Study On Fast Convergence For Wind Farm Integrated Generation System Reliability Evaluation

Compared with analytical method, Monte-Carlo method is easier to find multiple outages and simplifies the modeling process at the same time, therefore, it gains more and more popularity in the power system reliability. However, Monte-Carlo needs a large amount of sampling times and more time to converge. Usually, facilities multiply in wind farm integrated generation system, computation time will greatly increase, which makes the reliability evaluation more difficult. This paper researches into this problem based on variance reduction techniques, and compares the convergence rate of different sampling ways. Main aspects of the paper are as follows:Firstly, this paper introduces necessary modeling work for power system reliability integrated with wind power, which involves working principle of wind turbine generator, wind speed forecast, output power of wind turbine generator and wake effect of wind farm. Distribution model and sequential model of wind speed are attached with more importance. The work above lays a foundation for reliability evaluation.Secondly, introduces advantages and disadvantages of Monte Carlo simulation, elaborates sequential and non-sequential Monte Carlo method of generation system reliability evaluation. Analyzing convergence of generation system reliability evaluation considering derated state by central limit theorem, confidence interval becomes narrow as sampling adds. The convergence curve comparison between IEEE RBTS and IEEE RTS79 explains convergence characteristics of Monte Carlo method.Next, sequential and non-sequential Monte Carlo method of wind farm integrated generation system reliability evaluation illustrates influence of integration of wind power on system reliability. Reliability evaluation of wind power integrated system needs more sampling times and computing time than conventional system.Lastly, Variance reduction technique can well cope with the problem above, realize fast convergence and improve sampling efficiency greatly, this paper researches into several sampling methods which realize fast convergence in reliability evaluation of generation system integrated with wind farm, and explains their flow charts by operation steps. The existing average and scattered sampling can reduce sampling times, however, add the computing time at the same time. This paper proposed control variate method, total probability formula method and mixed sampling method for wind farm integrated generation system reliability evaluation. Control variate method can greatly improve sampling efficiency and reduce computing time largely compared with average and scattered sampling method, it’s greatly influenced by wind power penetration. Total probability formula method is an advantageous method which is very convenient to apply. Mixed sampling of average and scattered method and total probability formula method has a greater efficiency than the methods above, which highlights superiority of mixed method and provides a possible direction for fast convergence study.