Study On New Measurement Methods Of Gas-liquid Two-phase Flow In Small Channels Based On Sensor Data Fusion

With the rapid development of micro-fabrication technology and new material technology, the miniaturized tendency of the industrial equipment becomes increasingly obvious. The study on gas-liquid two-phase flow parameter measurement in small channels has become a popular field in current two-phase flow research. This dissertation mainly aims at the online measurement of the flow pattern, void fraction of gas-liquid two-phase flow in small channels. Based on an optical position sensitive detector and a capacitively coupled contactless conductivity detection(C4D) sensor, and by using the information processing technology, the sensor data fusion technology, the research work on these parameters measurement are carried out.The main work and innovations are listed as follows:1. A new optical position sensitive detector and a new C4D sensor are developed, and an online parameter measurement system is built for gas-liquid two-phase flow in small channel. Research results show that the developed optical position sensitive detector and the developed C D sensor are successful, and they are both suitable for the parameter measurement of gas-liquid two-phase flow in small channel.2. Based on the optical position sensitive detector, flow pattern identification and void fraction measurement are studied for gas-liquid two-phase flow in small channel. First, using the obtained optical signal, three feature extraction methods are compared:1) using statistical analysis method for feature extraction;2) using the combination of statistical analysis and wavelet decomposition method for feature extraction;3)using the combination of statistical analysis and Empirical Mode Decomposition (EMD) method for feature extraction. Then, the effectiveness of these three feature extraction methods are verified by flow pattern identification experiment. Research results show that, these three feature extraction methods are all effective for flow pattern identification. The method which uses statistical analysis for feature extraction has good real-time performance, and the identification results using this method are satisfactory. In three pipes with the inner diameters of4.0mm,3.0mm and1.8mm, the identification accuracies for typical flow patterns are higher than85.0%,75.0%and83.0%, respectively. Meanwhile, based on the obtained optical signals, two void fraction measurement models are developed for slug flow based on physical model and LS-SVM regression method, respectively, and the effectiveness of the void fraction measurement models are testified by dynamic experiments. Experiment results show that these two developed void fraction measurement models are both effective. The measurement model based on LS-SVM can obtain higher accuracy of void fraction measurement (the maximum error are is than5.0%based on LS-SVM regression model, while the maximum errors is less than10.0%based on physical model).3. Based on C4D sensor, flow pattern identification and void fraction measurement are studied for gas-liquid two-phase flow in small channel. First, using the C4D sensor’s signals, flow pattern identification are carried out in different pipes based on the statistical analysis method. Then, using the C4D sensor’s signals, void fraction measurement models for typical flow patterns are developed, and the effectiveness of the models are verified by dynamic experiments. Research results show that the developed C4D sensor is effective for flow pattern identification and void fraction measurement of gas-liquid two-phase flow in small channel. In the pipes with the inner diameters of4.0mm,3.0mm and1.8mm, the identification accuracies for typical flow patterns are higher than78.0%,88.0%and83.0%, respectively. Besides, in the pipe with the inner diameter of3.0mm, the maximum errors of void fraction measurement under typical flow patterns (slug flow, bubble flow, stratified flow and annular flow) are all less than10.0%,3.5%,3.0%and5.0%, respectively.4. Based on sensor data fusion technique, and combing with the optical position detector and C4D sensor, a new flow pattern identification method and a new void fraction measurement method are proposed for gas-liquid two-phase flow in small channel. The proposed flow pattern identification method using the signals from the optical position detector and C4D sensor to identify the flow pattern respectively, and then the identification results obtained by two sensors are fused based on D-S theory, finally the flow pattern are determined. The proposed void fraction measurement method determines the current flow pattern firstly, and then selecting the proper void fraction measurement model according to the flow pattern identification results (if the flow pattern is slug flow, the LS-SVM model based on optical position detector is selected, else, the LS-SVM model based on C4D sensor is selected), finnally the void fraction is predicted by the selected model. Research results show that the proposed flow pattern identification method and the void fraction measurement method are both effective. Based on sensor data fusion technique, the performance of the flow pattern identification and the void fraction measurement are improved. In the pipes with the inner diameters of4.0mm,3.0mm and1.8mm, the identification accuracies for typical flow patterns are higher than85.0%,97.0%and88.0%, respectively. Besides, in the pipe with the inner diameter of3.0mm, the maximum errors of void fraction measurement under typical flow patterns (slug flow, bubble flow, stratified flow and annular flow) are all less than5.0%,3.5%,3.0%and5.0%, respectively. As the proposed flow pattern identification method and the void fraction method can sufficiently using the complementarity of the two sensors, the performance of the flow pattern identification and the void fraction measurement are improved.