【摘要】 神经网络专家系统为稠油开采方式的筛选提供了一种比人工估测更为全面精确的手段。文中采用三层BP网构建神经网络专家系统 ,利用优化算法改进神经网络的结构 ,并在动态调整网络参数时 ,不仅考虑了误差的变化 ,还考虑了误差变化的变化趋势 ,从而改进了BP网络的学习性能 ,大大增强了BP网络的自适应性 ,提高了学习速度。实例分析证明系统学习速度快、性能稳定 ,结果与实际情况吻合很好 ,具有一定的通用性 ,可广泛用于生产生活多个领域更多还原

【Abstract】 The Nerve Network Expert System(NNES)serves as a more comprehensive and accurate solution to screening of heavy oil production schemes than manual estimation.This paper adopts a three-layer back-propagation(BP) model in constructing the NNES.Furthermore,the structure of the nerve network is constructed by using optimization algorithm.In the dynamic adjustment of network parameters,not only the variation of errors but also the tendency of the variation of errors are taken into consideration,which enhances the learning ability and self-adaptability of the BP model.In-situ data analysis proves that the system yields satisfactory results with a quick learning speed and a stable performance.The system has considerable versatility and can be used in other areas of production or research.更多还原