Study On Voidage Measurement Of Oil-gas Two-phase Flow

Two-phase flow widely exists in industries such as power, chemical engineering, petroleum, metallurgy and nuclear energy. The parameters measurement of two-phase flow is very important for the metering, control and environment protection in production process. However, the complexity of two-phase flow results in many difficulties in the parameters measurement. Therefore, the parameters measurement of two-phase flow is always a frontier subject in research on two-phase flow. Voidage is one of the most important parameters of gas-liquid two-phase flow. Based on 12-electrode Electrical Capacitance Tomography (ECT), the voidage measurement of oil-gas two-phase flow is studied.The main works of the dissertation are listed as follows:1. Independent Component Analysis (ICA) and Support Vector Machine (SVM) are introduced to the voidage measurement of oil-gas two-phase flow. Features are extracted from the 66 measured capacitances obtained from ECT sensors by ICA. It not only reduces the data dimension and computation cost, but also improves the generalization performance of the model. Least Squares Support Vector Machine (LS-SVM) is a new kind of SVM with the advantage of simple computation. Two kinds of improved LS-SVMs—Weighted Least Squares Support Vector Machine and Sparse Least Squares Support Vector Machine are introduced, for the standard LS-SVM is bad in robustness and has no sparseness.2. Evolution Strategy (ES) is introduced to solve the difficult problem of parameters selection in LS-SVM, and a parameters selection method based on ES is proposed. The issue of parameters selection in LS-SVM is regarded as an optimization problem, objective function is established, and ES with global searching capability is employed to search the optimal parameters in LS-SVM. This method is universal for the parameters selection in SVM.3. Based on ECT, a new method is proposed for voidage measurement of oil-gas two-phase flow. In the measurement process, with the 66 measured capacitances obtained from ECT sensors being the inputs, the voidage is computed by using the voidage model. In the modeling stage, features are extracted from the 66 measured capacitances by ICA firstly, and then LS-SVM is used to establish the regression function, in which the features are the inputs and the corresponding voidage value is the output. Also, ES is employed to search the optimal parameters in LS-SVM. Static experimental results prove that the proposed method is effective.