摘要
为了解决输油管道多传感器在线检测数据存在不确定性和模糊性问题,提出了输油管道在线检测数据融合模型。从参与融合的数据源头和智能融合器两方面,利用检测数据间互补信息,分两级逐步地降低检测数据误差。第一级融合,从检测数据自身蕴藏的信息提取出综合模糊支持度,由此确定各传感器自身数据的优劣程度,筛选组成最优融合数据源。第二级融合采用正交基神经网络,通过训练,获取检测数据与融合输出结果间的非线性映射关系,实现不同时段、不同组检测数据间的融合。仿真验证表明,建立的模型不但提高了检测数据融合结果准确性,而且数据融合实时处理能力强,并且硬件易实现。
The data fusion model for on -line inspection in oil tubing was proposed, in order to solve the uncertainty and ambiguity in the online testing data using multi - sensors. This model utilizes complementary, information between the testing data, in two ways of the data source involved in fusion and the intelligent fusion tool, so that it can gradually reduce the testing data error by two phases. In first - phase fusion, the comprehensive fuzzy support degree, corresponding to specific sense data, was gained from their information. Thereby the importance of each sense data was determined, and the optimal data sources are filtered out. The orthogonal basis neural network waas used to implement second phase fusion. The non - linear mapping relations between the test data and the fusing out- put were obtained through training, for achieving testing data fusion in different time periods and groups. The simulative verification shows that this model not only improves the accuracy of data fusion results and the capability of real - time process, but also simplifies the implementation of hardware in the process of data fusion.
出处
《计算机仿真》
CSCD
北大核心
2013年第2期322-326,共5页
Computer Simulation
关键词
输油管道
传感器
数据融合
模糊支持度
正交基神经网络
Oil tubing
Sensor, Data fusion
Comprehensive fuzzy support degree
Orthogonal basis neural network