基于LSSVM与GA的混凝土抗冻性多目标配合比优化研究

Multi-objective Mix Proportion Optimization of Frost Resistance of Concrete based on LSSVM and GA

目的提出一种LSSVM-GA模型,以实现对混凝土抗冻性的精确预测及快速而高效的混凝土配合比优化设计。方法首先利用最小二乘支持向量机(LSSVM)实现对混凝土抗冻性的高精度预测,然后将抗冻性回归预测函数作为适应度函数,以混凝土抗冻性和经济成本为优化目标,结合工程要求及相关规范建立配合比参数约束条件,最后通过遗传算法(GA)实现混凝土配合比的多目标优化设计。结果以吉林省某工程项目为例进行应用分析,计算结果表明该模型对混凝土抗冻性预测的RMSE低至0.0025,R 2高达0.976,预测结果精度较高。将所得LSSVM预测回归函数作为适应度函数,结合GA算法进行多目标优化,能够获得符合规范和工程要求的混凝土最优配合比。结论笔者构建的LSSVM-GA模型,以混凝土抗冻性和经济成本为目标,实现了更加智能化和精确化的混凝土配合比多目标寻优。

Propose a LSSVM-GA model to achieve accurate prediction of concrete frost resistance and rapid and efficient concrete mix optimization design.First,the least squares support vector machine(LSSVM)is used to achieve high-precision prediction of the frost resistance of concrete.Then the frost resistance regression prediction function is used as the fitness function,the concrete frost resistance and economic cost are the optimization objectives,and the mix ratio parameter constraint conditions are established in combination with engineering requirements and related specifications.Finally,the genetic algorithm(GA)is used to realize the multi-objective optimization design of concrete mix ratio.Taking an engineering project in Jilin Province as an example,the calculation results show that the RMSE of the model for predicting the frost resistance of concrete is as low as 0.0025,and the R 2 is as high as 0.976.The accuracy of the prediction results is high.Using the obtained LSSVM prediction regression function as the fitness function,combined with the GA algorithm for multi-objective optimization,the optimal mix ratio of concrete that meets the specifications and engineering requirements can be obtained.Taking concrete frost resistance and economic cost as the goal,the LSSVM-GA model is constructed to achieve a more intelligent and accurate multi-objective optimization of concrete mix ratio in this paper.