Study On Wind Power Prediction,Synergistic Dispatch And Voltage Security Assessment Of Power Grid

The problems caused by conventional fossil energy consumption, carbon emissions and environmental pollution are becoming increasingly gravely, thereby, it has been widely attached importance to seek the clean and efficient renewable resources to replace the exhausting fossil energy. It plays an important role that using wind power, photovoltaic and the other renewable resources instead of fossil energy to reduce carbon emission and improve energy structure, however, the difficulty of the renewable resources connected to power grids is increasing for the renewable resources posses obvious intermittency and volatility. In recent years, the plug-in electric vehicles that are considered as decrease carbon emissions, economize conventional energy and reduce environmental pollution on the purpose, are randomly connected into power grids without guidance to charge for free, thereby, all of the above make both the power supply side and the load one emerge as the non-controllability, which bring a new challenge to the regulation of the power grids. So under the context of constructing a strong smart grids, in order to cope with the impact of Ultra High Voltage AC and DC, as well as the renewable resources and plug-in electric vehicles widely connecting to the power grids, carrying out the study on regulation and controlling of the power grids and safety assessment, which is regarded to reduce security and stability operation risk and improve the operation economics of power grids, has important economic and practical significance.Consequently, on the basis of analyzing wind power fluctuations law, regarding constructing a strong smart power grids as the core, this thesis carries out the following researches, such as wind power prediction, the collaborative dispatching between wind power and plug-in electric vehicles and the security analysis of power grids with wind power integration. The main works and innovative achievements of the thesis are as follows:(1) Chaotic theory is used to reveal the internal dynamic property of wind power time series. The largest Lyapunov exponent of wind power time series is calculated on the basis of phase space construction to verify the chaotic characteristics of wind power sets. The wind power forecasting would produce larger errors by using the Volterra filter multi-step prediction, thereby, the prediction results of Volterra filter are corrected by combining the ones predicted by Local-region Multi-steps Method and the Largest Lyapunov exponent method with weighted Markov chain and ordered operator, The simulation results illustrate that the correction forecasting model improves high predictive accuracy effectively under carrying out the ultra-short-term wind power prediction, which provides a useful reference for wind power forecasting by using the Volterra filter multi-steps method.(2) According to the short-term wind power prediction, a combined model of wind power prediction based on entropy, extreme learning machine(ELM) and reservoir(ESN), respectively. The wind power time series is decomposed into a series of sub-sequences with different characteristic scales, which is given a complexity analysis by using entropy to generate a new sub-sequence according to the classifying and superimposing with different entropies sub-sequence. Finally, the parameters and input vector dimensions of each learning machines are determined by cross validation and chaotic phase space theory. Then, the forecasting model of each subsequence is created with least squares support vector machine(LSSVM), ELM and ESN, respectively. The simulation results illustrate that the combining prediction model based on ELM and ESN is better in the training speed and forecasting accuracy than the one based on LSSVM method. In contrast to the chaotic theory, the proposed combining prediction models based on ELM and ESN that not only suit for the ultra-short-term forecasting but also can be used to short-term forecasting, have a higher prediction accuracy and provide a new useful reference for wind power forecasting in online engineering application.(3) Considering the extensive development of plug-in vehicles(PEVs) and wind power, an approach for PEVs charging dispatching in regional power grids with the uncertain outputs of wind power is proposed based on the wind power forecasting results. Firstly, in order to reduce the difference between the peak and the valley for equivalent load and purchasing power cost, a multi-objective non-linear mixed integer optimization model for PEVs charging dispatching is established. Secondly, the fuzzy theory is introduced to this paper to fuzzy the output of wind power and photovoltaic power. Therefore, the multi-objective fuzzy optimization model is reformulated as a single objective non-linear optimization problem. Thus, the data of example on regional power grid is analyzed to prove to the validity of model and the feasibility of solving for problems with improved particle swarm algorithm. An effective way is provided for the collaborative optimal dispatch of PEVs.(4) In the context of the influence on the safety and stability operation of power grids produced by wind power connected into power grids, considering the idea of overshoot, both overshoot and adjustment time on stability theory, assessments for voltage stability performance of power grid that contains fully the dynamic model characteristics, such as:signal energy aggregate index(SEAI) and cumulating index(CI), are proposed based on the signal energy method. On the basis of the dynamic simulation capabilities of PSS/E software to obtain the information of voltage amplitude of power grid, the two new criterions named as SEAI and CI are constructed considering different period signal energy spectrums to confirm the weak bus of voltage stability of power grids. Finally, the operation mode data about the winter of2010in Shandong power grid is used to compare to the conventional methods of voltage stability analysis to verify that the proposed methods have better credibility under the premise of full account of the power supply and fan models.