目的 针对飞机货舱配载方案评估中多属性决策的复杂性,以及现有评估方法在权重确定上过度依赖专家经验导致的主观偏差,或单纯依赖客观数据忽略决策者偏好的局限性,提出一种融合主观先验与客观数据驱动的混合赋权评估模型,以提供更为合理、可靠的配载方案择优决策支持。方法 首先,引入大型语言模型（Large language model,LLM）,构建“虚拟专家委员会”,通过精心设计的提示词工程,获取多维度、多情境下的主观权重。其次,针对传统熵权法对数据分布敏感、难以有效区分指标优劣等问题,提出一种改进的数据预处理熵权法（Improved data preprocessing entropy weighting method,IDPEW）,该方法结合指标值的辨识度和信息熵的均衡性来确定客观权重。最后,将LLM生成的主观权重与IDPEW计算的客观权重进行加权组合,构建综合评价函数,对飞机货舱配载方案进行全面评估和排序。结果 实验结果表明,LLM模拟专家意见时最关注“装载率”（主观权重0.225 0）,而IDPEW方法从数据中识别出“横向不平衡度”最具区分力（客观权重0.248 1）。混合赋权模型（α=0.5）有效平衡了主客观偏好,在24个方案中精准识别出综合性能最优的方案,验证了模型在复杂情境下的稳定性。结论 创新性地利用LLM低成本构建“虚拟专家”获取先验知识,并通过耦合指标辨识度与均衡性的IDPEW方法,提升了客观赋权精度。该模型克服了单一赋权的局限,为飞机货舱配载方案的科学评估提供了一种兼具可解释性和实用性的新范式。
Abstract
The work aims to propose a hybrid evaluation model fusing subjective priors with objective data insights to provide rational decision support to address the complexity of Multi-Attribute Decision Making (MADM) in aircraft cargo hold loading evaluation and the limitations of existing weighting methods, specifically the subjective bias from excessive reliance on expert experience and the neglect of decision-maker preferences in purely data-driven approaches. Firstly, Large Language Models (LLMs) were employed to construct a "virtual expert committee," utilizing prompt engineering to extract subjective weights across diverse dimensions and scenarios. Secondly, to mitigate the sensitivity of the traditional Entropy Weight Method (EWM) to data distribution, an Improved Data Preprocessing Entropy Weighting method (IDPEW) was introduced, determining objective weights by coupling indicator discernibility with entropy equilibrium. Finally, a comprehensive evaluation function was constructed by integrating subjective and objective weights to rank loading schemes. Experimental results indicates that the LLM-simulated committee subjectively prioritizes "Load Factor" (subjective weight 0.225 0), whereas IDPEW objectively identified "Lateral Imbalance" (subjective weight 0.248 1) as having the highest discernibility. The hybrid model (α=0.5) effectively balanced these preferences, accurately identifying the optimal scheme among 24 candidates and demonstrating robustness in complex scenarios. In conclusion, this study innovatively leverages LLMs for the low-cost acquisition of expert priors and enhances objective weighting precision via IDPEW. By overcoming the limitations of single-weighting approaches, the proposed model offers a novel paradigm featuring both interpretability and practicality for the scientific evaluation of aircraft cargo hold loading.
关键词
飞机货舱配载 /
多属性决策 /
大型语言模型 /
主观赋权 /
熵权法 /
混合赋权
Key words
aircraft cargo hold loading /
multi-attribute decision making /
large language model /
subjective weighting /
entropy weight method /
hybrid weighting
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] HWANG C L, YOON K.Methods for Multiple Attribute Decision Making[M]// Multiple Attribute Decision Making. Berlin: Springer Berlin Heidelberg, 1981: 58-191.
[2] SAATY T L.Decision Making with the Analytic Hierarchy Process[J]. International Journal of Services Sciences, 2008, 1(1): 83.
[3] J E J, LINSTONE H A, TUROFF M. The Delphi Method: Techniques and Applications[J]. Technometrics, 1976, 18(3): 363.
[4] SHANNON C E.A Mathematical Theory of Communication[J]. The Bell System Technical Journal, 1948, 27(3): 379-423.
[5] DIAKOULAKI D, MAVROTAS G, PAPAYANNAKIS L.Determining Objective Weights in Multiple Criteria Problems: The Critic Method[J]. Computers & Operations Research, 1995, 22(7): 763-770.
[6] YUE C, HUANG R B, TOWEY D, et al.An Entropy-Based Group Decision-Making Approach for Software Quality Evaluation[J]. Expert Systems with Applications, 2024, 238: 121979.
[7] 吴尚智, 任艺璇, 葛舒悦, 等. 用Critic赋权法加权邻域粗糙集的属性约简算法[J]. 北京航空航天大学学报, 2025, 51(1): 75-84.
WU S Z, REN Y X, GE S Y, et al.An Attribute Reduction Algorithm of Weighting Neighborhood Rough Sets with Critic Method[J]. Journal of Beijing University of Aeronautics and Astronautics, 2025, 51(1): 75-84.
[8] 徐泽水, 达庆利. 多属性决策的组合赋权方法研究[J]. 中国管理科学, 2002, 10(2): 84-87.
XU Z S, DA Q L.Study on Method of Combination Weighting[J]. Chinese Journal of Management Science, 2002, 10(2): 84-87.
[9] 张洋, 丘东元, 张波, 等. 基于层次分析-熵值法的DC-DC变换器拓扑综合评价[J]. 北京航空航天大学学报, 2025, 51(5): 1469-1479.
ZHANG Y, QIU D Y, ZHANG B, et al.Comprehensive Evaluation of DC-DC Converters Based on Analytic Hierarchy Process and Entropy Method[J]. Journal of Beijing University of Aeronautics and Astronautics, 2025, 51(5): 1469-1479.
[10] 孙学安, 王寅, 周齐贤. 基于综合赋权的目标可攻击价值综合评估排序[J]. 北京航空航天大学学报, 2024, 50(5): 1731-1737.
SUN X A, WANG Y, ZHOU Q X.A Comprehensive Air-Ground Target Attackability Value Ranking Based on Comprehensive Weighting[J]. Journal of Beijing University of Aeronautics and Astronautics, 2024, 50(5): 1731-1737.
[11] YANNIS G, KOPSACHEILI A, DRAGOMANOVITS A, et al.State-of-the-Art Review on Multi-Criteria Decision-Making in the Transport Sector[J]. Journal of Traffic and Transportation Engineering (English Edition), 2020, 7(4): 413-431.
[12] BROWN T, MANN B, RYDER N, et al.Language Models are Few-Shot Learners[J]. Advances in Neural Information Processing Systems, 2020, 33: 1877-1901.
[13] TEAM G, ANIL R, BORGEAUD S, et al. Gemini: A Family of Highly Capable Multimodal Models[EB/OL]. arXiv, 2023. [2025-12-01]. https://arxiv.org/abs/2312.11805.
[14] SVOBODA I, LANDE D. Enhancing Multi-Criteria Decision Analysis with AI: Integrating Analytic Hierarchy Process and GPT-4 for Automated Decision Support[EB/OL]. arXiv, 2024. [2025-12-01]. https://arxiv.org/abs/2402.07404.
[15] SVOBODA I, LANDE D. AI Agents in Multi-Criteria Decision Analysis: Automating the Analytic Hierarchy Process with Large Language Models[EB/OL]. SSRN, 2024. [2025-12-01]. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5069656.
[16] VANCROONENBURG W, VERSTICHEL J, TAVERNIER K, et al.Automatic Air Cargo Selection and Weight Balancing: A Mixed Integer Programming Approach[J]. Transportation Research Part E: Logistics and Transportation Review, 2014, 65: 70-83.
[17] 赵向领, 左蕾. 货运飞机装箱与配载组合优化[J]. 航空动力学报, 2024, 39(11): 482-492.
ZHAO X L, ZUO L.Optimization of Cargo Aircraft Packing and Stowage Combination[J]. Journal of Aerospace Power, 2024, 39(11): 482-492.
基金
中央高校基本科研业务费-高水平成果培育项目（3122025TD08）; 中国交通教育研究会教育科学研究重点课题（JT2024ZD066）
PDF(2368 KB)
渝公网安备 50010702501716号 渝ICP备15012534号-2
