变分贝叶斯Kriging模型预测混沌时间序列

基于变分贝叶斯及Kriging数学思想,提出了一种含噪混沌时间序列的相空间域预测模型。在相空间域中利用变分贝叶斯推断方法估计模型中的回归系数,采用Kriging数学方法估计模型中的随机部分,将该模型对含加性高斯噪声的Lorenz及Mackey-Glass混沌时间序列进行了预测研究;结果表明该文方法能够有效地预测含噪混沌时间序列,且具有较强的抗噪能力以及有效地克服了过拟和现象;同时预测精度对重构相空间的嵌入维数和时间延迟的变化不敏感。

Chaotic time series prediction based on the variational Bayesian kriging model

The paper presents a prediction model of the phase space domain of noisy chaotic time series based on the variational Bayesian method and the kriging mathematical idea.In the phase space domain the variational Bayesian method is adopted to estimate the regression coefficients and the kriging mathematical method is used to estimate random values in the model.The model is used to predict Lorenz and Mackey-Glass chaotic time series with additive gaussian noise.The results show the presented method can predict noisy chaotic time series effectively.The model is robust to noise and overcomes the overfitting effectively.The prediction effect is not sensitive to the change of embedding dimensions and time delay.