摘要
在复杂环境下进行多传感器测试,其数据分布往往不规则和不一致。针对该情况,提出一种基于聚类的多传感器数据融合方法。该方法不按权重相加,侧重于分析数据整体分布状况。采用模糊梯形函数对数据进行一致度量化,使用聚类算法对数据分布进行聚类分析,按照最大支持度原则寻找最优点。实验结果表明,该方法能得到较精确的融合值,并可以查找在测试过程中可能出现的故障。
Multi-sensors are often used when testing in a complex environment, however the results sometimes are difficult to interpret especially when data distribution is irregular and even inconsistent. This paper presents a new analysis and fusion method which focuses on the distribution analysis for all the data rather than just making the weights for each single sensor in the traditional ways. It uses a fuzzy gradient function to quantify the consistent degree of the sensor data, and makes an algorithm for cluster analysis. The method integrates data by using the supportive degree. Experimental results show that the method can make a better integration and also can help to find faults.
出处
《计算机工程》
CAS
CSCD
2013年第5期61-64,68,共5页
Computer Engineering
基金
航空科学基金资助项目(20101024006)
关键词
数据融合
多传感器
一致度
聚类
支持度
data fusion
multi-sensor
consistency
clustering
support degree
