Optimization of direct-regulated wind power automatic generation control (AGC) system and grid online optimal dispatch system requires the establishment of an accurate wind farm available power estimation model. Based on the new energy stand-alone information management system, this paper put forward a method of accurate estimation of wind power plant considering power generation conditions and station losses. The Spearman correlation coefficient was used to determine the number of key historical moments in the estimation moment, and the theoretical power estimation model of wind turbines was established based on the long short-term memory (LSTM) network. The operating conditions of the wind turbines were subdivided into six types (named as waiting wind, power generation and outage etc.), and the equivalent circuit model of the loss in the wind farm was established. Finally, the actual data of a wind farm was used to perform simulation calculations. The calculation results show that the root mean square error is reduced by 40% when the theoretical power model of a single machine considers the historical wind speed;When the available power model of wind farm takes into account the power generation condition and in-station loss, the root square error is reduced by 76.9%. The available power estimation model for wind farms proposed in this paper will facilitate the optimization of online dispatching and direct-regulation wind power AGC system strategies, and improve the level of wind power consumption.
