On Supervisory Systems For Gas Liqud Cylindrical Cyclone Multiphase Flow Meters

GLCC (Gas Liquid Cylindrical Cyclone) multiphase flow meter is an efficient separation measurement device. Micro GLCC multiphase flow meters have been widely used. The performance of GLCC depends on the gas-liquid separation effect. The conventional method is PID single loop control. Due to the mutual effect of gas and liquid phase, the control effect is poor. Oil production prediction is the basement of optimization decisions and dynamic analysis in oil exploitation. The existing GLCC multiphase flow meter systems can only statistic the flow of air, oil and water but can’t predict the production. Therefore, the study on supervisory systems for GLCC multiphase flow meters with the main content of separation optimum control and oil production prediction has great significance.The GLCC multiphase flow metering system is multiple variables, strong coupling and large delay. The thesis mainly study on model predictive control and support vector machine method. Based on the theories described above, design the supervisory systems for GLCC multiphase flow meters to control the liquid level of GLCC as well as predict the oil production. The main contributions of the thesis are as follows:1. To use the model predictive control as the main control method of GLCC multiphase flow meters, firstly we must ensure the stability of the algorithm. Since the systems are multiple variables, we presents a method of stability analysis of incremental predictive control scheme for linear systems with open-loop stability. By using the Lipschitz condition, weak controllability and strong duality of steady-state problem assumptions, quadratic performance indexes are transformed into new performance indexes. This leads to the Lyapunov functions of the closed-loop system. Then results on closed-loop stability of the incremental predictive control scheme are derived. An example of GLCC control simulation is used to demonstrate the effectiveness of the method proposed here.2. A brief introduction to the theoretical basis of Support Vector Machine and its principle. In the situation of sample number limited, use the SVM algorithm to predict the oil production. The simulation results show that the algorithm can be used to achieve the desired effect.3. Based on the theories described above, design the supervisory systems for GLCC multiphase flow meters. Use model predictive control as the main method, combined with the support vector machine to control the liquid of GLCC as well as predict the oil production. The running results show that the system has a good performance at controlling and monitoring the system.A number of illustrative simulations are given to show the effectiveness of proposed algorithms. And the reliability of the system was verified by the experimental. Finally, the conclusion and future work are presented.