• Energy Research

Abstract The potential of long short-term memory network on ultra-short term wind speed forecast attracted attentions of researchers in recent years. Extending a probabilistic long short-term memory network model to provide an uncertainty estimation than to make a point forecast is more valuable in practice. However, due to complex recurrent structure and feedback algorithm, large scale ensemble forecast based on resampling faces great challenges in reality. Instead, a reliable forecast method needs to be devised. Gaussian process regression is a probabilistic regression model based on Gaussian Process prior. It is reasonable to integrate Gaussian process regression with long short-term memory network for probabilistic wind speed forecast to leverage the superior fitting ability of the deep learning methods and to maintain the probability characteristics of Gaussian process regression. Hence, avoid the repeated training and heavy parameter optimization. The method is evaluated for wind speed forecast using the monitoring dataset provided by the National Wind Energy Technology Center. The results indicated that the proposed method improves the point forecast accuracy by up to 17.2%, and improves the interval forecast accuracy by up to 18.5% compared to state-of-the-art models. This study is of great significance for improving the accuracy and reliability of wind speed prediction and the sustainable development of new energy sources.

Abstract The potential of long short-term memory network on ultra-short term wind speed forecast attracted attentions of researchers in recent years. Extending a probabilistic long short-term memory network model to provide an uncertainty estimation than to make a point forecast is more valuable in practice. However, due to complex recurrent structure and feedback algorithm, large scale ensemble forecast based on resampling faces great challenges in reality. Instead, a reliable forecast method needs to be devised. Gaussian process regression is a probabilistic regression model based on Gaussian Process prior. It is reasonable to integrate Gaussian process regression with long short-term memory network for probabilistic wind speed forecast to leverage the superior fitting ability of the deep learning methods and to maintain the probability characteristics of Gaussian process regression. Hence, avoid the repeated training and heavy parameter optimization. The method is evaluated for wind speed forecast using the monitoring dataset provided by the National Wind Energy Technology Center. The results indicated that the proposed method improves the point forecast accuracy by up to 17.2%, and improves the interval forecast accuracy by up to 18.5% compared to state-of-the-art models. This study is of great significance for improving the accuracy and reliability of wind speed prediction and the sustainable development of new energy sources.

Abstract Reliable streamflow forecasts are very significant for reservoir operation and hydropower generation. But for monthly streamflow forecasting, the forecasting result is unreliable and it is hard to be utilized, although it has a certain reference value for long-term hydro generation scheduling. Current researches mainly focus on deterministic scheduling, and few of them consider the uncertainties. So this paper considers the forecasting error which exists in monthly streamflow forecasting and proposes a new long-term hydro generation scheduling method called forecasting dispatching chart for Xiluodu and Xiangjiaba cascade hydro plants. First, in order to consider the uncertainties of inflow, Monte Carlo simulation is employed to generate streamflow data according to the forecasting value and error distribution curves. Then the large amount of data obtained by Monte Carlo simulation is used as inputs for long-term hydro generation scheduling model. Because of the large amount of streamflow data, the computation speed of conventional algorithm cannot meet the demand. So an improved parallel progressive optimality algorithm is proposed to solve the long-term hydro generation scheduling problem and a series of solutions are obtained. These solutions constitute an interval set, unlike the unique solution in the traditional deterministic long-term hydro generation scheduling. At last, the confidence intervals of the solutions are calculated and forecasting dispatching chart is proposed as a new method for long-term hydro generation scheduling. A set of rules are proposed corresponding to forecasting dispatching chart. The chart is tested for practical operations and achieves good performance.

Abstract Reliable streamflow forecasts are very significant for reservoir operation and hydropower generation. But for monthly streamflow forecasting, the forecasting result is unreliable and it is hard to be utilized, although it has a certain reference value for long-term hydro generation scheduling. Current researches mainly focus on deterministic scheduling, and few of them consider the uncertainties. So this paper considers the forecasting error which exists in monthly streamflow forecasting and proposes a new long-term hydro generation scheduling method called forecasting dispatching chart for Xiluodu and Xiangjiaba cascade hydro plants. First, in order to consider the uncertainties of inflow, Monte Carlo simulation is employed to generate streamflow data according to the forecasting value and error distribution curves. Then the large amount of data obtained by Monte Carlo simulation is used as inputs for long-term hydro generation scheduling model. Because of the large amount of streamflow data, the computation speed of conventional algorithm cannot meet the demand. So an improved parallel progressive optimality algorithm is proposed to solve the long-term hydro generation scheduling problem and a series of solutions are obtained. These solutions constitute an interval set, unlike the unique solution in the traditional deterministic long-term hydro generation scheduling. At last, the confidence intervals of the solutions are calculated and forecasting dispatching chart is proposed as a new method for long-term hydro generation scheduling. A set of rules are proposed corresponding to forecasting dispatching chart. The chart is tested for practical operations and achieves good performance.