The distribution of wind energy is uneven, and improving the assessment method of wind energy resource characteristics to enhance its accuracy and comprehensiveness is crucial for wind farm construction and efficient use of wind energy. A modeling method is proposed for the joint probability distribution of wind speed and wind direction based on a binary exponential polynomial. The parameters of the binary exponential polynomial of this model are solved by using linear least squares. A normalization constant is added to make the binary exponential polynomial satisfy the characteristics of the probability density function. It combines multiple goodness-of-fit index functions to solve the optimal index of the binary exponential polynomial, so that obtains the optimal fitting performance of the joint probability distribution of wind speed and direction. The model is used to fit the measured data of wind farms in multiple regions and compared with the Copula model for verification. The results show that due to the more fitting parameters of the binary exponential polynomial, the proposed model is superior to Copula model in the aspects of root mean square error, coefficient of determination, Akaike information criterion and average absolute percentage error. It is proved that the fitting model based on the binary exponential polynomial can more accurately fit the wind speed and direction data of the wind farm.