证据理论结合灰色神经网络混合识别方法在苏通长江大桥状态预测中的应用

为解决千米级斜拉桥在施工过程中由于观测噪声及初始构件误差等因素影响下的结构响应预测问题，提出了D-S证据理论与灰色神经网络结合的混合识别方法。融合后的混合识别方法，在数据不完备的情况下，具有良好的预测功能。根据实际结构识别结果验证了方法的有效性和准确性，分析并比较了本文方法与传统识别方法的结果。实践研究表明，参数预测的D-S证据理论与灰色神经网络混合识别方法能够显著改善传统方法的识别准确性问题。

Condition prediction based on evidence theory and grey-neural network model for Su-Tong bridge

To improve the accuracy of condition prediction for thousand-meters scale cable-stayed bridge under the influences of indeterminate factors such as measurement noise and initial component error during the construction process, a condition prediction method based on evidence theory and grey-neural-network model was proposed. The application of fusion theory makes the effective prediction possible even in the case of insufficient samples or incomplete data. The validity and accuracy of the proposed method were demonstrated through an example of actual structure. The results of traditional method and the proposed one were analysed and compared. Practical results show that the parameter prediction method based on evidence theory and grey-neural-network model can get better estimation accuracy than the traditional method.