基于投影寻踪模拟退火的多传感信息融合技术  

Multi-Sensor Information Fusion Technology Based on Projection Pursuit Simulated Annealing

作　　者：张友好 郑艳芳 

机构地区：[1]江苏科技大学张家港校区电气与信息工程学院,江苏 苏州

出　　处：《传感器技术与应用》2020年第2期11-22,共12页Journal of Sensor Technology and Application

基　　金：国家自然科学基金(11703009) (针对平移误差检测的电容式边缘传感器改进研究)。

摘　　要：本文吸取加权平均法、概率统计法等算法的优点,设计了基于投影寻踪模拟退火法的多传感器信息融合算法。设计创新点在于对投影权值调整过程中加入概率窗口约束及聚类约束,使系统的计算复杂度减少41.6%,得出的结果符合峰值回归判断。本设计对于解决系统的多传感器信息融合有了新的改进,对于实时性和适用范围均有了一定的提升。Based on the advantages of weighted average method and probability statistics method, a multi-sensor information fusion algorithm based on projection pursuit simulated annealing method is designed in this paper. The innovation of the design lies in the addition of probability window constraint and clustering constraint in the process of projection weight adjustment to reduce the computational complexity of the system by 41.6%, and the results obtained are in line with the peak regression judgment. This design has a new improvement to solve the multi-sensor information fusion of the system, and has a certain improvement to the real-time and applicable range.

关 键 词：多传感器信息融合 投影寻踪 模拟退火法 K-MEANS 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]