【作者】 刘长鑫；

【导师】 柴天佑；

【作者基本信息】 东北大学 ， 控制理论与控制工程， 2008， 硕士

【摘要】 在选矿生产过程中,工艺指标是企业经济效益和生产控制之间的联系纽带。相对于局部工艺指标独立控制,对选矿生产全流程工艺指标进行协同控制能够取得更高的经济效益,且有利于企业节能降耗、提升生产效率、提高产品质量。选矿过程具有强非线性、参数时变、工况变化频繁、关键工艺指标不能在线实时检测等综合复杂性,难以通过建立精确的数学模型实行优化控制,采用的人工操作主观性强、随意性大,难以保证生产运行于优化状态。而目前选矿过程控制系统多以局部工艺过程的优化控制为目标。因此设计与开发一个用于选矿生产全流程工艺指标整体优化决策的决策系统软件平台,为复杂优化算法的研究提供一个研究和验证环境,具有重要的研究意义和实用价值。针对这些问题,结合某赤铁矿选矿厂具体背景,在东北大学“985工程”流程工业综合自动化科技创新平台的支持下,本文在总结现有工艺指标决策方法的基础上,设计和开发了面向Web的选矿生产全流程工艺指标决策系统软件平台,具体实验结果表明该软件平台功能可用、有效,人机交互界面友好、易用,可以帮助研究者在复杂工业过程实际环境下研究验证各种决策算法。本文具体研究主要包括:1)设计了面向Web的选矿生产全流程工艺指标决策系统软件平台。平台中工艺指标决策策略由工艺指标预设定模块、综合生产指标预报模块、反馈分析调整模块、预报分析调整模块、过程数据统计分析模块组成。各模块决策算法具有可替换性且其参数允许静态动态配置;算法运行的动态指示与工艺指标实时监控、预警分析处理增加了工艺指标决策的科学性,同时更好的向研究者揭示指标之间潜在的内部关系;通过设计独立于异构数据源的统一数据表达机制,提供灵活、易用、可扩展的接口,实现各算法间有效的数据交互,对研究者屏蔽不同的指标信息来源细节,使其可以集中精力在工艺指标决策算法的研究上。2)综合利用多种先进技术,开发了具有丰富有好可视化功能的面向web的工艺指标决策系统软件平台。包括采用Flash/Javascript实现的决策策略执行过程动态可视化指示,辅助研究者了解决策策略执行中的关联影响;采用Ajax实现无刷新界面的多种视图下指标数据图形联动的统计分析功能;采用Web Service和XML实现开放式决策算法库,允许研究者利用平台现有,包括其它研究者的算法,设计添加自己新的决策算法,实现研究工作的可持续性,而且有效利用了分布式计算资源,实现了算法的高速运行;采用DIV/CSS/XST开发的友好人机交互界面,在用户角色控制下,提供了研究者在任何时间以任何方式接入平台研究的途径,并记录其操作,为研究者提捕捉瞬间灵感提供了有力支持;开发的计划调度层、控制系统层接口及工艺过程指标采集与存储设计,为研究者提供了复杂工业实际环境下的决策算法整体研究验证和测试平台。3)具体的验证测试实验表明了工艺指标决策算法可替换,软件平台功能可用、有效,人机交互界面友好、易用,可以帮助研究者在复杂工业过程实际环境下研究验证各种优化决策算法。更多还原

【Abstract】 The technical indices of ore-dressing process link the economic efficiency of the enterprise to the process control. Comparing to the process control based on local technical indices, the process control based on coordination technical indices can achieve higher economic benefits, better efficiency and quality, and less consumption in energy and material. It’s difficult to build a mathematical model of the ore-dressing process and carry out optimized control because of its complexity, such as strong nonlinearity, time-varying, frequently changing operation conditions, being difficult to measure on-line the key technical indices and so on. Nonetheless, manual control of the process cannot guarantee the continuity and the stability of the production due to operators’ subjectivity and non-preciseness. Most of the current control systems set local optimization as their targets. Therefore it is significant to design and develop a decision-making platform system for deciding the technical indices of the processes targeting the optimization of the overall ore-dressing process. The platform provides an environment for the research and validation of the decision-making algorithms of the complex industrial processes.In order to solve these problems, the existing algorithms are summarized in this thesis, and then a web-oriented software platform of decision-making system for technical indices of overall ore-dressing process is designed and developed based on a hematite ore-dressing plant. The thesis is supposed by Northeastern University "985 Project" Process Industry Integrated Automation Technological Creativity Platform. The experiments prove that the algorithms are replaceable and the platform is functional and efficient. The platform can be used by the researchers to validate their algorithms under the complex industrial process circumstance through its user-friendly and easy to use human computer interface.The details are described as follows:1) This thesis introduces the design of the web-oriented software platform of decision-making system for the technical indices of overall ore-dressing process. The strategy used in the system is composed of the technical indices presetting module, the global production indices prediction module, the feedback analysis and adjustment module, the prediction analysis and adjustment module and the process data statistics analysis module. The algorithm used in each module is replaceable, and the static or dynamic parameter configuration is allowed. The technical indices decision-making process is more scientific by methods of dynamic instructions of algorithm running state, real-time monitoring of the technical indices, early warning analysis and adjustment. At the same time the inner relationships among the indices are illustrated. The design of the unified data expression mechanism which is independent of the heterogeneous data sources provides a flexible, usable and expandable interface. It also realizes an effective mean of data interaction between algorithms which masks the data and the details of different indices sources to help the researchers to concentrate on the decision-making algorithm research.2) This thesis develops the user-friendly web-oriented software platform of decision-making system for technical indices of overall ore-dressing process through a variety of technologies. The Flash and JavaScript are used to dynamically visualize the decision-making strategy running state. They are used to notify the researchers the relative influence during the decision-making operation. The statistics analysis with multi-view of the technical indices is realized by Ajax technology, which can provide non-refreshing user interface. Web service and XML technologies are used to realize the open decision-making algorithm library. This is helpful for keeping the relevant research actives coherent and improving the utilization of the distributed computing resources. The technologies of DIV/CSS/XST are used to develop user-friendly human computer interface that provides a way for researcher to access to the platform at any time in any way. The operation records provide powerful support to capture researcher’s instantaneous inspiration. Several interfaces are developed, including the interface of the planning and scheduling, the interface of the control system and the interface of the technical indices acquisition. These interfaces provide an environment for the research and validation of the decision-making algorithms under the complex industrial process circumstance.3) The experiments prove that the decision-making algorithms of technical indices are replaceable, and the platform is functional and efficient. The conclusion is that the human computer interface is user-friendly and easy to use. The platform can be used by researchers to validate their algorithms effectively under the complex industrial process circumstance.更多还原