论文部分内容阅读
本文在分析研究BP网络油气预测存在问题的基础上，提出了一种模糊神经网络。这种网络与BP网络相似，它是将模糊的概念结合于网络之中，使神经网络可以处理结构化的知识，亦即由专家给出的规则，从而提高了油气预测结果的可信度。同时，在网络训练时采用同伦学习算法，大大提高了网络训练的收敛速度，避免了用梯度下降法训练网络所产生的局部收敛现象。模糊神经网络已在大庆探区多个高分辨率区块进行了油气预测的实际应用，取得了较好的效果。
Based on the analysis of the existing problems of BP network oil and gas prediction, this paper proposes a fuzzy neural network. Similar to BP network, this kind of network integrates the concept of fuzzy into the network so that the neural network can deal with the structured knowledge, that is, the rules given by experts, so as to improve the credibility of the prediction result. At the same time, the homotopy learning algorithm is adopted in network training, which greatly improves the convergence speed of network training and avoids the local convergence caused by training the network with the gradient descent method. Fuzzy neural network has been in the Daqing exploration area a number of high-resolution blocks for oil and gas prediction of the practical application, and achieved good results.