Application of Artificial Neural Networks in Oil and Gas Multiphase Metering

摘要

In order to study the law of multiphase flow in pipeline and solve the on-line multiphase metering problem without separation of gas and liquid, a new type of multiphase flowmeter was developed and a series of water-gas two phase flows experiments in horizontal pipeline were carried out. And Artificial Neural Networks was used to process data after the experiments. The results show that Artificial Neural Networks could be used to simulate the relationship of the variables that were affected by many uncertain factors very well. And the relative error of liquid phase is less than 10% as well as the relative error of gaseous phase is less than 20%.