人工智能方法在土木工程监测中的运用
|
摘要
随着计算机技术的快速发展,土木工程领域的监测、预测方法得到了不断的更新。以大体积混凝土浇筑过程为工程背景,结合BP、GA-BP、PSO-BP、SOM、CNN、SVM和PNN算法建立预测模型。通过实测数据和预测模型得出:大体积混凝土水化放热会使得内部温度在2 d内先升高后下降;以统计率理论为基础的SVM、PNN神经网络和以深度学习为基础的CNN神经网络所建立的预测模型与实测数据非常吻合,其误差在2%以内;BP神经网络预测误差在10%左右,但通过遗传算法进行改进后误差在5%左右。结合七种人工智能方法,选择合适的算法并进行优化,可为今后土木工程领域监测-预测-预警提供依据。
With the rapid development of computer technology, monitoring and forecasting methods in the field of civil engineering have been constantly updated. Taking the pouring process of mass concrete as the engineering background, combined with BP, GA-BP, PSO-BP, SOM, CNN, SVM and PNN, the prediction model was established. According to the measured data and prediction model, it can be concluded that the internal temperature of mass concrete increases first and then decreases within 2 days due to hydration and heat release. The error of prediction model based on SVM, PNN neural network and CNN neural network is less than 2%. The prediction error of BP network is about 10%, but the error is 5% after improvement by genetic algorithm. Combining seven kinds of artificial intelligence methods, choosing the appropriate algorithm and optimizing can provide the basis for monitoring, predicting and early warning in the field of civil engineering in the future.
With the rapid development of computer technology, monitoring and forecasting methods in the field of civil engineering have been constantly updated. Taking the pouring process of mass concrete as the engineering background, combined with BP, GA-BP, PSO-BP, SOM, CNN, SVM and PNN, the prediction model was established. According to the measured data and prediction model, it can be concluded that the internal temperature of mass concrete increases first and then decreases within 2 days due to hydration and heat release. The error of prediction model based on SVM, PNN neural network and CNN neural network is less than 2%. The prediction error of BP network is about 10%, but the error is 5% after improvement by genetic algorithm. Combining seven kinds of artificial intelligence methods, choosing the appropriate algorithm and optimizing can provide the basis for monitoring, predicting and early warning in the field of civil engineering in the future.
引文
1杜德润,仇德伦,李爱群,等.无损检测,2004,26(8),383.
2吴大宏,赵人达.四川建筑科学研究,2002,28(3),4.
3 Albuthbahak O M,Alkhudery H H.International Journal of Civil engineering&Technology,2018,9(2),265.
4 Chen F C,Jahanshahi M R.IEEE Transactions on Industrial Electronics,2018,65(5),4392.
5张帆,胡伍生.东南大学学报(英文版),2013,29(4),441.
6肖进胜,刘恩雨,朱力,等.光学学报,2017(3),96.
7 Han B,Xiang T Y,Xie H B.Engineering Structures,2017,142(7),46.
8丁杨.河北工程大学学报(自然科学版),2016,33(2),30.
9 Hoseinian Fatemeh Sadat,Abdollahzadeh Aliakbar,Rezai Bahram.Journal of Central South University,2018,25(1),151.
10戴隆州,吴永明,李少波,等.计算机应用研究,2018,35(1),145.
11丁杨,陈希杰,杜欣,等.安徽理工大学学报(自然科学版),2017,37(4),32.
12 Deng F,He Y,Zhou S,et al.Construction and Building Materials,2018,175,562.
13张莉,卢星凝,陆从林,等.中国科学技术大学学报,2017(1),1.
14王彦富,李玉莲,张彪,等.安全与环境学报,2016,16(5),66.
15翟社平,郭琳,高山,等.小型微型计算机系统,2018,39(5),995.
2吴大宏,赵人达.四川建筑科学研究,2002,28(3),4.
3 Albuthbahak O M,Alkhudery H H.International Journal of Civil engineering&Technology,2018,9(2),265.
4 Chen F C,Jahanshahi M R.IEEE Transactions on Industrial Electronics,2018,65(5),4392.
5张帆,胡伍生.东南大学学报(英文版),2013,29(4),441.
6肖进胜,刘恩雨,朱力,等.光学学报,2017(3),96.
7 Han B,Xiang T Y,Xie H B.Engineering Structures,2017,142(7),46.
8丁杨.河北工程大学学报(自然科学版),2016,33(2),30.
9 Hoseinian Fatemeh Sadat,Abdollahzadeh Aliakbar,Rezai Bahram.Journal of Central South University,2018,25(1),151.
10戴隆州,吴永明,李少波,等.计算机应用研究,2018,35(1),145.
11丁杨,陈希杰,杜欣,等.安徽理工大学学报(自然科学版),2017,37(4),32.
12 Deng F,He Y,Zhou S,et al.Construction and Building Materials,2018,175,562.
13张莉,卢星凝,陆从林,等.中国科学技术大学学报,2017(1),1.
14王彦富,李玉莲,张彪,等.安全与环境学报,2016,16(5),66.
15翟社平,郭琳,高山,等.小型微型计算机系统,2018,39(5),995.