Study on mass-flow measure data fusion based on neural network

In order to solve the problem of mass-flow sensor measure precision influenced by the environment, the mass-flow sensor measure data fusion problem under different pressures is considered. The multi mass-flow sensors are used to obtain multi-channel flow data. The self-adaptive weight fusion method is firstly adopted to obtain the first-stage fusion values whose precision can be largely improved. To further restrain the mass-flow measure precision affected by the pressure, the second-stage fusion is carried out by BP neural network and RBF neural network. The data fusion results show that the fusion values precision by neural networks is significantly improved, and the fusion values by RBF neural network are with higher precision and minimal errors.