【摘要】 在综合考虑经典投影寻踪算法特点的基础上,针对投影寻踪计算中存在的一些不利因素,给出相应的解决思路。利用数值仿真技术进行基于粒子群优化算法与厄密特多项式构建的投影寻踪回归模型建模能力与计算精度的检验,再将其应用于多维地震时间序列综合建模预测中。计算结果和进一步分析表明,基于粒子群优化算法与厄密特多项式构建的投影寻踪回归投影寻踪模型具有简单、快速、有效的特点,在实际地震综合预测建模中取得了满意的效果,可作为地震综合预测的一种回归分析方法。更多还原

【Abstract】 Considering its characteristics,we give the solution to the flaws of the conventional Project Pursuit Regression(PPR)algorithm.We use the numerical emulation technique to test the modeling ability and calculating accuracy of PPR that based on particle swarm optimization and Hermite multinomial,and apply HPPPR to the synthetic,seismic forecasting modeling of multi-dimensional seismic time series.The result shows that PPR model that based on particle swarm optimization and Hermite multinomial is simple,fast and effective,and it is satisfying in the real seismic synthesis forecasting modeling and it can serve as a regressing method.更多还原