基于BP神经网络的石油运聚系数预测模型

Forecast Model for Oil Migration and Accumulation Coefficient Based on BP Neural Network

根据一系列石油运聚单元的解剖结果,筛选出影响石油运聚系数的主控地质因素。建立了以主控地质因素为输入向量,运聚系数为输出向量的BP神经网络石油运聚系数预测模型。应用结果表明,所建立的BP神经网络模型具有较好的预测效果,平均相对误差为10.89%,有效性指数达到92.51%,且运聚系数预测精度高于多元非线性回归模型的预测。

Based on the data from a series of petroleum migration and accumulation units,the major geologic factors for controlling oil migration and accumulation coefficient are selected.The forecast model for oil migration and accumulation coefficient based on BP neural network is developed by taking the major geologic factors as the input vectors and the oil migration and accumulation coefficients as output vectors.It is indicated that the applied result of this model is in good agreement with the observed data with average relative error of 10.89% and the corresponding agreement index is about 92.51%.Moreover,the prediction precision by using this model is much higher than that by using multi-element nonlinear regression model.