仿真程度高且简洁实用的风模型，对于风力机系统的气动与结构设计、风场布置与发电功率预测等十分重要。实际风速时间序列普遍具有自相似性，传统的谐波叠加法生成风速不具有自相似特征。基于随机型Weierstrass-Mandelbrot函数，通过其功率谱与Kaimal脉动风速功率谱建立联系，生成具有自相似分形特性的风速时间序列。将该方法生成风速与谐波叠加法生成风速从功率谱、概率密度函数及均值、极值、标准差等统计参数方面进行比较。结果表明：相比于谐波叠加法，随机型W-M函数能有效仿真具有自相似分形特性的风速时间序列。

It is of importance for wind energy system design and simulation to have accurate,simple and practical wind model that may improve aerodynamic and structural design,wind farm layout and electricity generation prediction. Actual wind velocity time series usually have self-similarity.Wind velocity generated by traditional harmonic wave superposition method does not have such selfsimilarity. Based on stochastic WeierstrassMandelbrot function,wind velocity time series with self-similarity can be generated according to Kaimal wind spectrum. Comparing the wind velocity time series with that generated by harmonic wave superposition method from power spectrum,and statistic characteristics such as probability density function,mean and variance,the results showed the two series have almost the same characteristics,indicating W-M function method can also effectively simulate wind velocity time series with self-similarity.

TK81

国家自然科学基金资助项目（51176129）；上海市科学技术委员会项目资助（13DZ2260900）