论文部分内容阅读
给出了建立门限自回归模型（ＴＡＲ） 的一套实用方法， 用作者提出的改进遗传算法可同时优化门限值和自回归系数，成功地解决了ＴＡＲ建模过程所涉及的大量复杂寻优工作这一难题。实例计算的结果说明了， 这套方法在海洋冰情预测中是可行的和有效的， 在各种自然灾害非线性时序动态预测中具有重要的理论意义和广泛的实用价值。
A set of practical methods for establishing threshold auto-regressive model (TAR) is given. The proposed improved genetic algorithm can simultaneously optimize threshold and auto-regression coefficients and successfully solve a large number of complex optimization problems involved in the TAR modeling process Work this problem. The result of example calculation shows that this method is feasible and effective in marine ice condition prediction and has important theoretical significance and extensive practical value in predicting nonlinear time series dynamics of various natural disasters.