基于相空间重构与最小二乘支持向量机的时延预测

发布时间：2018-03-12 10:41

本文选题：网络控制系统　切入点：相空间重构　出处：《电子学报》2017年05期 　论文类型：期刊论文

【摘要】：针对网络控制系统的时延预测问题,提出一种基于相空间重构与最小二乘支持向量机的时延预测方法.首先利用0-1测试法确定时延序列具有混沌特性,引入相空间重构技术提高预测精度.对实际采集的时延序列进行Hurst指数分析,选择最小二乘支持向量机作为预测模型.然后利用C-C方法确定时延序列相空间重构参数,通过递归图确定时延序列的局部可预测性,利用遗传算法对最小二乘支持向量机的参数进行离线优化.最后通过优化后的最小二乘支持向量机并结合相空间重构对时延序列进行在线预测.与其它预测方法进行了仿真对比,结果表明本文方法具有更高的预测精度与更小的预测误差,同时并未降低预测算法的实时性.
[Abstract]:For the networked control system with time delay prediction problems, put forward a prediction method of phase space reconstruction and least squares support vector machine based on the time delay. First determine the delay time series has chaotic characteristics by using the 0-1 test method, the phase space reconstruction technique to improve the prediction accuracy. The Hurst index analysis of the actual delay sequence collection, selection of the least squares support vector machine as the prediction model. Then use the C-C method to determine the parameters of phase space reconstruction delay sequence through recursive delay sequence diagram to determine local predictability, off-line optimization of the parameters of least squares support vector machine using genetic algorithm. Finally, the optimized least square support vector machine combined with phase space reconstruction is applied to the online prediction of the delay time series. Compared with other forecast methods, the results show that this method has higher prediction accuracy and smaller The prediction error does not reduce the real-time performance of the prediction algorithm.

【作者单位】：沈阳工业大学信息科学与工程学院;东北大学计算机科学与工程学院;
【基金】：国家自然科学基金(No.11273001) 辽宁省博士启动基金(No.20141070)
【分类号】：TP18;TP273

【相似文献】