| Popis: |
The investigation of wind energy describes a potential way forward in electricity generation in New Zealand. To meet the future demands of end users, the management of the uncertainty in wind power needs to be developed along with the integration into the power system. This thesis studies probabilistic models of wind power prediction in the short term period. Previous research topics pertaining to wind power demonstrate an understanding of its persistence and uncertainty properties. Several issues that the researchers consider to be relevant are studied before a new statistical model for wind power behaviour is derived. For example, the basic summary of the time series detects an autocorrelation with time lags; correlations to market factors reveal the general relationships between wind power and market components with different measurements and time frames; a Weibull model, using the case of Tararua wind farm, identifies the variability levels of wind power transformed from wind speed; a Double Exponential distribution (DEPD) characterises the wind power change which is conditional on present power information. When examining the conditional probability in extracted smoothing subsets, exercises relating to probabilistic models suggest an advanced Mixture Beta model (PBMB) that combines the accuracy of the non-parametric model and the modelling efficiency of the parametric model. Considering the power boundaries as additional variables, an improved model enhances the accurate forecasting from two sides, the probabilistic and the time series. To validate the proposed models, simulations of historical wind power and prospective national planning are implemented. The stochastic behaviour of wind power in a small market penetration can be concluded from the summarized results depending on average wind capacity factor. However, a large scale of wind power leads to different consequences. The average generation and variation of the wind power become instable in the long time periods. Further practical evidences need to be tested to form a conclusion. The integration of simulated wind power into the New Zealand power system develops the investigation into wind power models to impact upon the systematic interaction, which is complex. The dispatch model (SWDM), in which wind power is one of input variables in association with information related to the national grid, optimises the electricity operations. The market reflection of energy flows, operational constraints and economic realisation is substantially influenced by wind uncertainty, especially during peak-load time periods. |
