Compayative Study Of Numerical Methods And Swarm Intelligence Algorithms On Optimizing Low Wind Speed Distribution Models

Recently, with energy crises and environmental problems around the world becoming increasingly prominent, the development and utilization of wind power have become a big concern. Due to regional imbalance of China’s economic development, it is hard and unable to consume a large amount of wind power for the local regions that are rich in wind energy resources but with undeveloped economies, which often leads to difficulties of wind power grid integration and difficulties of transportation. Meanwhile, China’s wind farm constructions in the area with high-quality wind energy resources have basically become saturated. Thus, the development of low-speed wind power has gradually risen to concern. The great Yangtze River Delta City Group, with developed economies, large population and huge energy consumption, is chosen as the study area in this paper.Therefore, in order to effectively explore the wind speed characteristics in the study area, Weibull, Rayleigh, Gamma and Lognormal are taken as four distribution models in assessing the low-speed wind energy resources. To obtain the unknown parameters in the distributions, three numerical methods-moment estimation (ME), maximum likelihood estimation (MLE), least squares method (LSM) are applied. Then bat algorithm (BA) is employed for the first time to optimize the parameters, and cuckoo search algorithm (CS) and particle swarm optimization (PSO) are taken into consideration. Additionally, parameter results of numerical methods and swarm intelligence algorithms are compared. The empirical results show that Weibull PDF displays better performance when fitting the actual frequency. Secondly, in comparison of numerical methods and swarm intelligence algorithms, parameters are more accurate optimized by swarm intelligence algorithms. Thirdly, PSO-Weibull and BA-Weibull perform the best. When assessing the low-speed wind energy resources, the average wind power density, the effective wind power density, the available factor and the capacity factor of wind turbine, are functions of the parameter in Weibull model. Consequently, accurate parameter estimation will lead to accurate and reasonable assessment, which demonstrates the importance of parameter estimation of the distribution model in wind energy resource assessment.