Research On Short-Term Load Algorithm For Chaotic Time Series

Load prediction is the core issue of planning, operation of power system. In recent years, there appeared a variety of effects of accuracy, reliability, such as weather, policy, economy. The truth shows that traditional load prediction methods no longer totally meet the adaptability and accuracy requirements of high load prediction. At the same time, the rise of new energy for accessing to grid brings the problem of the source of load where it comes from..Based on the background above, this paper aimed at designing load prediction method with better adaptability and accuracy. We clarified the definition of chaotic time series and analyzed single model load prediction methods and multi-model load prediction methods separately to develop adaptive load prediction algorithm. Furthermore, we presente interval prediction algorithm based on add-weigthted one rank local region predition. The specific research contents are as follows:1) Clarified the definition of chaos and summarizes three measures of chaotic time series. Enumerated actual maneuvering trajectories of chaotic time series.Introduced global method and local region methods, and pointed out that traditional local region method loses sensibility for prediction when it reaches interval predition, and introduced one rank local region multi-steps method which can fix this problem. Then presented performance evaluation indicators of one rank local region multi-steps method.2) Studied interval prediction based on add-weighted one rank local region method. Analyzed traditional method such as chaotic time series, and pointed out the limit of fixed-parameter model in changing environment adaption. Based on existing interval prediction, we proposed one rank local region method and add-weighted one rank local region multi-steps method based on interval prediton.The chaotic time series model was presented to better describe the flexibility at last. Simulation was implemented to verify the performance.3) Studied the prediction based on chaotic time series algorithm. Multi-steps method is a better choice in load prediction while singer step method cannot fit the variety of motion. Based on one rank local region model, we introduced add-weighted one rank local region mode algorithm. Then we proposed the add-weighted one rank local region method based on new energy to join in. And then simulation was implemented to verify the effect.4) Studied the intersection with interval prediction based on add-weighted one rank local region multi-steps method.Proposed interval prediction based on one rank local region multi-steps method, and discussed the reasons why the methods could bring in intersection. Proposed the mathematical proof, and simulation was implemented. At last, we presented the process of interval prediction based on add-weighted one rank local region multi-steps method with single model or multiple models.