For quantitative description of turbulence wind field and solving the blindness and inaccuracy of turbulent wind spectrum model, the wind speed fluctuation curve is generated based on different turbulent wind spectrum models and considering turbulence intensity, the influence of surface roughness and space height. Based on the technology of image recognition and box-counting dimension, the fractal dimension of time series curve is calculated. The results show that the different turbulence wind spectrum model have different fractal characteristics. Turbulent wind is affected by turbulence intensity and surface roughness. Under the same physiognomy and climate conditions, the wind speed time series data of different height have different fractal dimensions.