针对大型旋转机械多故障同时性诊断问题，基于人工神经网络，构造了一种由多个子网络组成的分级诊断网络（ＨＤＡＮＮ）．该网络旨在将一个大的分类模式空间划分为几个小的子空间，以便对各子网络进行有效的训练，提高各子网络的分类能力，从而使整个网络具有高精度的多故障同时性诊断能力．测试结果表明：ＨＤＡＮＮ网络不仅能准确地对单故障进行诊断，而且多故障同时存在的情况下，也能有效地识别出各种故障，该网络具有较高的诊断精度，可用于旋转机械工况实时监测和诊断场合． Aiming at the problem of multi-fault simultaneous diagnosis of large rotating machinery, a hierarchical network (HDANN) composed of several sub-networks is constructed based on artificial neural network. This network aims to divide a large classification pattern space into several small subspaces in order to train each sub-network effectively and improve the classification ability of each sub-network so as to make the whole network have high-precision multi-fault simultaneity Diagnostic ability. The test results show that HDANN network not only can accurately diagnose a single fault, but also can effectively identify various faults under the condition of multiple faults. The network has high diagnostic accuracy and can be used in rotating machinery Real-time monitoring and diagnosis of occasions.