ppt课件-penetration level impactof wind powered generationto power（渗透水平impactof风动力generationto权力）.pptVIP

* * * * When applying to wind generation forecast, the neural network will be used to construct the optimal transformation between the affecting variable such as wind speed to the output which is the wind power generation. The manufacture generally provide the turbine power curve to represent the characteristic of each wind turbine. The simplest method to project the output power is to use this power curve. This is called a direct transformation. The validity of the approach subjects to many constraint, which is likely not be be valid in real operation of wind farm. For example, wind speed is not necessary measured at the same height which the power curve was generated. So it’s difficult to get a good result from applying he curve. Also if the information for each individual wind turbine is not available, then power curve cannot be applied. What ANN does in this case is to develop the optimum transfer function based on the presentation of recorded historical data. For each level of wind speed, the wind power generation from a wind farm may varied in a certain range. After training, ANN will develop the curve that best match this information which later can be used to project wind power forecast in the future. * * * * * * * * * * * * * * * * * * Non Coincident Peak Wind Generation in Texas Source: ERCOT Unit Commitment Scheduling Market clearing price for energy and total wind generation on February 22, 2005 Price Dips Market clearing price for energy (MCPE) and total wind generation on April 27, 2007. Variability Increases Operating Costs Committing unneeded generation Scheduling unneeded generation Allocating extra load-following capability Violation of system performance criteria (For example, spinning reserve) Requirement of reactive power supply for induction-type generator These will increase the Ancillary Service Costs Voltage Fluctuation Buses % Real Power deliver by wind Generator 100% 75% 50% 25% IG operate at 0.85 leading PF, Fix Qc=70.5 MVAR (Unity P