在对桥梁损伤评估系统模型的基础上,提出了基于DE-BP神经网络的损伤评估方法。该方法引入差异演化算法(DE算法),通过对已有的桥梁损伤评估实例对神经网络进行训练,可使神经网络较好地表达评估结果与评价因素之间的关系。在网络学习过程中充分引入DE算法,避免了BP神经网络容易陷入局部最值及收敛慢的缺点,经该方法训练好的网络可以对实际桥梁进行评估。 Based on the bridge damage assessment system model, a damage assessment method based on DE-BP neural network is proposed. This method introduces the differential evolution algorithm (DE algorithm). By training the neural network on the existing example of bridge damage assessment, the neural network can well express the relationship between the evaluation results and the evaluation factors. The DE algorithm is fully introduced in the process of network learning to avoid the disadvantage that BP neural network easily falls into the local maximum and converges slowly. The network trained by this method can evaluate the actual bridge.