基于随机森林-NSGAⅡ高性能混凝土耐久性配合比的多目标优化研究  

Multi-Objective Optimization of High-Performance Concrete Durability Mix Ratio Based on RF-NSGAⅡ

作　　者：吴贤国[1] 王雷[1] 陈虹宇[2] 冯宗宝 覃亚伟[1,3] 徐文胜 WU Xianguo;WANG Lei;CHEN Hongyu;FENG Zongbao;QIN Yawei;XU Wensheng(School of Civil and Hydraulic Engineerng,Huazhong University of Science and Technology,Wuhan 430074,China;School of Civil and Environmental Engineering,Nanyang University of Technology,Singapore 639798;Wuhan Huazhong University of Science and Technology Test Technology Co.,Ltd.,Wuhan 430074,China)

机构地区：[1]华中科技大学土木与水利工程学院,武汉430074 [2]南洋理工大学土木工程与环境学院,新加坡639798 [3]武汉华中科大检测科技有限公司,武汉430074

出　　处：《材料导报》2022年第17期111-117,共7页Materials Reports

基　　金：国家自然科学基金(51378235;71571078;51308240)。

摘　　要：针对高寒复杂环境下高性能混凝土耐久性预测和优化问题,本工作研发了一种结合随机森林与带精英策略的非支配排序遗传算法(NSGAⅡ)的混合模型,以实现高精度的混凝土性能预测以及多目标配合比优化。以混凝土耐久性的重要指标氯离子渗透系数和28 d抗压强度为研究目标,基于正交试验设计和工程实际试验样本建立混凝土数据集。利用随机森林对混凝土氯离子渗透系数和强度进行预测,得到氯离子渗透系数和强度与配合比的非线性映射关系函数,将其作为对应优化目标的适应度函数,再引入混凝土成本作为另一个优化目标的适应度函数。依据规范和工程要求,建立原材料及配合比之间的约束,采用NSGAⅡ进行混凝土配合比的多目标优化。研究表明,利用随机森林对混凝土性能进行预测的精度很高,且利用NSGAⅡ算法进行多目标配合比优化的效果很好。将优化配合比方案进行试验验证,发现模型优化结果与实际试验结果误差很小。这说明混凝土配合比符合规范且满足工程项目对耐久性能、强度和工作性能的要求,体现了该模型的智能化、精准化,可对工程实践中混凝土配合比的优化提供指导。Aiming at the prediction and optimization of the durability of high-performance concrete in the alpine complex environment, this work developed a hybrid model that integrates random forest(RF) with nondominated sorting genetic algorithm-Ⅱ with elite strategy(NSGA Ⅱ) to achieve high-precision concrete performance prediction and multi-objective mix ratio optimization. Taking the chloride ion permeability coefficient and 28 d compressive strength, which are important indicators of concrete durability, as the research objective, a concrete data set was established based on the orthogonal experimental design and engineering actual test samples. The random forest model was used to predict the chloride ion permeability coefficient and strength of concrete. A non-linear mapping function of chloride ion permeability coefficient and strength and mix ratio was obtained, which acted as the fitness function of the optimization target, while the concrete cost was introduced as the fitness function of another optimization target. According to the specifications and requirements of the project, the constraints of raw materials and mix ratios were established, and the NSGA Ⅱ algorithm was applied for multi-objective optimization of concrete mix ratio. Results show that the accuracy of using random forest to predict concrete performance is fairly high, and that the use of NSGA Ⅱ algorithm for multi-objective mix ratio optimization is quite effective. The optimized mix ratio scheme was tested and verified, and it was found that the error between the model optimization results and the actual test results was very small. It suggests that the concrete mix ratio meets specifications and engineering project requirements of durability, strength and work performance, reflecting the intelligence and precision of the model, which can provide guidance for the optimization of concrete mix ratio in engineering practice.

关 键 词：混凝土配合比优化 氯离子渗透系数 28d抗压强度 随机森林(RF) NSGAⅡ 

分 类 号：TU528[建筑科学—建筑技术科学]