基于粒子群支持向量机的海杂波序列回归预测

Sea Clutter Sequences Regression Prediction Based on PSO-SVM Method[JZ)]

摘要:

在雷达数据处理中,为更好地抑制海杂波,预测海杂波是必要的；海杂波具有混沌特性,而支持向量机算法能够有效地对混沌序列进行回归预测,文章提出了一种改进的支持向量机海杂波序列回归预测算法；文中给出了算法的框架结构,采用了互信息法和改进的伪邻近点法提取海杂波混沌特性的延迟时间和嵌入维数,利用相空间重构求取SVM训练样本,应用改进的PSO算法优化SVM的核函数参数以及惩罚系数,并仿真了预测模型；仿真实验结果表明:海杂波回归预测能达到满意的精度,而PSO-SVM方法比SVM方法的预测精度更高。

Abstract:

In the radar data processing, in order to better suppress sea clutter, sea clutter prediction is necessary. Sea clutter has chaotic features, and support vector machine algorithm can effectively carry out regression prediction of chaotic sequences, this paper proposes an improved support vector machine regression sea clutter sequence prediction algorithm. The paper gives the algorithm framework, using a mutual information method and improved pseudo-point method to extract neighboring sea clutter chaotic time delay and embedding dimension, using phase space reconstruction to strike SVM training sample, applying improved PSO algorithm to optimize kernel function parameter and penalty coefficient of SVM and simulate a prediction model. Simulation results show that:regression model to predict sea clutter is feasible and PSO-SVM method can further improve the prediction of sea clutter prediction accuracy.