【摘要】 本义以反应谱的特征参数提取为主题，讨论了被估计目标特征的自动提取问题．文中给出了反应谱特征参数提取的神经网络模型，该应用模型通过学习积累有关特征的知识，然后利用已学到的知识自动完成新目标相应特征的提取，体现出神经网络用于设计智能系统的潜力。该模型的输入为地震反应谱，输出为提取出的特征参数，可以是标准谱形参数、拐点周期、谱平台宽度等。在本文中，通过由反应谱直接提取标准谱形参数的例子，说明了模型和方法、并进行试验计算。参照这些可设计出类似的其它应用模型。更多还原

【Abstract】 The present paper focuses on automatically extracting the characteristic parametersof response spectrum based on neural network method. The application model established isfirst trained using some sample data to accumulate "knowledge", and then it extracts thecharacteristic parameters of the new input vector automatically by means of the obtained"knowledge". The input of the model is earthquake response spectrum, and the output of themodel is some characteristic parameters, such as the shape parameters of response spectrum,period value of Tg, platform width of standard spectrum, etc. In this paper, the model andmethod about extracting shape parameters of response spectrum is described. An example isshown and computing experiment is made. According to above mentioned method some similarmodels can be designed and developed.更多还原