【作者】 张振；

【导师】 柯庆镝； 王玲；

【作者基本信息】 合肥工业大学 ， 机械工程（专业学位）， 2025， 硕士

【摘要】 本文针对机电产品全生命周期绿色设计中知识服务不足的问题,展开了全生命周期下用户多维特征分析及其绿色设计知识推荐方法的深入研究,旨在为设计人员提供精准、高效的绿色设计知识推荐。研究的主要内容与成果如下:(1)提出面向生命周期绿色设计的多维用户画像构建模型方法。首先,根据全生命周期绿色设计概念,结合物元理论建立产品-操作-效果-效果影响矩阵知识表达框架,细化产品层级、操作流程及环境影响指标,实现知识量化关联。然后,通过分析研发人员在全生命周期绿色设计过程中的数据信息,构建包含了职位、年龄、主要职责等基本信息的用户基础特征;设计知识、操作记录、时间信息、使用工具行为数据的用户行为特征;根据建立知识表达框架从产品元素、操作元素和效果元素等方面构建用户绿色设计特征;生命周期特征进一步量化了用户在原材料获取、制造与装配、运输、使用和回收处理等阶段的影响,全面反映了用户在全生命周期绿色设计中的需求和行为模式。(2)提出了基于多维用户画像模型的绿色设计知识推荐方法。首先,针对新用户冷启动问题,设计了基于迁移学习的用户画像匹配与特征迁移策略,基于用户基础特征进行相似度计算,将用户画像的特征简化并使用卡方检验进行特征迁移。然后,通过用户行为特征中的知识操作、时间衰减以及使用工具函数计算相似度,用户绿色设计特征中的产品约束条件、操作发散以及效果关联函数计算相似度以及用户生命周期特征中矩阵以及向量计算相似度。最后,根据用户对知识的重视程度和启发程度进行排序和推送并通过用户反馈不断调整权重系数,优化推荐的绿色设计知识,提升用户满意度和推荐效果。(3)以2.5MW风力发电机为案例,详细构建了用户画像,实现了新用户的用户画像迁移,并对绿色设计知识进行了推送。首先对风力发电机的组成、工作原理以及失效部件进行了深入分析,确定了齿轮箱等关键部件的绿色设计需求。然后,构建了多个用户的画像,通过基础特征相似度计算和特征迁移,完成了新用户的用户画像构建,根据用户画像的行为特征、绿色设计特征以及生命周期特征的相似度计算进行了用户画像的匹配。最后,根据用户与绿色设计知识关联匹配,为用户推送了相关的绿色设计知识,验证了推荐系统的有效性。通过该案例,证明了所提方法在实际应用中的可行性和优越性,为风力发电机的绿色设计提供了有力支持。更多还原

【Abstract】 This paper presents an in-depth study on multi-dimensional user portrait modeling and personalized recommendation methods for life cycle green design,aiming to solve the problem of insufficient knowledge service in the green design of mechanical and electrical products and provide accurate and efficient green design knowledge recommendations for designers.The main research contents and results are as follows:(1)Combined with knowledge representation methods and life cycle phase division,the four elements of green design features are extracted according to the characteristics of green design knowledge,and a product-operation-effect-effect life cycle matrix knowledge representation framework is constructed.By analyzing the data of R&D personnel in the green design process across the entire life cycle,a multi-dimensional user profile model is built,including user basic features,behavioral features,green design features,and life cycle features.User basic features cover basic information such as position,age,and main responsibilities.User behavioral features collect data on operations like searching,browsing,downloading,and collecting green design knowledge,as well as tools used daily,contributing to mining user interests and preferences.Green design features depict users’expertise in this field from product,operation,and effect elements.Life cycle features further quantify users’green design features across life cycle data.(2)To address the cold start problem for new users,a user profile matching and feature transfer strategy based on transfer learning is designed.By calculating the similarity of user behavioral,green design,and life cycle features,precise knowledge recommendation is achieved.A feedback mechanism is introduced to optimize recommendation weights,enhancing the real-time and adaptability of knowledge services.During recommendation,the similarity of user behavioral,green design,and life cycle features is comprehensively considered to ensure comprehensive and accurate results.Meanwhile,knowledge is ranked and pushed based on users’emphasis and inspiration levels,with weights continuously adjusted via feedback to optimize recommendations,improving user satisfaction and recommendation effectiveness.(3)A case study of a 2.5MW wind turbine is used to construct user profiles,implement user profile transfer for new users,and push green design knowledge.First,an in-depth analysis of the wind turbine’s composition,working principles,and failed components identified key components such as the gearbox and their green design requirements.Then,multiple user profiles were constructed,and new user profiles were completed through basic characteristic similarity calculations and feature transfer.User profile matching was performed based on behavioral characteristics,green design characteristics,and life cycle characteristics.Finally,relevant green design knowledge was pushed to users based on associations between user profiles and green design knowledge,validating the effectiveness of the recommendation system.This case demonstrates the feasibility and superiority of the proposed method in practical applications,providing strong support for the green design of wind turbines.更多还原