两级数据融合算法在木材干燥过程中的应用

本文提出了一种基于RBF神经网络和证据理论的两级数据融合方法。利用RBF神经网络实现特征层数据融合,建立基本信任分配函数,具有最佳一致逼近特性,同时解决了D-S证据理论确定基本信任分配函数困难的问题。基于D-S证据理论的传感器故障诊断方法的研究,可有效地判断工业现场传感器的工作状态。实验结果表明该方法可正确定位并准确分离出木材含水率检测系统中失效传感器。

Application of Two-Level Data Fusion Algorithm in Wood Drying Industry Process

Two-level fusion method based on RBF neural network and D-S theory is presented in the paper. Feature-layer data fusion is realized by RBF neural network. RBFNN is adopted to establish basic belief assignment function, achieve the optimum uniform approximation characteristic, and solve the problem that basic belief assignment function is difficult to create in D-S evidence theory. The operating condition of sensors is diagnosed in industry process effectively. The method is used in the timber moisture measuring system. The simulation results show that the method can locate the fault sensors＇ position and isolate the fault sensors accurately.