城中村局部拆整区建筑布局多目标优化研究OACHSSCD

Given the trend towards urban regeneration in built-up areas,the transformation of urban villages has gradually shifted from large-scale demolition and reconstruction towards a more refined and incremental selective redevelopment pattern.Through partial demolition,spatial improvement,and functional upgrading,selective redevelopment aims to improve the living environment while preserving the existing spatial fabric and social networks.However,high-density urban villages in the Pearl River Delta are characterized by complex building relationships and narrow streets and alleys.Changes in local building massing,height,and location may affect ventilation,shading,sunlight access,and solar heat gain.Therefore,the coordination of development intensity,spatial order,and environmental performance within a limited scope of intervention has become a critical issue in selective redevelopment. A case study based on the selective redevelopment area of Shipai Village in Guangzhou was carried out to develop a layout-assisted design approach integrating parametric modelling,building performance simulation,artificial intelligence-based prediction,and multi-objective optimization.The plots which can be developed were divided into basic units.By using unit construction status and building height as decision variables,the optimization objectives focused on the cumulative duration of thermal comfort at evaluation points,the cumulative duration of sunlight exposure at evaluation points,and the cumulative solar radiation received by building surfaces.Air temperature and relative humidity at an open reference point within the site were first monitored and compared with synchronous online meteorological data.The two datasets exhibited a high degree of consistency in their diurnal variation trends,indicating that online meteorological data could be used as boundary inputs for environmental simulations.On this basis,field measurements were further conducted to validate the wind speed and sunlight simulations.For the wind speed simulation,the coefficient of determination(R2)was 0.989,the normalized mean bias error(NMBE)was+2.86％,and the coefficient of variation of the root mean square error(CV[RMSE])was 7.18％.For the sunlight simulation,the corresponding values were 0.972,+8.75％,and 16.58％,respectively.This proves that the simulation model can effectively reproduce variations in wind speed and sunlight conditions within complex street-and-alley environments. To address the high computational cost of thermal comfort simulations,a surrogate model based on a back-propagation(BP)neural network was developed,with an R2 of 0.95 and a mean absolute percentage error(MAPE)of 1.61％.It reduced the computation time for a single design scheme by about 88.9％,from approximately six minutes to 40 seconds. Based on the surrogate model,the NSGA-Ⅱ algorithm was employed to conduct the three-objective collaborative optimization,yielding 48 Pareto-optimal solutions.The results indicate that increased building height and more continuous building masses are conducive to shading but may reduce sunlight access and weaken local ventilation.Increasing open space improves ventilation and daylight access but may reduce shading.Although continuous street frontages could be formed by increasing the amount of development,it also may increase the building surface area exposed to solar radiation.Therefore,selective redevelopment should not pursue the optimum of any single indicator but should achieve a balance among building height gradients,open spaces at critical nodes,continuous street interfaces,and landscaped open spaces. Candidate schemes were subsequently selected according to floor area ratio ranking,a minimum green space ratio of 30％,and a minimum street-wall alignment ratio of 40％.The selected schemes were translated into implementable architectural proposals through the consolidation of building masses,the creation of open spaces at key nodes,and the reorganization of street interfaces.In the final scheme,the cumulative duration of thermal comfort reached 64,457 h,corresponding to a thermal comfort duration ratio of 68.03％,which was 50.62 percentage points higher than that under the existing condition.The cumulative duration of sunlight exposure reached 24,021 h,corresponding to a sunlight duration ratio of 50.71％,showing an increase of 18.81 percentage points.The cumulative solar radiation received by an individual building was reduced by 76.97％to 709.12 kWh.Compared with the average performance of the Pareto-optimal solution set,the final scheme maintained comparable or superior performance across the evaluated indicators,demonstrating that the architect-led engineering adjustments prevent the scheme deviation from multi-objective optimization directions. The findings indicate that the proposed method can transform experience-based building layout decisions into a quantifiable and comparable design evaluation process.The selective redevelopment of high-density urban villages should avoid uniformly high-rise development or extensive low-rise coverage.Instead,an integrated balance of environmental performance,spatial quality,development benefits,and implementation feasibility shall be achieved through differentiated building heights,strategic open spaces at key nodes,continuous street walls,and the coordinated layout of green spaces.