| Curtain grouting construction scheme optimization is a complex problem with multiple interrelated design variables and optimization objectives,and its advantages and disadvantages will directly affect the final construction effect.However,the determination of the existing curtain grouting construction scheme still relies mainly on limited grouting test holes and highly relies on the personal experience of grouting engineers,which is highly subjective.And a few combined with numerical calculation of curtain grouting scheme optimization research only from a single goal,it is difficult to obtain the optimal construction scheme of safety and economy,and there are shortcomings of low calculation efficiency and poor accuracy.In view of the above deficiencies,this paper proposes a three-dimensional fine fracture network curtain grouting numerical simulation method considering variable pressure grouting,and establishes a multi-objective optimization model of curtain grouting construction scheme based on combined residual correction agent model.The main research contents are as follows :(1)In view of the lack of comprehensive consideration of safety and economy,and the low computational efficiency and poor accuracy of existing curtain grouting scheme optimization,a curtain grouting construction scheme optimization method based on grouting effect prediction agent model is proposed.Firstly,combined with the curtain grouting construction specification,a multiobjective optimization mathematical model of curtain grouting construction scheme is established with the optimization objectives of post-irrigation water permeability and curtain cost.Secondly,a model solving method based on NSGA-Ⅱ algorithm is proposed to realize the fast solution of Pareto solution set.In order to further improve the efficiency of the model,the grouting effect prediction combined residual correction surrogate model is used to replace the time-consuming and labor-intensive grouting numerical simulation,and the high-precision mapping between the grouting construction parameters and the post-grouting effect is constructed to realize the fast and high-precision solution of the optimization objective function of the curtain grouting construction scheme.(2)For the existing proxy model will focus on the algorithm itself and parameter optimization,ignoring the effective information hidden in the prediction residual,resulting in poor prediction accuracy,a combined residual correction proxy model for grouting effect prediction based on ICSA optimization is proposed.Firstly,based on the combination prediction theory,the Gaussian process regression with outstanding performance in dealing with small sample problems is used as the basic prediction method,and the combination residual model is constructed by combining the BP neural network with good generalization performance and the ELMAN neural network.At the same time,the improved Cuckoo Search Algorithm(ICSA)is used to optimize the combination weight,and then the combination residual correction proxy model(ICSA-COGPR)for the prediction of grouting effect is established,which not only overcomes the shortcomings of the single algorithm that the prediction accuracy is prone to oscillation.More deeply mining the effective information hidden in the prediction residual;in order to reduce the subjectivity of the combination weight,the global and local search capabilities of the general cuckoo search are improved by using the composite method and chaos theory,which effectively improves the optimization accuracy and robustness of the algorithm and ensures the high-precision prediction of the grouting effect.(3)In view of the existing grouting numerical simulation mostly adopts fixed grouting pressure and water cement ratio,which is inconsistent with the construction technology of variable pressure grouting in practical engineering,and the accuracy of fracture modeling is poor,a three-dimensional fine fracture network curtain grouting numerical simulation method considering variable pressure grouting is proposed.Firstly,the correlation consistency criterion is introduced to improve the Latin hypercube sampling to reduce the randomness of sampling results,and a threedimensional fine fracture network model modeling method based on improved Latin Hypercube Sampling(ILHS)is established.Secondly,according to the construction specifications and the implementation process of pressure water test,the grouting pressure and water cement ratio are dynamically adjusted by the flow rate and cumulative grouting volume in the grouting hole in the statistical simulation process,and the simulation method of multi-hole sequential grouting and post-grouting permeability considering variable pressure grouting is established to realize the fine simulation of curtain grouting construction,so as to obtain the safety and economic indexes of grouting scheme optimization.(4)Taking the curtain grouting project of 18 # dam section of a high dam in Southwest China as an example,the optimization of curtain grouting construction scheme is studied to verify the superiority of the proposed method.Firstly,the sample set of surrogate model for grouting effect prediction is established by using the proposed numerical simulation method of three-dimensional fine fracture network curtain grouting considering variable pressure and slurry.Compared with the numerical simulation method of grouting with fixed pressure and water-cement ratio,it is proved that the proposed method has higher simulation accuracy.Secondly,the ICSA-COGPR surrogate model was used to predict the grouting effect.The results showed that ICSA-COGPR performed better than CSACOGPR,COGPR,BP-GPR,ELMAN-GPR and GPR models in mean absolute percentage error,mean absolute error,root mean square error and multiple correlation coefficient.Finally,the optimization method proposed in this paper is used to optimize the curtain grouting construction scheme in this region.The post-irrigation permeability of the optimized construction scheme is 33.67 % higher than that of the initial scheme,and the cost is reduced by 2.53 %.It proves that the optimized construction scheme has been improved both in safety and economy,and the superiority of the proposed method is verified. |