基于多模型SVR的抗野值Kalman滤波器

• 作 者: 袁伟
• 机 构: 江苏科技大学电子信息学院,
• 出 处: 科技信息(学术研究)
• 发表日期: 2008年34期

为了抑制观测数据中野值对Kalman滤波的不利影响,本文首先基于支持向量机(SVM)和多模型理论,建立了Kalman滤波新息过程的多模型支持向量回归(SVR)预测模型,然后给出了结合该预测模型的Kalman滤波的修正算法。仿真结果表明,该Kalman滤波算法不仅有效地剔除了观测数据的野值,提高了滤波器的稳定性和精度,而且算法比较简单。

In order to restrain the unfavorable influence of outliers on Kalman filtering, this paper first builds a multi-model support vector regression (SVR) prediction model based on support vector machine (SVM) and multi-model theory, Kalman Filter Correction Algorithm Based on the Prediction Model. The simulation results show that the Kalman filtering algorithm not only effectively eliminates the outliers of the observed data and improves the stability and accuracy of the filter, but also has a simple algorithm.