油气井压裂效果通常用有效期的长短来反映。但有效期影响因素多,预测模型难以确定。BP神经网络能考虑多变量因素,把影响有效期的各种地层和压裂施工参数自行组织起来,建立起广义、准确的预测模型。实例研究表明,所建预测模型可大大提高压裂井有效期预测的精度、速度,具有较广阔的应用前景。 Oil and gas well fracturing effect is usually reflected in the length of validity. However, the effective period of many factors, the forecast model is difficult to determine. BP neural network can take multivariate factors into consideration, and organize various formation and fracturing parameters that affect the validity period to establish a generalized and accurate prediction model. The case study shows that the prediction model can greatly improve the prediction accuracy and speed of the fracturing well and has a broad application prospect.