安全与环境学报

现代工程项目多目标的协同是项目管控的难点，基于项目的多目标协同优化是解决这一难点的关键。研究探讨了在碳减排目标下的项目工期、成本、质量、安全多目标协同关系和协同优化机制。此外，基于多目标优化（Multi-Objective Optimization Problem，MOP）理论开展系统研究，通过纳入建造阶段碳排放指标，建立了涵盖工期–成本–质量–安全–碳排放的五维度多目标优化模型，并基于改进非支配排序遗传算法（Non-dominated Sorting Genetic Algorithm Ⅱ，NSGA–Ⅱ）算法通过实证研究，验证了所提模型的有效性及算法的可行性。结果显示，NSGA–Ⅱ在求解该规模多目标优化问题时具备显著优势，通过对NSGA–Ⅱ求解得到的Pareto解集进行客观量化分析，可实现最优方案的科学筛选。研究不仅为项目多目标协同管控提供了应用工具，还在多目标问题构建、求解及算法选择方面提供参考。

This study develops a five-objective optimization framework for the construction phase of building projects under carbon emission constraints. The framework integrates project duration， cost， quality， safety， and carbon emissions into a unified multi-objective model， treating construction activities and their alternative execution modes as decision units. The model quantifies the attributes of each candidate mode and transforms the actual construction organization problem into a computable combinatorial optimization problem by incorporating precedence relationships， exclusive mode-selection constraints， schedule limits， cost bounds， and quality and safety thresholds. In the model structure， quality and safety are treated as constraint-type objectives to ensure minimum engineering and risk-control requirements， while duration， cost， and carbon emissions are regarded as core optimization objectives for trade-off analysis. To solve the proposed problem， an Non-dominated Sorting Genetic Algorithm II based optimization procedure is employed to generate Pareto non-dominated solutions. An empirical analysis is conducted using a public building project in Xi’an， China. The project is decomposed into 17 key construction activities， each associated with two or three feasible execution modes， resulting in 17 006 112 candidate construction schemes. The parameters for duration， cost， quality， carbon emissions， and safety for each activity are determined through schedule analysis， quantity takeoff， pricing software， carbon emission factor calculation， and safety-cost-based risk evaluation. The proposed solution method is further compared with the Genetic Algorithm (GA) algorithm under identical parameter settings using Hypervolume (HV)， Inverted Generational Distance (IGD)， and spacing indicators. The results demonstrate that the proposed NSGA–Ⅱ–based approach achieves superior convergence， distribution uniformity， and solution stability compared to the GA in solving the five-objective problem. The obtained Pareto front effectively illustrates the nonlinear trade-offs among duration， cost， quality， safety， and carbon emissions， providing differentiated decision support for scenarios prioritizing time， cost， low carbon emissions， or balanced solutions. This study shows that the proposed model and solution framework can facilitate the scientific selection of construction schemes in complex multi-constraint environments.