关键词: 水头预测;水电站调度;Kriging代理模型;大、二江电站;葛洲坝电站;

Abstract: In the dispatching and operation of hydropower stations, water head is an important data index, which plays a key role in the safe and stable operation of units.After selecting non water abandonment period and water abandonment period, under different operation modes of Gezhouba Hydropower Station, the influence of output distribution and abandoned water discharge on the water head of Dajiang and Erjiang hydropower stations was analyzed.The Kriging surrogate model was introduced to establish a head prediction models in two typical periods.The predicted value of water head based on Kriging surrogate model was compared with the actual calculated value.The result shows that in the non water abandonment period, the average head deviations of Dajiang and Erjiang hydropower stations are 0.22 m and 0.26 m respectively, and in the abandonment period, their average head deviations are 0.20 m and 0.16 m respectively, namely the prediction accuracy meets the needs of real-time dispatching.The model aims to provide a new analysis method for the trend change law of hydropower station's water head, and then provide decision-making reference for real-time optimal dispatching.

Key words: water head prediction; operation of hydroelectric station; Kriging surrogate model; Dajiang and Erjiang Hydropower Stations; Gezhouba Hydropower Station;