Applications On Muti-sensor Information Fusion Technology In The Combination Station Oil-water Separation Process

Multi-sensor information fusion technology,as more than one of an interdisciplinary science,had been promoted and applied in the practical program,such as industrial process monitoring,industrial robot,intelligent manufacturing system, medical image processing etc.Study of multi-sensor information fusion technology is the focus in domestic and abroad recently, Mainly because changes of the parasitic capacitance and the ambient temperature and salinity, useing capacitor method measurement of crude oil exports moisture cannot achieve high-precision. Multi-sensor information fusion technology algorithm solve practical problems in order to improve the measurement accuracy and to eliminate the impact of other interference and to improve on the target parameter identification capability.In this paper,we had learned the knowledge of multi-sensor information fusion technology algorithm.Since the features of multi-sensor information fusion technology,we concerned to combine it with artificial neural network and genetic algorithm.The contributions of this paper are as follows:First,according to the learning ability and powerful mapping ability for nonlinear system of neural network,BP neural network combined with multi-sensor information fusion technology,which measured the water content ratio of the combination station oil-water separation process.Simulation results demonstrate the multi-sensor information fusion technology based on BP neural network can approximate measurement accuracy.Second,It is the fact BP neural network is easily fallen into local minimums and learning time long.So a method based on L-M optimization combined with multi-sensor information fusion technology is presented.L-M optimization train BP network weights and thresholds,and the network convergence speed is greatly improved.At the same time the multi-sensor information fusion technology based on RBF neural network model presented. Simulation results demonstrate these two methods both can approximate measurement accuracy.Third, GAL-M combines the advantages of neural network and genetic algorithm.Given the multi-sensor information fusion technology based on GAL-M model.The model can not only solve the network initial weights and thresholds, but also effectively improve system reliability.Model designed through the program proposed in this paper can make system immunity.