Search

Accurate wind speed forecasting ensures the controllability for the wind power system. In this paper, a structure with density-weighted active learning (DWAL)-based model selection strategy from the perspective of meteorological factors is proposed to improve the accuracy and stability for wind speed deterministic and probabilistic forecasting. To improve training efficiency and accelerate the sample selection process, DWAL is employed. The multi-objective flower pollination algorithm is used to combine best models selected from model space with optimal weights for higher accuracy and stability. Except deterministic forecasts, as large-scale wind power generation integrated into power grid, the wind direction should also be forecasted and the estimation of the wind speed and direction uncertainty is vital, offering various aspects of forecasts for risk management. Thus, both deterministic and probabilistic forecasting for the wind speed vector are included in this paper. Eight datasets from Ontario Province, Canada, are utilized to evaluate forecasting performance of the model selection and the proposed structure. Results demonstrated: (a) the proposed structure is suitable for wind speed vector forecasting; (b) the proposed structure obtains more precise and stable forecasting performance; (c) the proposed structure improves the accuracy of deterministic forecasting and provides probabilistic information for wind speed vector forecasting. • Propose a novel forecasting structure based on meteorological analysis. • The model selection is employed. • The multi-objective flower pollination algorithm is used. • Density-weighted active learning algorithm is applied. • Both deterministic forecasting and probabilistic forecasting are applied. [ABSTRACT FROM AUTHOR]

Abstract: Electricity is an important asset that influences not only the economy, but political or social security of a country. Reliable and accurate planning and prediction of electricity demand for a country are therefore vital. In this paper, electricity demand in Ontario province of Canada from the year 1976–2005 is modeled by using an (adaptive neuro fuzzy inference system) ANFIS. A neuro fuzzy structure can be defined as an ANN (artificial neural network) which is trained by experimental data to find the parameters of (fuzzy inference system) FIS. Inputs for the model include number of employment, (gross domestic product) GDP, population, dwelling count and two meteorological parameters related to annual weather temperature. The data were collected and screened using statistical methods. Then, based on the data, a neuro-fuzzy model for the electricity demand is built. It was found that electricity demand is most sensitive to employment. [Copyright &y& Elsevier]

Area feeder long-term load forecast (LTLF) is one of the most critical forecasting tasks in electric distribution utility companies. Cost effective system upgrades can only be planned out based on accurate feeder LTLF results. However, the commonly used top-down and bottom-up LTLF methods fail to combine area and feeder information and cannot effectively deal with component-level LTLF. The previous research effort on hybrid approach that aims to combine top-down and bottom-up approaches is very limited. The recent work only focuses on the forecast of the next one-year and uses a one-fit-all model for all area feeders. In response, this paper proposes a novel selective sequence learning method that can convert a multi-year LTLF problem to a multi-timestep sequence prediction problem. The model learns how to predict sequence values as well as the best-performing sequential configuration for each feeder. In addition, unsupervised learning is introduced to automatically group feeders based on load compositions ahead of learning to further enhance the performance. The proposed method was tested on an urban distribution system in Canada and compared with many conventional methods and the existing hybrid forecasting method. It achieves the best forecasting accuracy measured by three metrics AMAPE, RMSE and R-squared. It also proves the feasibility of applying sequence learning to multi-year component-level load forecast. • Converts a multi-year LTLF problem to a multi-timestep sequence prediction problem. • Can incorporate area and feeder level information in multiple historical years. • Can be used to forecast multiple years in the future. • Best sequential configurations can be automatically learned for different feeders. • Unsupervised learning is adopted to further enhance forecasting accuracy. [ABSTRACT FROM AUTHOR]