专家神经网络是由专家系统产生出的事件驱动无环路神经对象网络。这些神经对象处理信息的非线性复合函数比正常的神经网络结点处理器更复杂。本文研究了无环路网络的ＢＰ学习算法，这一算法将传统ＢＰ算法与专家神经网络特征结合起来，提供了一种获取知识的方法。 Expert neural network is an event driven loop-free neural network produced by expert system. The nonlinear complex functions that these neural objects process information are more complex than normal neural network node processors. This paper studies BP learning algorithm for loop-free networks. This algorithm combines traditional BP algorithm with expert neural network features to provide a method to acquire knowledge.