【摘要】 概率神经网络(PNN)是一种基于概率密度函数理论且泛化能力很强的神经网络,并且能够广泛地应用于模式识别等领域。针对储层产能的预测问题,提出了一种具体的概率神经网络方法,包括网络模型的构造、学习训练和预测识别等步骤。基于MATLAB6.5设计出概率神经网络的具体应用软件,实际应用表明,在储层产能预测中效果显著。与BP网络进行对比实验,其预测正确率优于BP网络。更多还原

【Abstract】 Probabilistic neural network, PNN, is that of a powerful generalization capacity based on the probabilistic density function. It can be widely used in the field of pattern recognition. In terms of the forecasting problem of reservoir productivity, a probabilistic neural network is brought forward in detail, including the construction of the neural network, the process of training of PNN and recognition forecasting. By used of the MATLAB 6.5,the neural network is programmed and applied to the forecasting of reservoir productivity. Actual example shows the result is perfectly tally with the fact. The comparing experiment shows the applied effect of PNN is superior to that of BP network.更多还原