确定大坝效应量分量比例的神经网络方法

为了评价水位、温度等环境量对大坝观测效应量的综合影响程度，并由此定量分析评价大坝的运行状态，提出了确定大坝观测效应量的各分量比例的神经网络方法。根据大坝的观测资料，利用误差反向传播的神经网络(BP网络)建立效应量与环境量关系的神经网络模型，BP网络的输入变量为水位、温度、时效等环境量因子，网络的输出变量为效应量。利用网络的权值来表示网络的输入变量对网络的输出变量的影响程度，从而确定水位、温度等分量占效应量的比例。文章通过工程实例验证了该方法的有效性。结果表明，该方法简便实用，可定量分析水位、温度等环境对效应量的影响程度，有助于进一步分析大坝的安全性态。

Neural network method for determining the componentproportion of dam effect variable

In order to evaluate the behavior of a dam and the comprehensive significance of environmental variables of water level and temperature on the effect variable, the authors presented a neural network method for determining the proportion of components of the effect variable. Based on observation data, the neural network model was set up by using BP network. The input variables were environmental variable, such as water level, temperature and time effect. The output variable was effect variable of the dam. The weight of the neural network was used to explain the relative significance of inputs on the output and the proportion of components of the effect variable was determined. The method was applied to a project and the result proved that the method was simple in practice and effective in determining the component proportion of the effect variable and evaluating the behavior of dams.