论文部分内容阅读
在大量生油门限的预测方法中,统计预测法还未受到广泛的重视,它是以一定数量的已知数据为基础,通过一定的办法建立经验表达式,然后应用于实际。Connan(1974)首次建立了生油门限温度与生油岩地质年龄之间的函数关系,在此基础上,于志钧(1981,1983)做了一些补充和改进,并通过逐步回归分析,得到了一个新的表达式,其多重相关系数为0.931,比Connan的方程高。在实际应用中,Connan和于志钧的公式都不大方便,因为其因变量都是生油岩年龄值的对数。事实表明,若以生油岩的现今地层温度为因变量,则所得模型将更为有效合理。对于新的模型,笔者采用了较先进的叠代变换回归法(Box和Tidwell,1962)进行拟合,产生了比于志钧公式更高的拟合度。用门限已知的其他一些盆地数据对方程进行检验,结果是令人鼓舞的。预测的门限地温与实测值最大误差只有4℃。
In the prediction method of a large number of oil production thresholds, the statistical forecasting method has not received much attention. Based on a certain amount of known data, it establishes the empirical expression by a certain method and then applies it to practice. For the first time, Connan (1974) established a functional relationship between the gas-generating threshold temperature and the geologic age of the source rock. On this basis, Yu Zhijun (1981, 1983) made some additions and improvements and through stepwise regression analysis, A new expression has a multiple correlation coefficient of 0.931, higher than Connan’s equation. In practical applications, Connan and Yu Zhijun’s formula are not very convenient, because the dependent variable is the age of oil rock age logarithm. The facts show that the current model of formation temperature will be more effective and reasonable if the present formation temperature is taken as the dependent variable. For the new model, the author uses the more advanced iterative transformation regression (Box and Tidwell, 1962) to fit, resulting in a higher fitting than the Zhijun formula. The results are encouraging with some other basin data with known thresholds. The predicted threshold temperature and the maximum error of the measured value is only 4 ℃.