Analysis of some flexible families of distributions for estimation of wind speed distributions

Overall Description of Wind Power

“…Along with the Weibull distribution the Rayleigh distribution is one of the most used statistical distributions in reliability theory; is a special case of Weibull distribution, where the shape parameter is equal to two, and the scale parameter is √ 2 times of the scale parameter of the corresponding Weibull distribution [21]. Other distribution that have been reported in literature are Lognormal distribution, which is a special form of the normal distribution, Gamma and Erlang distributions which uses exponential approximations, inverse Gaussian distribution is also used and is popular in meteorology studies, additionally other alternatives for wind speed analysis include Skewed generalized error distribution, Skewed t distribution and Burr distribution [24]. In [25] and [26], Maximum Entropy Principle (MEP) is proposed to overcome the inaccuracy of Weibull and other distributions in some cases.…”

A comparative analysis of wind speed probability distributions for wind power assessment of four sites

“…A number of other PDFs have also been proposed by various researchers to characterise wind speed frequency distributions; they include gamma, lognormal, inverse Gaussian, beta, Burr, Wakeby, kappa, and hybrid distributions [12][13][14][15][16]. In [17], two flexible families of distributions, the skewed generalised error and skewed t, have been proposed for the description of wind speed. Both of these distributions were found to have the flexibility of accommodating the shape of the wind speed data, which included some well-known distributions as special cases.…”

“…Six probability density functions, namely Weibull, Rayleigh, gamma, lognormal, inverse Gaussian, and maximum entropy principle (MEP) derived PDFs, were evaluated [19] using six goodness-of-fit criteria for five representative sites in North Dakota. Statistical distributions have been analysed in the literature for examining differences between the results for day and night, between various seasons, and for monthly and yearly data [17,20]. The suitability of these distribution functions for a particular application depends upon a number of factors such as the type of wind regime, availability of data, and recording intervals [3,16].…”

“…Mixture distributions have been used to fit the bimodal shape of wind speed histograms [9][10][11]. A number of other PDFs have also been proposed by various researchers to characterise wind speed frequency distributions; they include gamma, lognormal, inverse Gaussian, beta, Burr, Wakeby, kappa, and hybrid distributions [12][13][14][15][16]. In [17], two flexible families of distributions, the skewed generalised error and skewed t, have been proposed for the description of wind speed.…”

“…In [16,18], different distribution functions are analysed for their suitability in offshore applications and in [18] Johnson S E distribution has been introduced for the first time. In [13], seven different distributions were analysed to identify suitable distribution for an urban area and the Burr distribution recently applied to wind speed problems [12] was found suitable for urban applications. Six probability density functions, namely Weibull, Rayleigh, gamma, lognormal, inverse Gaussian, and maximum entropy principle (MEP) derived PDFs, were evaluated [19] using six goodness-of-fit criteria for five representative sites in North Dakota.…”