Research On Stochastic Optimal Dispatch Of Power System With Multiple Wind Farms

In order to solve the problem of energy shortage and environmental pollution, renewable energy generation is widely concerned. Wind becomes an important form of renewable energy generation due to its non-pollution, renewablity and other excellent features. However, unlike traditional energy generation, the randomness and intermittence of the wind speed, making wind power also has a random, non-scheduling feature, which brings enormous difficulties and challenges to the safe and economic operation of power system, when large-scale wind power incorporated into power grid. Therefore, studying optimal scheduling problem of power systems with wind power is of great theoretical significance and application value.Power converted by wind must be in command and control of the dispatching system with such links as transmission, transformation, distribution and use, ultimately you can feed to the users. However, wind power depends entirely on the wind conditions, present strong randomicity, intermittent, periodicity and volatility. Along with the increase of wind turbines capacity installed, the existing technical level also unable to accurately forecast the wind power, which makes the wind power scheduling more difficult. In the optimal scheduling of power system with wind power, it is necessary to calculate the Probability Distribution of wind power from wind farm, in order to assess the maximum wind power that can be incorporated into the system, so that to achieve security and economic dispatch of the power syetem. However, with the construction of Smart Grid, large-scale wind power connected to power grid, and wind at different locations may come from the same origin, so that their wind power has a significant degree of correlation. So that outputs from multiple wind fams characteristics is different from single wind farm, therefore it is necessary to analysis the joint distribution for outputs from multiple wind fams. In view of this, this paper uses Weibull distribution function to characterize the probability distribution of wind speed, and summarizes the estimations for the shape parameter and scale parameter of Weibull distribution. Also, based on the power output function of the wind-turbine, pushs the probability distribution of the power from a single farm. Then according to the basic theory of Copula function and the characteristic analysis for the output from multiple wind farms, the Gumbel-Copula function is employed to characterize the Joint Probability Distribution(JPD) of wind power from multiple wind farms. And puts forward the goodness-of-fit testing method, so as to provide a powerful tool to characterize JPD and depicting its tail correlation for outputs from multiple wind farms.Moreover, since the randomness and uncertainty wind power, we can not get exact information of the wind power from farms when we plan to develop a scheduling about the system, which reduces the reliability of the scheduling. Therefore, to ensure the security, economy, flexibility and robustness of system operation, including power transmission capacity constrains of the lines, an optimal dispatch model of power system with multiple wind farms is proposed based on Chance Constrained Programming (CCP),which describes the randomness and uncertainty wind power in the probability form. For the chance constraints optimization model with uncertain facters, the general is to convert them into certainties for solutions. Sample Average Approximation (SAA) method is used to transform the CCP to non-continuous, non-differentiable, computable and certainty problem, then to find the existing optimal method for the solutions. For this, we use the improved Particle Swarm Optimization (PSO) algorithm with better solving rate and inertia factor.Through the two wind farms in Netherlands, for example, show that the Gumbel-Copula distribution can better describe the JPD of the output from wind farms, improve the estimation precision output from wind farms and effectively describe the relevance of its tail. Inspection based on IEEE-9 nodes system showed that the proposed stochastic optimal model and its transformation methods improve the model solving rate, and increase the feasibility, flexibility, robustness of the system scheduling, which provides a theoretical basis for optimal scheduling with large-scale wind incorporated power systems and provides a powerful tool for system dispatcher to make a reasonable optimal scheduling quickly under conditions with uncertainty information.