基于SVM的入库径流混沌时间序列预测模型及应用

Chaotic Time Series Forecasting Model Based on SVM for Reservoir Runoff

针对常用的入库径流混沌预测模型只能做短期预测,且需要大量样本数据的问题,将支持向量机理论与混沌预测理论相耦合,建立基于支持向量机的入库径流混沌时间序列预测模型。该模型利用混沌理论中的相空间重构技术将原始入库径流序列映射到一个高维相空间,以相空间中的相点为基础构造训练样本和测试样本,然后利用支持向量机理论进行预测。经实例计算,模型比基于最大Lyapunov指数的混沌预测模型、人工神经网络模型和自回归模型拟合效果好,预测精度高,丰富和发展了入库径流预测理论和方法。

Aiming at the problem that traditional chaotic forecasting model is only for short-term forecasting and needs a lot of data, a chaotic time series forecasting model based on support vector machine for reservoir runoff was established which combines the support vector machine theory and chaotic forecasting theory. The original time series was reconstructed to a high dimension space through the skills of state space reconstruction. The training sample and testing sample was conformed based on the states in the state space, and then the support vector machine theory was used for forecasting. By example study, the supposed model has better curve fitting and higher forecasting precision compared to the chaotic forecasting model based on large Lyapunov index, the artificial neural network model and autoregressive model. It develops the theory and method for reservoir runoff forecasting.