考虑竞拍者选择行为的鲜活品拍卖收益优化策略和方法研究

Facing the complex and dynamic perishables auction revenue optimization requirements, regarding the problem that current fixed auction combination cannot adapt to dynamic demand and choice behavior changes to achieve maximized revenue, this project proposes a new paradigm of flexible auction combination revenue optimization considering the choice behavior of bidders through refining three key research contents, namely auction demand forecasting model, bidders’ choice behavior model and auction combination control optimization model. Firstly, based on the identification of key factors that influence auction demand, multiple regression models are adopted to predict auction demand. Then it analyzes the characteristics of the choice behavior of the strategic bidders and builds the discrete choice model to calculate the choice probability. Taking auction demand prediction and choice probability as inputs, a multi-family, multi-class and multi-stage auction combination control optimization model is being built using dynamic planning method. Related algorithm is also designed. Optimal auction combination configuration is thus achieved in terms of different demand distribution and choice behavior modes. Finally, the theoretical mechanisms of this project are integrated and verified through the perishables auction combination revenue optimization system. This study accurately reveals the quantitative relation between flexible control of auction combination and revenue levels, transforms the traditional and fixed arrangement of auction sequence/quantity, as well as solves the problem of large quantities of trading with optimal revenue and efficient operations simultaneously for a perishable goods auction center. This research will lay a theoretical foundation for auction trading mechanism innovation of multi-category perishable goods.

面向复杂和动态的鲜活品拍卖收益优化需求，针对当前固化参拍组合无法根据动态需求和选择行为实现收益最大化的问题，提出考虑选择行为的柔性参拍组合收益优化新模式，提炼拍卖需求预测模型、竞拍者选择行为模型和参拍组合控制优化模型三个关键研究内容。首先，在识别关键因素对竞拍需求影响的基础上，采用多元回归模型预测竞拍需求；进而分析策略竞拍者选择行为的特征，构建离散选择模型计算选择概率；以预测需求值和选择概率为输入，基于动态规划方法构建多品类多等级多阶段参拍组合控制优化模型并设计求解算法，得到不同需求分布和选择行为模式下的最优参拍组合配置；最后，通过鲜活品参拍组合收益优化系统将本项目理论成果进行整合和验证。本课题精确揭示参拍组合柔性控制对收益水平的量化关系，转变以固定次序安排参拍的传统模式，解决鲜活品拍卖中心大批量交易的同时实现收益优先并流畅运转的难题，为多品类鲜活品拍卖交易机制创新提供理论支撑。

面向复杂动态的拍卖运营与收益联合优化需求，提炼覆盖鲜切花、电商货品及具有易逝品属性的物流服务的系统化理论模型、机制、算法，将模型算法与本项目自研的鲜活品拍卖智能系统相嵌融合，分别在实验室搭建的仿真环境和实际场景（例如昆明鲜花拍卖中心、香港电商海外仓）进行小规模、大规模算例的机制验证、实验分析和结果比较，取得与本项目相关的如下重要成果：（1）考虑需求预测的柔性参拍组合控制优化模型，以收益最大化为目标建立拍卖笼车分配模型，验证了拍卖收益优化方法及基于鲁棒模型所获得的收益平均高于传统先到先得模式20%-34%；（2）考虑竞拍者博弈与信号传递的序贯拍卖批量优化与收益策略，凝练竞拍人数随机到达和多阶段信息传递等核心特点，针对静态博弈构建多阶段拍卖平台均衡收益模型；（3）数据驱动的拍卖中心存储位置优化方法；（4）基于需求预测的拍卖中心存储分配机制，以上两种方法与传统的存储位置策略相比，减少了总行程距离；（5）基于预测和延迟的拍卖中心订单分批方法，该方法考虑了时间间隔和买家完成率，平均缩短了订单处理时间(淡季76.69 min，旺季195.15 min)和系统响应时间(淡季65.90 min，旺季108.29 min)；（6）电商拍卖中心两阶段跨区拣选与分拣同步机制；（7）基于CPS的电商拍卖物流系统与案例研究，通过案例与实验验证了以上所提方法的有效性并考虑不同的订单履行场景如淡季/旺季进行了绩效权衡分析；（8）基于CPS的按需物流在线双边拍卖机制设计，结果表明当市场供求不平衡时，与固定定价机制相比，设计的拍卖机制更适合采用；（9）考虑实时转运的按需物流拍卖方法，设计了一种两阶段组合拍卖机制来实现动态按需任务(重)分配。

内容获取失败，请点击重试