城市生活垃圾规划管理智能决策支持系统

【作者】 李劲；

【导师】 王华；

【作者基本信息】 昆明理工大学 ， 生产过程物流学， 2008， 博士

【摘要】 城市生活垃圾治理是城市社会经济发展总体规划中的一项重要内容,对垃圾进行科学的治理,对于保护城市环境,促进区域社会经济的可持续发展有着十分重要的作用。但目前,在实际管理过程中,主观随意性太强,缺乏一定的科学决策,片面追求形式而忽视质量,注重局部治理,而没有从全局出发,综合治理,从而使垃圾管理流于形式。另外,在管理的技术方法上,虽然计算机的应用已逐步代替了传统的手工方法,但管理中的决策过程仍以人工为主,且信息管理主要限于属性数据,而没有涉及到空间布局的决策过程。因此,本论文针对当前城市生活垃圾治理中存在的问题,以建立相应的辅助决策模型为核心,将其与人工智能技术、GIS技术相结合,建立城市生活垃圾规划管理智能决策系统,为科学合理、快速高效地规划管理城市生活垃圾提供技术支持。智能决策系统(简称IDSS)是将人工智能技术引入到决策支持系统中形成的一种新型信息系统,主要解决半结构、非结构化的决策问题。本文对MSWIDSS (Municipal Solid Waste Intelligence Decision Support System)的原理、模型、实现方法等关键技术作了探索性研究,提出了“城市生活垃圾规划管理智能决策方法”,并结合应用实际,突出研究了决策模型体系的组织和模型库系统的构建、智能模型与可视化模型的建立及其实现方法。首先,本文在查阅大量国内外文献的基础上,对城市生活垃圾治理研究做了全面的综述。从城市生活垃圾治理业务的实际出发,定义了城市生活垃圾治理系统的概念及内涵。以城市生活垃圾治理可持续发展的角度,指出了对城市生活垃圾治理系统实施规划管理的必要性,并论述了其原理。在此基础上,进一步提出了城市生活垃圾规划管理的关键问题就是垃圾处理处置设施的定量与定位以及系统规划,由此构建了垃圾治理决策模型库的模型组织体系。针对昆明市的实际和特点,通过决策模型库、数据库以及专家知识库的协作进行多层次决策,首次设计并实现了一个较完善的昆明城市生活垃圾规划管理智能决策系统。其次,从城市生活垃圾规划管理的实际应用角度出发,将传统的数学建模与人工智能建模技术相结合,通过垃圾处置设施选址的多准则专家决策支持模型、垃圾产生量及组分的网络预测模型的建立,阐明了如何应用智能模型解决城市生活垃圾治理系统中不确定的、非结构化以及难于精确数学建模的问题,使得常规应用系统具有了智能决策的功能。第三,在传统BP网络算法基础上,引入最优化方法中的Levenberg-Marquardt算法,开发了基于LMBP算法(BP神经网络Levenberg-Marquardt算法)的城市生活垃圾网络预测模型,该模型性能稳定,收敛速度快。与当前常用的城市生活垃圾预测模型(如：灰色预测模型、人口增长法)相比,该模型预测精度最高。第四,针对复杂的垃圾处置设施选址需要借助大量领域专家知识的特点,以垃圾焚烧厂选址为例,首次建立了“多属性加权评估决策准则表”。该表内容详细且客观,是将专家知识转换成系统的知识数据库,并结合数学规划模型、多属性加权法进行垃圾焚烧厂选址的基础。第五,以昆明市为例,通过实地详细的调研,并结合昆明市的实际情况和垃圾治理现状,首次用规划管理决策模型对昆明市生活垃圾治理系统进行了较全面的优化,为昆明市政府决策者进行规划提供辅助决策依据。第六,在决策模型构建的基础上,将其与GIS进行高效集成。以ARCGIS为基础平台,以VS.NET为开发工具,采用类库开发方式,实现了数据库管理、图形数据和基础数据的可视化查询、城市生活垃圾处理处置设施数量确定、城市生活垃圾处理处置设施空间定位、系统规划、图表输出等功能。系统由数据库、模型库及相应的管理系统和人机交互界面等部分构成,采用了模型驱动的运行方式,并通过较为完备的模型接口函数实现模型与数据之间的连接,系统界面友好,操作简单,灵活性强。总之,根据MSWIDSS的功能、目标、技术要求、体系结构及设计原则,分别从决策模型组织体系、数据库系统、模型库系统、人工智能建模以及GIS组件技术等方面,阐述了MSWIDSS系统的实现机制。最后,设计并实现了面向城市生活垃圾的MSWIDSS应用系统。更多还原

【Abstract】 Municipal solid waste (MSW) management is an important content of social economy development general planning. Scientific and reasonable MSW management is very important for protecting city environment and promoting sustainable development of social economy. However, now the factual management is always lack of scientific decision-making and objectivity, pursuing the form one-sidedly while neglecting the quality, paying attention to local management while neglecting comprehensive management due to lacking whole conception, which leads to MSW management in form. In addition, in technical method of management, although application of computer substitutes for conventional manual method step by step, the decision-making of management is still manual work and that most of information management are restricted to attribute data, not involving in decision process of spatial distribution. So aiming at the problems of present MSW management, the dissertation sets up a decision support system for MSW planning and management by forming relevant aid decision-making models and integrating with artificial intelligence and GIS to realize decision visualization and intelligentizing, which can provide technique support for MSW planning and management scientifically, reasonably and efficiently.Intelligent decision support system (IDSS), a latest-model information system that introduces artificial intelligent technology into decision support system, concentrates on semi-structural and non-structural matter. Based on the groping research of the principle, the model and the realization method of MSWIDSS(Municipal Solid Waste Intelligence Decision Support System), the dissertation presents an intelligent decision-making method for municipal solid waste planning and management and studies the organization of decision-making model system the design for model base system, the establishment of intelligent model and visualization model outstandingly combining with practice.Firstly, after consulting many domestic and foreign literatures, the studies of MSW management were comprehensively summarized in this dissertation. Starting with the practice of MSW management operation, the dissertation defines the concept and meaning of MSW management system. And then according to the sustainable development of MSW management, the necessity of MSW planning and management is pointed out and its principle is dissertated. On the basis of these, the dissertation put more forward the key of MSW planning and management is the quantity determination and spatial distribution of MSW treatment facilities as well as system planning from which an organizing system of MSW decision-making models is set up. Aiming at the practice and characteristic of Kunming city, by means of the mutual collaboration among model-base, data-base and expert knowledge-base, multilayer decision-making is come true and a comparatively perfect MSWIDSS is designed and realized with visual and intelligent characteristics for the first time.Secondly, starting with the practical of MSW planning and management, combined traditional mathematics modeling method with artificial intelligence, the dissertation elucidates how to solve the uncertain, and the non-construct problem and the problem of the accurate model which is hardly built by means of mathematical formulation, through multi-attributes expert decision model of MSW facilities site selection and artificial network model of MSW amount and ingredients forecast. Thus, it is possible that the application system possesses the function of intelligent decision-making.Thirdly, in traditional BP neural network modeling, by introducing Levenberg-Marquardt algorithm, a MSW prediction model based on LMBP algorithm is developed. The model has stable performance and speedy convergence ability. Compared with the MSW prediction models in common use (e.g. grey-prediction model, MSW prediction method based on population growth), LMBP model has upmost prediction precision.Fourthly, aiming at that the complicated waste disposal facilities site selection need a great deal of expert knowledge, take a waste incinerator siting for example, multi-attribute utility evaluation decision criteria table is set up for the first time. The table with detailed and objective content is an important base by which expert knowledge can be transformed into systemic knowledge-database, and waste incinerator siting can be realized combining with mathematics planning model and multi-attribute utility evaluation method.Fifthly, take kunming city for example, on the basis of detailed research, MSW management system of kunming city is optimized roundly using planning and management model for the first time. It’s result can serve decision-making of kunming municipality. Sixth, based on the decision-making models and integrating with GIS efficiently. The software platform of this system is ARCGIS and the development tool is VS.NET. Class libraries development method has been adopted and the functions include data base management, visualization inquiring of basic data and graph data, quantity determination of MSW disposal facilities, spatial distribution of of MSW disposal facilities, system planning and outputs of graphic chart have been realized. The system consists of spatial database, attribute database, model base, their corresponding management system and human-computer interactive interface, with the operating mode of models driving. The connection of models and data is realized by the perfect interface functions. The human-computer interactive interface provides information by Figure and data, while decision maker can set and modify model parameters by menu and dialog box. The system has friendly interface and with simplicity of operation and good adaptability.In conclusion, on the basis of the function, goal, technology requirement, system framework and design principle, the realization mechanism of MSWIDSS’s system integrating from the organization system of decision-making models, data-base system, model-base system, artificial intelligent modeling and GIS module technique is expounded. Finally, an application system aimed to MSW planning and management is designed and realized.更多还原