基于神经网络的油基钻井液体系CQ-WOM高温高压流变性预测

TO PREDICT HPHT RHEOLOGICAL PROPERTIES OF OIL BASED DRILLING FLUID(CQ-WOM) BASED ON NEURAL NETWORK

应用人工神经网络,对油基钻井液体系CQ-WOM在不同密度和高温高压条件下的Φ3读值、塑性黏度和动切力三个流变性参数进行预测。在模型训练中,文章采用了神经网络集成来提高人工神经网络的泛化能力。模型检验结果表明,采用神经网络集成后,预测精度大幅提高,可以快速准确地预测油基钻井液模型的高温高压流变性。运用神经网络集成进行预测,能大幅提高预测精度,可快速准确预测。

Artificial neural nestwork was used to predict accurately Φ3 reading values,plastic viscosity and yield point of oil based drilling fluid( CQ-WOM) under different densities and HPHT conditions. The neural network assembly was used to improve the generalization ability of the artificial neural network model in the training. The model test results indicate that neural network assembly can improve the accuracy of prediction,can predict the rheological parameters of oil based drilling fluid( CQ-WOM) under high temperature and high pressure rapidly and accurately.