论文部分内容阅读
了解油井生的井底流压大小是现场生产测试和分析中的一项重要工作。自喷井的井底流压值与产油量、含水、气油比、流体性质等参数呈复杂的非线性关系。神经网络具有表达任意非线性映射的能力，可以将其应用于建立自喷井井底流压预测模型。用一定数量的实测井底流压及相应的有关参数，根据BP神经网络学习算法对网络进行训练。训练后的网络就可用于预测一定生产条件下的井底流压，减少实测次数。实例计算表明：用神经网络建立自喷井井底流压预测模型是可行的，计算精度高。
Knowing the size of the bottomhole flow pressure in an oil well is an important task in field test and analysis of production. The bottom hole flow pressure value of self-injection well has a complex and nonlinear relationship with the parameters of oil production, water content, gas-oil ratio and fluid properties. Neural networks have the ability to express any nonlinear mapping and can be applied to establish a prediction model of bottom hole flow pressure from a blowhole. With a certain number of measured bottom hole pressure and the corresponding parameters, according to BP neural network learning algorithm to train the network. The trained network can be used to predict the bottom hole flow pressure under certain production conditions and reduce the number of measurements. The case study shows that it is feasible to establish a prediction model of bottom hole flow pressure of self-propelled well by neural network, and the calculation accuracy is high.