摘要
相空间重构是通过一维的时间序列反向构造出原系统的相空间结构,它的基本思想就是系统任一分量的演化是由与之相互作用的其他分量所决定的,而这些相关分量的信息就隐含在任一分量的发展过程中。建筑业系统变化过程是一个非线性变化过程,具有非常复杂的非线性动力学特征,因此对建筑业增加值的时间序列进行空间重构分析可以对建筑业整个系统有更全面的认识。通过计算关联维和最大Lyapunov指数分析,证实建筑业增加值时间序列的混沌特性。对建筑业增加值时间序列应用递归图进行定性分析,并用RBF神经网络进行单步和多步的混沌预测。预测结果表明:建筑业产业增加值季度时间序列的单步预测值和实际值比较接近,而多步预测效果差,即有短期可预测性、长期内不可预测。
Phase space reconstructio n is to construct the original phase space structure reversely through one-dimensional time series. The basic idea is that the evolution of every component of the system is determined by other interacting components and the information of the relative components is hidden in their developing progresses. The progress of construction system is non-linear and complicated in non-linear dynamics characteristics. The space reconstruction analysis of time series for construction added-value can make the entire construction system known more comprehensively. The chaotic characteristic of time series for construction added-value is confirmed by calculating the correlative dimensions and analyzing the maximum Lyapunov exponent. Qualitative analysis of the time series for construction added-value is carried out using recurrence plot and chaotic prediction of one-step and multi-step are made by using the RBF neural network. The results show that the values of one-step prediction are close to the true values of the quarter time series for construction industry added-value and multi-step prediction performs well in that it can be predicted precisely only in the short term.
出处
《土木工程学报》
EI
CSCD
北大核心
2008年第8期99-104,共6页
China Civil Engineering Journal
关键词
建筑业增加值
相空间重构
混沌
关联维
LYAPUNOV指数
construction added-value
phase space reconstruction
chaotic
correlative dimension
lyapunov exponent
