| The rapid development of urbanization has led to an increase in impervious area,the number of people living in the city,and the extreme rainfall events,resulting in urban waterlogging and overflow due to insufficient drainage capacity of the drainage network in the city.In the face of these problems,scholars often use rainfall models to model the study area and simulate the effect of low impact development facilities to mitigate urban flooding and overflow disasters.How to quickly and accurately determine the model and optimize the layout of facilities has become a hot topic of research.In this paper,the SWMM model of Xi’an University of Technology Jinhua campus is used as the study area,and the parameters of the SWMM model are combined with BP neural network algorithm to make the model rate more efficient.The SWMM model is coupled with the NSGA-Ⅲ algorithm to build an optimization model for LID facility layout with the optimization objectives of maximum overflow reduction rate,minimum whole-life cost and minimum footprint,so as to realize the optimization of LID facility layout scheme.The main conclusions obtained based on the study results are as follows:(1)The initial SWMM model of the study area is established,and the parameters to be rate set and the reasonable value interval are determined by referring to the research results of related scholars and the SWMM user manual.Using the Latin hypercube sampling method,1000 sets of parameters to be rate-setting were extracted,and the dynamic database of SWMM was called through MATLAB to realize the replacement,updating and simulation of parameters in the model file,and the extraction of node water depth simulation results,so as to establish the BP neural network model between the parameters to be rate-setting and simulation results,and the R values of training,validation,testing and the whole stage were all greater than 0.93,indicating that the established BP neural network works well.The measured node water depth data are input and the BP neural network inversion calculation is carried out to obtain the values of the parameters to be rate-determined.The model parameters were then validated by two measured rainfall cases with measured water depth data in two nodes,and the results showed that the Nash coefficients were 0.86,0.82,0.81 and 0.82,which were all greater than 0.8,indicating that it is reasonable to rate the model parameters based on the BP neural network.(2)With the Chicago rainfall type as the design rainfall type,the flow rate,overflow rate,surface runoff,nodal overflow and pipeline overload in the study area were simulated for rainfall return periods of 1a,3a,5a,10a and 20a,respectively.The results show that with the increase of rainfall recurrence period,the overflow,runoff,nodal overflow rate and pipeline overload rate increase significantly,and the simulated values are 3243m3,49.783mm,73.6%and 86.8%at the rainfall recurrence period of 20a.The flow process line at Outfall 1 was higher than the other two outfalls.The simulation analysis of the new bioretention zone and water storage module in the study area found that these two combinations of facilities can effectively reduce the road runoff,and the reduction rate of runoff from Mingli Road can reach more than 80%under different rainfall return periods,and the reduction rate of peak runoff is also more than 74%,in which the storage module can store a maximum of 8.57 m3,36.68 m3,80.38 m3,and 40.68 m3 under five rainfall return periods,The maximum storage capacity of the water storage module under the five rainfall recurrence periods is 8.57m3,36.68m3,80.38m3 and 97.79m3.(3)Combining the soil texture,full life-cycle cost and suitable sites of each LID facility in the study area,rain gardens,depressed green areas,green roofs and permeable pavers were selected as the types of LID facilities to be laid out,and the location of each facility was determined,and a total of 25 LID facilities were laid out.With the objectives of maximum overflow reduction rate,minimum life-cycle cost and minimum footprint,the layout of LID facilities is optimized by coupling the SWMM model with the NSGA-Ⅲ algorithm to obtain Pareto solutions for different rainfall return periods.The Pareto solution set under the rainfall return period of 3a is used as an example,and the TOPSIS comprehensive evaluation method is applied to evaluate each optimization scheme,and the optimal LID facility layout scheme is finally obtained.The rain garden under this optimal scheme covers an area of 4854.55 m2,the recessed green space covers an area of 4565.27 m2,the green roof area is 4237.49 m2,the permeable pavement deployment area is 1902.26 m2,the reduction rate of the overflow is 78.01%,the whole life cycle cost is 6.95 million yuan,and the land area is 1.56ha. |
PDF Full Text Request