基于RBF神经网络瓦斯水合物生成压力预测研究

Generate Pressure Prediction of Gas Hydrate Formation Based on RBF Neural Network

瓦斯水合物生成是复杂的结晶过程,不同组分和浓度瓦斯生成水合物时压力等热力学参数的获取对水合物技术的应用具有非常重要的意义。鉴于此,利用径向基神经网络方法对瓦斯水合物生成压力进行了预测。针对瓦斯水合物生成边界条件,确定了RBF神经网络的输入、输出向量,建立了RBF神经网络瓦斯水合物生成压力计算及预测模型,并用实验数据进行了验证。结果表明,该模型对瓦斯水合物生成压力的拟合和预测具有计算精度高、速度快等优点。RBF神经网络研究为瓦斯水合物生成压力预测提供了一种新途径。

The formation of gas hydrates is a complex crystallization process, and the acquisition of thermodynamic parameters （such as pressure） when gas of different components and concentrations form hydrates is very important for hydrate technology application. In view of this, the formation pressures of hydrates were forecasted by utilizing the method of RBF neural network in this paper. The input and output vectors of RBF neural network were determined and the calculating and forecasting model of hydrate formation pressures was built, according to boundary conditions of gas hydrate formation. And the model was verified by utilizing many experimental data. The results indicated that the model has many advantages such as high calculation precision and fast speed when fitting and forecasting gas hydrate formation pressures. The research of RBF neural network will be a new approach to the gas hydrate formation pressure forecast.