基于神经网络的多传感器数据融合  

Multi-Sensor Data Fusion Based on BP Neuron Network

作　　者：胡仲毅[1] 陈进军[1] 鲁绪文[2] 

机构地区：[1]贵州大学电气工程学院,贵州贵阳550003 [2]贵州大学资环学院,贵州贵阳550003

出　　处：《贵州工业大学学报（自然科学版）》2005年第4期50-53,共4页Journal of Guizhou University of Technology(Natural Science Edition)

摘　　要：当前,多传感器技术是研究热点。通过不同功能,不同精度,不同位置的传感器,可以对所需要的被测量进行多方位多角度的测量。但是,目前对于多传感器所测的数据还没有一种通用的行之有效的处理思想。借用模糊控制策略,与改进的BP神经网络算法结合对其进行数据融合。在仿真中取得了比已往算法更高的精度,控制策略制定准确、可靠。是一种较有潜力的多传感器数据融合方法。At present, the multi - sensor technique is a study focus. From different positions, through different functions, the different accuracy sensors can be used to carry out measurements to the target. But currently, there is not a very useful principle that can handle data of multi - sensor system universally. This paper uses the fuzzy calculation approach, combined with the BP neuron network whose calculating-algorithm has been improved into data fusion, identifying, predicting, controlling, have got more accurate results in the emulating than ever before,and control- strategy obtained is precise, credible. It is a kind of potential data fusion methods.

关 键 词：多传感器 数据融合 BP神经网络 Lenenberg-Marquardt算法 模糊控制 滤噪 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]