压力信号干扰抑制的质量流量数据融合研究  

Study on Mass Flow Data Fusion for Restraining Pressure Disturbance

作　　者：汪洪波[1] 唐志国[1] 马培勇[1] 林胜[1] 

机构地区：[1]合肥工业大学,合肥230009

出　　处：《中国机械工程》2012年第18期2223-2227,共5页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(51006031);安徽省自然科学青年科学基金资助项目(11040606Q40);中央高校基本科研业务费专项资金资助项目(2010HGBZ0600)

摘　　要：质量流量测量精度受压力的影响,且随着压力的增大其测量精度变差。采用多个质量流量传感器进行多处测量,对质量流量测量数据进行自适应加权融合。在此基础上,为了消除压力对流量测量值的影响,采用BP神经网络进行压力干扰抑制的质量流量数据融合研究。研究结果表明,BP神经网络质量流量融合值的精度较自适应加权融合值的精度大大提高,且附加动量法获得的BP网络融合精度最高,自适应学习速率调整法次之,梯度下降法最差。The mass flow measurement precision was affected by the pressure,and the precision was becoming worse with the pressure grew.Multi mass flow sensors were applied to obtain the measurement data which were fused by an adaptive weighted data fusion method.Besides,the back propagation（BP）neural network was utilized for carrying out the mass flow data fusion for restraining the pressure disturbance,so as to eliminate the sensitiveness of the mass flow measurement to pressure disturbance.The research results demonstrate that the mass flow measurement precision can be improved largely after the BP neural network data fusion than after the adaptive weighted fusion,and the BP neural network is with the highest fusion precision by additional momentum algorithm,worse by adaptive learning speed regulating algorithm and the worst one by gradient descending algorithm.

关 键 词：质量流量 压力 数据融合 神经网络 

分 类 号：TH814[机械工程—仪器科学与技术]