基于人工智能技术的原油含水率动力学模型研究

基于介电常数法测量原油含水率，往往受油水两相流态、温度、矿化度、非线性特性等多因素的影响。本文设计了一套基于多传感器的油水两相流实验室模拟系统对多个参量进行测定，提出基于智能信息处理方法的粗糙集预处理器、支持向量机分类器和遗传神经网络预测器建构了原油含水率预测模型，对原油含水率的高精度、智能化测量进行研究。实验结果表明，该模型在很大程度上改善了油水乳化液相转变、黏度、温度、矿化度等因素对原油含水率测量的影响，具有良好的模式识别性能和稳健的泛化能力，克服了原油含水率常规测量方法的弊端，是一种高精度智能化新型测量方法。

Research into Kinetic Model of Water Content in Crude Oil Based on Artificial Intelligent Technique

The measurement of water content in crude oil based on the method of dielectric coefficient is affected by multi-factor, including flow states of oil/water mixture, temperature, salinity content, non-linear character, etc. In this paper, some influencing-parameters are measured using a multi-sensor measuring system designed for oil/water two-phase flow experiments, and a compound prediction model of water content in crude oil is established by combining rough set preprocessor, support vector machine classier and genetic neural network predictor. The research result shows that the intelligent hybrid prediction model overcomes the short-comings of the present measuring methods, evidently improving the measuring precision for water content under the influence of flow states, temperature, salinity content, viscosity of emulsion, gains a good performance of pattern recognition and generalization ability, is proved to be a novel intelligent measuring method with a extensive application prospect in petroleum industry.