Abstract-Accurate and reliable forecast of wind power is essential to power system operation and control. However, due to the nonstationarity of wind power series, traditional point forecasting can hardly be accurate, leading to increased uncertainties and risks for system operation. This paper proposes an extreme learning machine (ELM)-based probabilistic forecasting method for wind power generation. To account for the uncertainties in the forecasting results, several bootstrap methods have been compared for modeling the regression uncertainty, based on which the pairs bootstrap method is identified with the best performance. Consequently, a new method for prediction intervals formulation based on the ELM and the pairs bootstrap is developed. Wind power forecasting has been conducted in different seasons using the proposed approach with the historical wind power time series as the inputs alone. The results demonstrate that the proposed method is effective for probabilistic forecasting of wind power generation with a high potential for practical applications in power systems.Index Terms-Bootstrap, extreme learning machine (ELM), forecasting, prediction interval, wind power.
NOMENCLATUREBiases of ELM, approximated biases.Number of bootstrap replications.Number of bootstrap replications for model uncertainty estimation.
Abstract-Accurate and reliable wind power forecasting is essential to power system operation. Given significant uncertainties involved in wind generation, probabilistic interval forecasting provides a unique solution to estimate and quantify the potential impacts and risks facing system operation with wind penetration beforehand. This paper proposes a novel hybrid intelligent algorithm approach to directly formulate optimal prediction intervals of wind power generation based on extreme learning machine and particle swarm optimization. Prediction intervals with associated confidence levels are generated through direct optimization of both the coverage probability and sharpness to ensure the quality. The proposed method does not involve the statistical inference or distribution assumption of forecasting errors needed in most existing methods. Case studies using real wind farm data from Australia have been conducted. Comparing with benchmarks applied, experimental results demonstrate the high efficiency and reliability of the developed approach. It is therefore convinced that the proposed method provides a new generalized framework for probabilistic wind power forecasting with high reliability and flexibility and has a high potential of practical applications in power systems.
The electricity market plays a key role in realizing the economic prophecy of smart grids. Accurate and reliable electricity market price forecasting is essential to facilitate various decision making activities of market participants in the future smart grid environment. However, due to the nonstationarities involved in market clearing prices (MCPs), it is rather difficult to accurately predict MCPs in advance. The challenge is getting intensified as more and more renewable energy and other new technologies emerged in smart grids. Therefore transformation from traditional point forecasts to probabilistic interval forecasts can be of great importance to quantify the uncertainties of potential forecasts, thus effectively supporting the decision making activities against uncertainties and risks ahead. This paper proposes a hybrid approach to construct prediction intervals of MCPs with a two-stage formulation. In the first stage, extreme learning machine (ELM) is applied to estimate point forecasts of MCPs and model uncertainties involved. In the second stage, the maximum likelihood method is used to estimate the noise variance. A generalized and comprehensive evaluation framework for probabilistic electricity price forecasting is proposed in this paper. The effectiveness of the proposed hybrid method has been validated through comprehensive tests using real price data from Australian electricity market.Index Terms-Artificial neural network, bootstrap, electricity price forecasting, maximum likelihood estimation, prediction intervals.
Abstract-This letter proposes a novel efficient probabilistic forecasting approach to accurately quantify the variability and uncertainty of the power production from photovoltaic (PV) systems. Distinguished from most existing models, a linear programming based prediction interval construction model for PV power generation is proposed based on extreme learning machine and quantile regression, featuring high reliability and computational efficiency. The proposed approach is validated through the numerical studies on PV data from Denmark.Index Terms-PV power, forecasting, prediction intervals, extreme learning machine, quantile regression.
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.