The economic value of wind power forecasting: a data-driven method and its application in various scenarios

The penetration of wind power is rapidly increasing under the carbon neutrality target, and the natural wind's randomness and volatility bring significant operational uncertainties to the system. Enhancing the accuracy of wind power forecasting is one of the effective means to reduce the impact of uncertainty. However, it's currently unclear how forecasting errors impact under different scenarios and to what extent the enhancement in forecasting accuracy can contribute to reducing operating costs. To address these issues, this paper proposes a method for conducting quantitative study on the economic value of wind power forecasting. Firstly, a data-driven method is developed to generate wind power time series under specified forecasting errors, serving as crucial inputs for subsequent evaluation. Compared to the traditional Monte Carlo generation model, the proposed method can maintain the mapping relationship between forecasted and measured data in actual conditions, thereby achieving better generation results. Then, a novel quantitative method is constructed from the perspectives of day-ahead economic dispatch and post-evaluation of real operating costs. Finally, the influence of wind power forecasting under various wind penetrations and energy storage allocation ratios is quantitatively studied. IEEE Three-machine Nine-bus system is taken as an example, the results indicate that when wind power penetration is no more than 50 % and the forecasting error is already less than 12 %, the economic benefits brought by further improving accuracy are limited. When the penetration exceeds 50 %, operating costs show a nearly linear downward trend as the forecasting error decreases, improving the accuracy is an eternal theme.