIM software, and, using collaborative operation under the EMAP model, the required capacity of SMES to compensate the fluctuation of both PV output and load demand is examined by simulation, using MAT-LAB/Simulink.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.