Multi Sensor Fusion Estimation Of Oil-water Two-phase Flow

Oil-water two-phase flow widely exists in industrial production process with a complex flow state.The characterization of flow state and identification of flow regime for oil-water two-phase flow is an important basis for accurate estimation of the process parameters.With the improvement of the requirement of measurement and environmental protection in industrial production,the requirement for accurate measurement of oil-water two phase flow parameters is improved.Although there are many kinds of detection methods for oil-water two-phase flow,the measurement method with single sensor information can not solve the parameter estimation problem of oil-water two-phase flow.Therefore,the multi sensor fusion technology is introduced into the parameter estimation of oil-water two-phase flow.Multi sensor fusion technology based on different time and space of multi-sensor data resources,can achieve more complete,accurate and reliable measurement and description than the single sensor information.In order to accurately describe the flow characteristics of the process and characterization of oil-water two-phase flow,and achieve the accurate identification of flow pattern and further achieve the estimation of the phase holdup of oil-water two-phase flow,Multi sensor information fusion technology is adopted.The measurement data of electrical resistance tomography system and ultrasonic transmission system are fused.The flow regime identification model is established by using the radial basis function neural network and the phase holdup estimation model is established by using the nonlinear partial least square method.The specific research work of the project includes:(1)The characteristics and mechanism of two phase flow are studied and the electrical resistance tomography and ultrasonic transmission technology are summarized.In view of the high data dimension and the large amount of data in the electrical resistance tomography system and the ultrasonic transmission system,the law of feature extraction is proposed for the analysis of the characteristics of oil-water two-phase flow.(2)A flow pattern recognition model based on radial basis function neural network is designed on the basis of feature extraction.The characteristics of typical oil-water two-phase flow pattern are analyzed,and the accurate identification of oil-water two-phase flow pattern is realized.(3)Linear method is not suitable for the complex nonlinear model of oil-water two-phase flow.The partial least square method is adopted to design the flow regime identification model of oil-water two-phase flow.In view of the single model can not be applied to a variety of typical flow patterns,a multi model structure is proposed to estimate the phase holdup of oil-water two-phase flow.(4)The flow regime identification model and phase holdup estimation model of oil-water two-phase flow are designed based on multi sensor data fusion.This method has higher accuracy than the single sensor model.