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

The oil-water two phase flow is a complex flow process, which widely exists inpetroleum, chemical and other industrial processes. Along with the improvement ofmeasuring requires, energy saving and control in the nature and industrial process,new challenges are proposed for the measurement precision of the oil-watertwo-phase flow. New ideas and methods are being developed constantly.In the flow process of multiphase flow, the distribution of each phase mediumis random because of the different mixed flow medium with diverse conditions, whichleads to complicated flow states and varied parameters. As a result, a single sensorcannot obtain comprehensive and accurate estimations on multiphase flow parameters.However, data fusion method with heterogeneous sensors can obtain morecomprehensive understanding about the process at a relatively low cost and widerange of application, it therefore has prospective applications in multiphase flowmeasurement.In order to investigate the accurate detection problem of the oil-water two phaseflow of individual flow rate estimation, multi-sensor data fusion method is used, theinformation obtained from different sensors of oil-water two-phase flow are fused, toexplore a more accurate flow detection method. The research work includes:(1) Based on the summary of the existing multiphase flow measurement methodand the multi-sensor fusion technology, parameter estimation of multiphase flowusing the kalman estimation theory is put forward to solve the accurate detectionproblem of the multiphase flow parameters in the industrial application,.(2) A conductance sensor array and a cone type differential pressure sensor areused to measure the flow process. In order to obtain the total flow estimation of oiland water, the kalman centralized and distributed fusion estimation methods areapplied to fuse the data obtained by the two sensors. The experimental results showthat the detection accuracy of two sensor fusion estimation is superior to the singlesensor test results; and the kalman estimation precision of distributed fusion method isbetter than kalman centralized fusion method.(3) The discrete kalman estimation theory is applied to fuse the two phase flowinformation of conductance sensor array and cone type differential pressure sensor,and the oil phase flow rate can be obtained. The experimental results show that theestimation accuracy of oil phase flow is improved under the steady-state conditions.(4) In view of the actual oil/water two-phase flow measurement, the flow statechanges, the fusion estimator does not track sensor signal change fast, which can beimproved by introducing the switching function and switching system noisecovariance matrix. This improves the tracking ability of the estimator in thecontinuous variable instantaneous working condition and meets the needs of practicalapplications.