Reliability Evaluation Techniques And Their Applications For Wind-integrated Power Systems Considering Complicated Wind Characteristics

With the accelerated depletion of fossil energy resources and growing awareness of environmental concerns,many countries around the world are trying their best to develop renewable energy resources.Compared to geothermal energy and tidal energy,wind energy is widely distributed,low-cost,and technically mature.In addition,many countries have established considerable incentive policies for developing and utilizing wind energy.All these facts have sparked a rapid growth of wind power capacity.Owing to the regional and seasonal differences of wind energy resources,wind has complicated characteristics.The wind speed probability distribution has the characteristics of diversity,and wind direction randomly varies with wind speed.Traditional reliability evaluation techniques have not adequately considered these characteristics and analyzed their impacts on power system reliability performance.Consequently,it is of great significance and importance to the optimal planning and operation of wind-integrated power systems by studying the complicated wind characteristics of wind energy resources,establishing wind models considering these characteristics,and analyzing their impacts on power system reliability performance.A mixture kernel density model for estimating wind speed probability distribution is proposed.The model considers the regional and seasonal differences of wind speed probability distribution.It uses optimal weight coefficients,which are obtained by combining multiple kernel densities into the mixture kernel density and a Lagrange method to minimize the asymptotic integrated mean squared error.The mixture kernel density model is ensured to accurately estimate wind speed probability distribution,and is applied to estimate wind speed probability distributions of eight wind farms from different regions and seasons at North Dakota in U.S.The results indicate that the proposed model demonstrates better goodness-of-fit than those of six conventional models,and verifies the accuracy of the proposed model.To consider the fact that wind direction varies randomly with wind speed,a wind speed-direction Markov chain model is proposed.This model classifies wind speed-direction states using Beaufort wind speed scale and cardinal wind direction scale.A transition rate matrix of wind speed-direction states is calculated and used to describe the random transition relationship between these states.The wind speed-direction Markov chain model is used to establish wind speed-direction model for four wind farms.The results indicate that the proposed model is capable of considering the basic statistical characteristics of actual wind speed-direction,such as mean value,standard deviation,cross-correlation coefficient,and so on.And also shows that the probability distribution of the wind speed-direction state duration follows the exponential distribution,and the state transition process of the states is a Markov process.Owing to the regional and seasonal differences of wind energy resources,wind speed probability distribution has the characteristics of diversity.To evaluate its impact on the power system reliability performance,a reliability evaluation technique for wind-integrated power system considering the diversity of wind speed probability distributions is proposed.This technique applies a wind speed sampling method to model the wind speeds using a spline interpolation.It enables to produce wind speed samples subject to the statistical characteristics of actual wind speed probability distributions with high accuracy.Then,a wind farm power output sampling method is proposed by considering the random failures of wind turbine generators(WTGs)and combining a two-state reliability model of WTG and a wind energy conversion model.Finally,combined with a load model and a reliability model of conventional generating units,a reliability evaluation technique for wind-integrated power system using non-sequential Monte Carlo simulation method is proposed to evaluate the reliability performance of RBTS system containing eight different wind farms.Simulation results indicate that wind speeds produced by the proposed model can accurately maintain the statistical characteristics of actual wind speed probability distributions.Moreover,the regional and seasonal differences of wind speed probability distributions have significant impacts on power system reliability performance.Owing to the fact that wind direction varies randomly with wind speed,the wake effect on the wind speeds of a wind farm may be different at distinctive time period.To evaluate its impact on power system reliability performance,a reliability evaluation technique for wind-integrated power system considering the wake effect is proposed.The method incorporates the wind speed-direction Markov chain model,two-state reliability model of WTG,Jensen wake model,and wind energy conversion model into a wind farm generation Markov chain model considering wake effect and random failure of WTG.Then,a wind-integrated power system generation Markov chain model is developed by combining with load model and reliability model of the conventional generating units.The probability and power output of each generation state can be calculated by the Markov equations.The proposed technique has been applied to evaluate the reliability performance of RBTS system containing four different wind farms.The results indicate that the wake effect has a negative impact on power system reliability performance.The power system reliability performance will be underestimated without considering the wake effect.An appropriate WTG power optimization strategy is helpful to improve the wind farm generation potential and system reliability performance.Complicated wind characteristics have considerable impact on WTG power,which raises a serious challenge to WTG power optimization.To address this challenge,a feed-forward neural network is employed to fit the WTG power function from the actual WTG operation data and a WTG power optimization model for each WTG operation point is proposed.The point-to-point and clustering WTG power optimization strategies for WTG operation period are developed.The latter strategy employs the K-means clustering technique on the basis of former strategy,which reduces the computational complexity and enables the real-time implementation.Simulation results on the reliability evaluation of the wind-integrated RBTS system with and without using the proposed WTG power optimization strategies show that the proposed WTG optimization strategies improve the WTG power output and the power system reliability performance of the wind-integrated RBTS system.