短期风速时间序列混沌特性分析及预测

针对短期风速时间序列的预测问题进行了研究. 首先通过0-1混沌测试法确定短期风速时间序列具有混沌特性. 采用相空间重构技术, 利用C-C算法确定延迟时间, G-P 算法确定嵌入维数. 然后提出一种参数在线修正的最小二乘支持向量机预测模型, 采用改进的粒子群算法进行预测模型中参数的优化. 最后通过仿真对比实验表明提出的预测方法在预测精度、预测误差、预测效果方面都要优于其他常见的预测方法, 证明该预测方法是有效的.

Chaotic characteristics analysis and prediction for short-term wind sp eed time series

A short-term wind speed time series prediction is studied. First, 0-1 test method for chaos is used to identify the short-term wind speed time series that has chaotic characteristics. Through phase space reconstruction, the delay time is determined by using C-C algorithm; and the embedding dimension is determined by using G-P algorithm. Then a least square support vector machine with parameters online modified is proposed, so that an improved particle swarm optimization algorithm may be used for the prediction of parameters optimization. Simulation experiment shows that the present method for its prediction accuracy, prediction error, and prediction effect is better than other prediction methods. Thus the proposed prediction method is effective, and feasible.