摘要
为提高集输管道中天然气水合物生成条件的预测精度,采用粒子群算法(PSO),优化支持向量机(SVM)的惩罚因子C和核函数参数g,对天然气水合物的生成压力进行了预测。结果表明,PSO-SVM模型选用径向基核函数的预测结果相对最优,最优惩罚因子C为97.5331,最优核函数参数g为0.6439,训练集和测试集的平均绝对比例误差(MAPE)分别为2.74%和2.84%;PSO-SVM模型对纯组分和多组分天然气水合物的适用温度分别为273.49~295.00 K和273.59~298.00 K;PSO-SVM模型预测纯组分和多组分天然气水合物时,平均平方误差(MSE)、平方相关系数(R~2)和MAPE分别为0.0003963、0.9996、2.84%和0.0006870、0.9983、2.74%。
In order to improve the prediction accuracy of natural gas hydrate formation conditions in pipeline, particle swarm optimization(PSO) was used to optimize the penalty factor C and kernel function parameter g of support vector machine(SVM)to predict the formation pressure of natural gas hydrate. The results show that the prediction results of PSO-SVM model using radial basis function are the best, the optimal penalty factor C is 97.5331, the optimal kernel function parameter g is 0.6439, the mean absolute proportional error(MAPE) of the training set is 2.74%, and the test set is 2.84%. The applicable temperature range of PSO-SVM model for pure component natural gas hydrate is 273.49~295.00 K, and for multi-component natural gas hydrate is 273.59~298.00 K. The mean square error(MSE), square correlation coefficient(R~2) and MAPE are 0.0003963, 0.9996 and 2.84% respectively when PSO-SVM model is used to predict pure component natural gas hydrate, and 0.0006870, 0.9983 and 2.74%respectively when predicting multi-component.
作者
范婕
许欣怡
周诗岽
周年勇
FAN Jie;XU Xinyi;ZHOU Shidong;ZHOU Nianyong(Jiangsu Key Laboratory of Oil-Gas Storage and Transportation Technology,School of Petroleum Engineering,Changzhou University,Changzhou 213164,Jiangsu,China)
出处
《天然气化工—C1化学与化工》
CAS
北大核心
2022年第5期171-176,共6页
Natural Gas Chemical Industry
基金
国家自然科学基金(51974037)。
关键词
天然气水合物
粒子群算法
支持向量机
生成条件预测
参数优选
natural gas hydrate
particle swarm optimization
support vector machine
formation conditions prediction
parameter optimization
