Uncertainty Evaluation Of Wind Power Ramping And Multi-objective Reactive Power Optimization Of Wind Storage Combined

With the gradual increase in the permeability of wind power in the power system,which had brought enormous economic and social benefits to the society,the uncertainty inclusive randomness and volatility of wind power output also had a great impact on the safe and stable performance of the power system.Particularly,with the frequent occurrence of extreme climate and wind power ramp events,the power imbalance of the power system might be easily caused,or even the limits of power voltage or frequency might be exceeded if the deviation enlarged,then,the secure and stable performance of the power system would be severely threatened.Therefore,improving the accuracy degree of wind power ramp prediction and identification,quantitatively evaluating the uncertainty of wind power ramp output,and providing a useful solution to stabilize the wind power ramp events,would be helpful to improve the voltage distribution of the power system with a high proportion of wind power,reducing the active power loss,increasing the wind power consumption,promoting the peak emissions of carbon dioxide,and achieving the carbon neutrality.On account of the problems that the prediction accuracy is reduced due to the ineffective decomposition of the high frequency components in the process of wind power ramping prediction,the time-domain modeling is vulnerable to noise interference and the frequency-domain modeling is difficult to detect the inflection point of ramp in the process of wind power ramp identification,a method for wind power ramp event prediction and identification considering time-frequency characteristics was provided in this article.Firstly,based on wavelet packet transform,the wind power ramp sequence was preprocessed to guarantee the decomposition of high-frequency components,and the noisy high-frequency components was denoised with the wavelet packet soft threshold denoising method.Secondly,a method for wind power ramp event prediction and identification considering the feature of time-frequency was proposed from the perspective of domain of time and frequency.Finally,the mean square root error,average relative error,and average absolute error of wind power ramp prediction were respectively reduced by 31.72%,7.94%,and 23.78% by means of actual wind power data case analysis and comparison with other methods.The comprehensive success index and accuracy rate of identification were respectively increased by 23.38% and8.77%,it means,the method proposed in this article would be helpful to effectively raise the wind power ramp prediction accuracy and the identification rate.Aiming to the difficulty in quantitatively evaluating the uncertainty of strong wind power ramp output,the wavelet packet variance entropy was proposed to evaluate the uncertainty of wind power ramp.Firstly,the random fluctuation wind power sequence was denoised and decomposed basis on the wavelet threshold denoising method and decomposition algorithm,then,the wind power variance sequence was obtained.Secondly,combined with the variance theory and the wavelet packet entropy theory,the wavelet packet variance entropy was proposed to measure the uncertainty of the wind power ramp.Finally,based on the actual wind power data,the occurrence of varied levels of wind power ramp events was evaluated,the result shown that the more severe of the wind power ramp fluctuation,the larger of the wavelet packet variance entropy,the smoother the wind power ramp fluctuation,and the smaller the wavelet packet variance entropy.On account of the problems inclusive active power loss and out of voltage limit resulted from the wind power ramp events,a multi-objective reactive power optimization model of wind-storage combination was built up in this paper.Basis on the results of wind power ramp prediction and identification in Chapter 2,the smallest wind power standard deviation(predicted value),the smallest active power loss,and the smallest voltage deviation were adopted as the objective function,the tap position of the on-load regulating transformer,the switching class number of the capacitor bank,the reactive output of the static var compensator,and the charge-discharge power of the energy storage system as the restrained variables.Intend to avoid the local convergence of the algorithm,the improved Pareto archive multi-objective particle swarm optimization algorithm based on niche technology was adopted to solve the problem.The peak-to-valley variation index of active power and the wavelet package variance entropy achieved in Chapter 3 were used as the evaluating indicators of wind power ramp event stabilization.The IEEE-33 node simulation shown that the various evaluating indicators of ramp stabilization on the wind power were better than that prior to the ramp stabilization,compared with unilateral reactive power optimization on the power grid,none of out of voltage limit event occurred,and active power loss declined to 13.21%.