Research On Ultra Short-term Wind Power Forecasting And Its Ensuing Integration Transient Tripping Control Issues

With the rapid increase of wind power integration, the intermittent nature and fluctuation of wind energy resource has become a technique bottleneck for large-scale wind power penetration. A reliable and accurate ultra-short and/or short term wind-power prediction will help to reduce system reserve and overall operation cost, even improve power grid security and stability. Moreover, increasing penetration of wind farms will make an impact on the stability of power system transient process. It has resulted in the need for the wind tripping measures in the power system emergency control. This thesis mainly obtains the following results:1. An overview of wind power prediction methods and its research around the world were outlined. Then, combined with actual historical information in a typical wind farm and its numerical weather prediction, the SCADA data was corrected and complemented. Based on that, the influential factors to the output were analyzed and the fuzzy C-means clustering algorithm was adopted to reflect the seasonal characteristic of the wind.2. The transient stability of grid-connected large-scale DFIG is outlined. The different type of the wind grid-connection transmission channel and the DFIG simulation model in PSASP are introduced. Then, the parameters of the case study in north-eastern and its difficulty in wind power utilization are illustrated.3. An ultra short-term wind power prediction method based on online sequential extreme learning machine (OS-ELM) was proposed in this paper. Combined with the batch processing and successive iteration, the real-time prediction of wind turbine power output and the NWP wind speed correction were accomplished by taking advantage of OS-ELM’s fast learning speed and strong generalization ability.4. The multi-ELM error evaluation network based on Bootstrap method was established to generate Bootstrap samples. Then the bootstrap wind power prediction intervals were constructed via Biased-corrected Percentile Bootstrap (BCPB) method.5. The doubly-fed induction generator (DFIG) transient operating characteristics were analyzed. The transient stability analysis of the case study and the combination of the wind power and thermal power tripping control under serious fault conditions were simulated via PSASP. The results showed that there existed an optimal proportion tripping decision-making.