介绍了LVQ神经网络的结构,学习算法及其步骤,利用煤矿风机振动频域的特征向量作为学习样本,建立与风机故障类型的映射关系,达到对风机的故障诊断。仿真结果表明,LVQ网络训练速度快,准确可靠,具有很好地泛化能力。LVQ网络无需对输入向量进行归一化,正交化,结构简单,是一种有效的诊断方法。 The structure, learning algorithm and steps of LVQ neural network are introduced. Using the eigenvectors of coal mine fan vibration frequency domain as a learning sample, the mapping relationship between fan fault types and LVQ neural network is established, and the fan fault diagnosis is achieved. The simulation results show that LVQ network is fast, accurate and reliable, and has good generalization ability. LVQ network without the need to normalize the input vector, orthogonal, simple structure, is an effective diagnostic method.