时间序列建模在卡钻类别判断中的应用研究

为了提高卡钻预测中卡钻类别判断的准确度,以青海地区地热勘探井实钻数据为基础,结合时间序列分析建模方法,提出了一种适合卡钻类别判断的方法。通过时序模型对未来钻井数据进行预测处理,运用Matlab软件对各个ARMA模型做功率谱估计,比较相邻两个ARMA模型的功率谱密度,计算各个参数的功率谱偏差值,进行数值仿真,当某一参数其功率谱偏差值出现明显异常时,则预判断这一时刻可能发生此参数对应类别的卡钻事故。引入多因素时序建模方法,运用SPSS软件做多因素模型,计算主要参数的预测区间,当预测值超出预测区间时,则可以判断发生对应类别的卡钻事故。最终证实,采用此方法能够实现对钻井过程中未来卡钻事故的类别判断,在实际钻井中有较高的可扩展性及应用价值。

Application of time sequence modeling in category judgment for sticking of drilling rig

In order to improve the accuracy of judging drilling rig sticking category in sticking prediction,on the basis of real drilling data of the geothermal exploration well in Qinghai region,a new method of category judgment for sticking of drilling rig is proposed in combination with the method of time sequence analysis modeling. The future drilling data is predicted and pro-cessed by the time-sequence model. Power spectrum estimation of each ARMA model is conducted with Matlab software. The power spectral density of two adjacent ARMA models is compared. The power spectrum deviation of each parameter is calculated and its numerical simulation is carried out. When the power spectrum deviation of a parameter is obviously abnormal,the sticking accident category corresponding to this parameter,which may occur in this moment,is judged in advance. The multi-factor time-sequence modeling method was introduced. The multi-factor model was built with SPSS software to calculate the prediction inter-val of main parameters. When the predicted value exceeds the prediction interval, the corresponding category of sticking acci-dent can be determined. The result confirms that the method can realize the category judgment of future sticking accident in the drilling process and has high extendibility and application value in the actual drilling.