【摘要】 针对现有风电数据清洗算法在处理风电数据效率较低的问题，提出一种基于雨滴侵蚀算法的数据清洗方法。该算法通过模拟雨滴对地形的冲击和侵蚀过程，有效去除了风电数据中的异常值。实验结果表明，雨滴侵蚀算法的风速与功率的相关性系数达到0.977，在降低数据离散性方面的效果显著，且运行时间为2.1 s。另外，采用雨滴侵蚀算法进行数据清洗后，结合贝叶斯优化的支持向量回归（SVR）模型，拟合了单台风电机组的最佳风功率曲线，并据此计算了调度后的盈余电量。随后，使用质子交换膜（PEM）电解槽和碱性水（ALK）电解槽的仿真模型，对两者利用盈余电量制氢的耗电量进行对比分析。仿真结果表明，PEM电解槽的电耗为39.4 kW·h/kg,ALK电解槽的电耗为53.9 kW·h/kg,PEM电解槽更适合盈余电量制氢。更多还原

【Abstract】 Addressing the inefficiency of existing wind power data cleaning algorithms, a data cleaning method based on the raindrop erosion algorithm is proposed. This algorithm effectively removes outliers from wind power data by simulating the impact and erosion of raindrops on terrain. Experimental results indicate that the correlation coefficient between wind speed and power using the raindrop erosion algorithm reaches 0.977, demonstrating significant effectiveness in reducing data dispersion with a runtime of 2.1 seconds. Additionally, after cleaning the data using the raindrop erosion algorithm, the optimal wind power curve for a single wind turbine is fitted using a support vector regression(SVR) model with Bayesian optimization, from which the scheduled surplus electricity is calculated. Subsequently, simulation models of proton exchange membrane(PEM) electrolyzers and alkaline water(ALK) electrolyzers are used for comparative analysis of electricity consumption for hydrogen production utilizing surplus electricity. Simulation results show that the electricity consumption is 39.4 kW·h/kg for the PEM electrolyzer and 53.9 kW·h/kg for the ALK electrolyzer, suggesting that the PEM electrolyzer is more suitable for hydrogen production from surplus electricity.更多还原