基于智能算法的河港码头混凝土碳化深度预测

Concrete carbonation depth prediction of river port using the intelligent algorithms

由于混凝土碳化所引起的河港码头构件材料性能劣化是导致其整体结构发生耐久性失效破坏的主要原因之一,混凝土碳化深度的预测以及结构物服役寿命的评估是河港码头运行维护过程中的关键。然而,基于Fick第一定律的传统碳化深度预测方法所得结果的精度与实测样本容量的大小呈正相关,当实测样本值有限时,其预测精度难以保证。为此,针对有限样本支撑下河港码头混凝土碳化深度预测难、精度低等问题,提出了一种基于人工智能的计算方法,用以更加精确地预测河港码头混凝土的碳化深度。该智能算法包括BP神经网络模块与樽海鞘群算法(SSA)模块,SSA模块负责修正BP神经网络模块的设置参数,完成BP神经网络设置参数的自适应优化;BP神经网络模块主要负责利用SSA模块得到的设置参数对样本进行训练及测试,两者交叉融合,实现对河港码头混凝土碳化深度的精确预测。对融合BP神经网络和SSA的算法流程及实施步骤进行了详细描述,并通过试验实测数据验证了智能算法预测混凝土碳化深度结果的精度及可行性。

The deterioration of material properties of river port components caused by concrete carbonation is one of the main reasons for the durability failure of the whole structure.The prediction of concrete carbonation depth and the evaluation of service life are the key in the wharf operation and maintenance.However,the accuracy of the traditional carbonation depth prediction method based on Fick′s first law is positively related to the size of the measured sample size.When the measured sample is limited,the prediction accuracy is difficult to guarantee.To resolve the difficulty and inaccuracy problem under limited samples,an artificial intelligence based calculation approach is proposed to do the concrete carbonation depth prediction of river port using back propagation neural network(BPNN)and salp swarm algorithm(SSA).SSA is applied to tune the setting parameters of BPNN and achieve the adaptive optimization of these parameters.BPNN is used to train and test the samples based on the parameters obtained by SSA.The combination of the two can accurately predict the concrete carbonation depth of river port.The algorithm flow and implementation steps of the fusion of BP neural network and SSA are described in detail,and the accuracy and feasibility of the proposed approach to predict the carbonation depth of concrete are verified by the experimental data.