基于风险矩阵和支持向量机的套管寿命预测

Risk matrix and support vector machine-based prediction of the service life of casing

普通预测方法无法有效预测套管寿命,为此,建立了套管寿命预测模型。该模型基于支持向量机方法建立,通过对32个套管失效因素的影响等级及相应因素发生概率等级的定量分析,求得各因素对于套管失效影响的风险矩阵,并确定了油气井套管8种失效形式的危险性能指标。对60口非训练样本的套损井寿命进行预测,通过寿命预测值与现场实际寿命值的对比分析表明,该套管寿命预测方法对中长寿命油气井套管具有较高的预测精度。

Considering the fact that the ordinary prediction method cannot predict the service life of casing effectively,a kind of prediction model for the casing service life was established.The model was established on the basis of support vector machine.It made a quantitative analysis of the effect grades of the failure factors of 32 casings and the probability grades of the corresponding factors,finding out the risk matrix of the effect of the factors on casing failure and determining the danger property index for eight failure forms of the oil-gas well casing.A prediction was conducted of the service life of 60 untrained sample casing damage wells.A comparative analysis of the life prediction value and the field real life value indicates that the method can reach a higher precision in predicting the medium length life of the oil-gas well casing,thus providing an engineering reference.