Research On The Optimal Reactive Power Allocation Of Power Systems With Random Wind Power Generation

The increasing wind power capacity not only influences the security, stability, andeconomic operation of the power system, but also poses a challenge to the traditionalpower system programming methods. With its random characteristics which differs windpower from traditional power generators, more uncertainty of reactive power, voltage aswell as voltage stability problems will be brought about to the power system, which isdifficult to evaluate with deterministic analysis method. This paper focuses on analyzingthe integration of multiple wind farms to the power system, establishes the stochasticmodel of the correlated wind speeds and correlated loads, and researches on reactivepower optimization problems in different aspects with wind power integrated, based onthe two models previously proposed.Two models for wind speed and load simulation are proposed, based on the analysisand summary of wind speed stochastic simulation methods. A bivariate wind speed timeseries model was proposed with the consideration of the capability of ARMA methods forsimulating the correlation of variables. Another model for wind speed and load based onLatin Hypercube Sampling was also proposed for its featured sampling efficiency. Thesetwo models are verified for its validity and compared for their characteristics andapplicability.As with the impact of the randomness of wind power being highlighted, probabilisticpower flow is becoming a hot research field, characterized by simulation methods basedon Monte Carlo Simulation. However, MCS based probabilistic power flow is difficult tomodel the timing characteristics of wind power, which leads to its infeasibility whendealing with short-medium term reactive power operation. The bivariate wind speed timeseries model proves to well fit the auto-correlation of wind speed of every single windfarm, and the cross-correlation of wind speeds between multiple wind farms. Studies werecarried out with the simulated wind speed data on the problems of optimal reactive poweroperation, and the results prove that the bivariate model can fit the original wind farmspeed characteristics well. And the results can be used a reference to the optimal reactivepower operation problem of power systems with the integration of stochastic wind power.As an important branch of VAR planning problems, optimal reactive power allocationserves to solve the unreasonable configuration of reactive power compensation sources,and plays an important part in power systems with plenty wind power generation and outstanding voltage problems. Probabilistic power flow based on Latin HypercubeSampling can take into account various random factors comprehensively, and thecorrelations, probabilistic distribution of wind speed, load as well. The proposed reactivepower allocation method has been proved to be effective in reducing investment as well asensuring system voltage stability. So, this model could have certain practical significanceon power system reactive power allocation, with large-scale randomly integrated windpower.