Based on thorough analysing the factors which affect the seismic damage of single - story reinforced concrete industrial building , this paper introduces two kinds of methods which are applicable to prediction of seismic damage for single - story reinforced concrete industrial building . one is bp ( back propagation ) neural network method , which is based on the theory of artifical neural network . in this method , lm ( levenberg - marquardt ) optimization algorithm is applied to improve the performance of standard bp algorithm . the other is grey fixed weight cluster method , which is based on the theory of grey system 本文在充分分析单层钢筋混凝土柱工业厂房震害影响因素的基础上，提出了两种用于单层钢筋混凝土柱工业厂房震害预测的新方法。一种是基于人工神经网络理论的bp神经网络方法。在bp神经网络方法中，采用改进的bp算法( bplm算法)对网络进行训练，有效地改善了标准bp算法的不足之处。
3 . on the base of grey fixed weight clustering model , this paper transfers the absolute partition into relative partition of grey class , therefore creates the grey clustering model of time sequence , so that the model can solve the application of grey clustering model in the sequence data and broads the field of grey clustering model 以灰色定权聚类模型为基础，将灰类的绝对划分转变为相对划分，建立了时间序列灰色聚类模型，从而解决了灰色聚类模型在序列数据处理方面的应用问题，拓展了灰色聚类模型的应用领域。
