基于LS-SVM的天然气水合物生成条件预测模型建立

Establishment of the Prediction Model of the Forming Conditions of Natural Gas Hydrate Based on Least Square Support Vector Machine

在天然气管线内生成的水合物会严重影响天然气的开采、运输,因而天然气水合物的预测方法和防治措施备受重视。针对天然气水合物生成条件,考虑天然气组分对水合物生成的影响,为简化计算、提高预测精度,引入一种能够很好解决复杂物理问题的最小二乘支持向量机(LS-SVM),并且通过Matlab语言编程,建立了一种包含CH4浓度、CO2浓度、H2S浓度以及水合物生成温度为输入,水合物生成压强为输出的天然气水合物生成条件预测模型,同时将实验数据作为最小二乘支持向量机训练数据并进行预测分析。结果表明,该预测模型不仅拥有较高的预测精度,而且方法简单、可行,为天然气水合物生成条件预测提供了一种新的解决方法。

The hydrate generated in the natural gas pipeline will seriously affect the natural gas production, transportation, and the prediction methods and control measures of natural gas hydrate （NGH） have attracted great attention. For natural gas hydrate formation conditions, considering the influence of the components on the gas hydrate formation, in order to simplify the calculation and improve the prediction accuracy, through introducing least squares support vector machines （LS - SVM） and using Matlab language programming, a prediction model of natural gas hydrate formation conditions was established with the concentrations of CH4 , H2S , CO2 and hydrate formation temperature as the input, the pressure of hydrate formation as the output. At the same time, the experimental data were used as Least Squares Support Vector Machine training data to carry out the forecast analysis. The results show that the forecasting model has higher prediction accuracy, and the method is simple and feasible, can provide a new solution for orediction of natural gas hydrate formation conditions.