Research On SRM-CoRBF Multi-fidelity Surrogate Model For Maintenance Ventilation Of Long-distance Water Transmission Pipeline

Maintenance ventilation is essential to ensure the maintenance safety of longdistance water transmission pipelines.However,the existing research is mainly based on computational fluid dynamics(CFD)model to predict the ventilation effect,which has the disadvantages of low computational efficiency and high modeling cost.In addition,some studies based on single fidelity surrogate model instead of CFD model to predict ventilation effect,but it is still difficult to balance the contradiction between prediction accuracy and modeling cost.By integrating the data of high and low fidelity analysis models,the multi-fidelity surrogate model can take into account the prediction accuracy and modeling cost.However,it obtains the model hyper-parameters by minimizing the prediction error,and ignores the influence of structural complexity on the generalization performance of the model.In view of the above problems,this study established an improved cooperative radial basis function multi-fidelity surrogate model for prediction the ventilation effect of long-distance water transmission pipeline maintenance ventilation,and carried out the prediction research on the maintenance ventilation effect combined with the actual long-distance water transmission pipeline project.The main research results are as follows:(1)A method for predicting the ventilation effect of long-distance water transmission pipeline maintenance ventilation based on SRM-CoRBF multifidelity surrogate model is proposed.Firstly,a multi-fidelity surrogate model of long-distance pipeline water transmission pipeline maintenance ventilation SRM-CoRBF is constructed to overcome the shortco-mings of low efficiency and high computational cost of traditional CFD simulation and the difficulty of taking into account the prediction accuracy and modeling cost of single fidelity surrogate model.Secondly,a three-dimensional mathematical model for long-distance water transmission pipeline maintenance ventilation was established to solve the response of sample points,Renault timeaveraged model was used to simulate and analyze the diffusion and migration of harmful gases released from the decay of biologically invasive mussel in the process of maintenance ventilation.The reliability of the numerical simulation model was verified by the ventilation scale model of 1 : 20.Finally,the evaluation index of ventilation effect for long-distance water transmission pipeline maintenance ventilation is put forward,including supply-demand ratio of air volume,the number of ventilation changes and the ventilation cost of long-distance water transmission pipeline.(2)An improved cooperative radial basis function multi-fidelity surrogate model is proposed.This model uses the structural risk minimization(SRM)criterion to consider the advantages of empirical risk and confidence range minimization at the same time.The L2 norm of structural risk is used to describe the structural complexity of CoRBF,and the dif-ferential relationship between CoRBF and model hyper-parameters is derived to optimize the shape factor of radial basis function,so as to make up for the deficiency of traditional parameter optimization methods that are easy to lose the sparse characteristics of the model.Then,the multi-fidelity surrogate model of SRM-CoRBF is established,and the test function is used to verify the superiority of the improved model in prediction accuracy and generalization performance.(3)Combined with a long-distance water transmission pipeline project in southern China,the ventilation effect prediction of long-distance water transmission pipeline maintenance ventilation based on SRM-CoRBF multifidelity surrogate model is studied.The numerical simulation of long-distance water transmission pipeline maintenance ventilation verified by model test is used to solve the response of sample points and the SRM-CoRBF multi-fidelity surrogate model is established to predict the ventilation effect.The results show that the proposed model can realize the highprecision prediction of the supply-demand ratio of air volume,the number of ventilation changes and the ventilation cost of long-distance water transmission pipeline maintenance ventilation.On the one hand,compared with the RBF single-fidelity surrogate model,the prediction accuracy of the proposed model in the supply-demand ratio of air volume,the number of ventilation changes and the ventilation cost is increased by 15.02%,9.29% and 6.03%,respectively.On the other hand,compared with the traditional CoRBF multi-fidelity surrogate model,SRM-CoRBF can effectively improve the generalization per-formance while ensuring the prediction accuracy of the model.