【摘要】 为有效预测套损发生,掌握油水井套管的状况,减小套损所带来的损失,基于大庆油田南一区井网的现有资料,综合分析采集到的各种因素,建立了基于主成分分析的遗传神经网络模型。该模型首先对原始数据进行主成分分析,并将得到的主成分作为神经网络的输入,然后用遗传算法确定了网络的最佳初始权值和阈值,最后用神经网络进行预测。结果表明,该方法油井和水井的预测准确率分别达85%和82. 5%,证明经过主成分分析和遗传算法优化的BP神经网络的准确性和可靠性。更多还原

【Abstract】 In order to effectively predict casing damage,grasp the casing condition of oil and water wells and reduce the losses of casing damage,a genetic neural network model based on principal component analysis was established based on the existing data of well pattern in Nanyi District of Daqing Oilfield. Firstly,the original data is analyzed by principal component analysis(PCA),and the obtained principal components are used as the inputs of neural network. Then the optimal initial weights and thresholds of the neural network are determined by genetic algorithm. Finally,the casing damage is predicted using the neural network. The prediction result of a case shows that the casing damage prediction accuracy of oil and water wells using this method is 85. 0% and 82. 5% respectively,which proves the accuracy and reliability of the BP neural network optimized by principal component analysis and genetic algorithm.更多还原