【作者】 刘胜荣；

【导师】 于军琪；

【作者基本信息】 西安建筑科技大学 ， 控制理论与控制工程， 2007， 硕士

【摘要】 环境空气质量计算机监测系统采集的数据量一般是非常巨大的，针对这些海量的数据，如何进行空气质量的评价，已是我们必须面对的问题。本论文将数据挖掘与数据融合两种技术应用于数据分析之中，建立空气质量评价模型，从而有效的进行空气质量评价。文中提出了目前大多数环境监测站所采用的计算机集散控制系统(DCS)、现场总线控制系统(FCS)后，根据其信息一般通过内部网、外部网实现本监测站内部互联，并且同行业内部的监测信息也实现互联，本文以环境空气监测监控系统为基础，对其中采集的信息进行抽象和简化，建立起空气质量评价系统的信息处理模型，实现了从数据到智能行为的转化。文中基于粗糙集理论对信息进行预处理，完成知识表达中不同属性的重要性分析，并进行知识表达空间的简化，再把简化后的系统特征参数作为神经网络的输入，建立起空气质量评价智能信息处理模型，进行空气质量评价最小属性评价因子研究。通过粗糙集和神经网络二者的结合，提高系统信息处理的容错能力及抗干扰能力，为处理海量数据的不确定性、不完整性提供了一条有效的途径。文中基于遗传蚁群(GAAA)算法对神经网络参数进行优化，提高网络逼近精度，克服所建立的空气质量评价信息处理模型的不足，从而减少计算量，使建立的空气质量评价信息处理模型更加切合实际。经过仿真试验，结果表明，本文所提出的智能信息处理技术，相对于目前存在的信息处理技术具有智能性、扩展性、伸缩性，评价结果具有客观、正确等优越性。此评价方法不但能对空气质量进行科学评价，而且还可以应用到水质综合评价等领域；不但可以通过4个因子进行空气质量评价分级，而且可以不受污染物种类和数目的限制，完成多个评价因子的空气质量综合评价。更多还原

【Abstract】 The gathering data of the environment air quality monitoring system is very enormous. Aiming at the magnanimous data, how to carry through air quality evaluation became a question which we must to face. The paper used the data mining and the data fusion technology to apply in data analysis, established the air quality evaluation model, and then carried on the air quality evaluation effectively.At present, the majority environmental monitoring stations have adapt computer distribution control system (DCS), field-bus control system (FCS), the information was interconnected to own interior station and even to internal monitor information system of craft brothers through intranet and extranet. On the base of environment air monitoring and supervisory system, the gathering information was simplified, and the information process model for air quality evaluation system was established. The model can realize transformation from the data to the intelligent behavior.Based on rough set method, the information was pretreatment, weightiness analysis of different attributes in the knowledge expression were completed, and we carried on simplification of the knowledge representation space, then put the simplified system characteristic parameters as inputs to the neural network, established a intelligent information process model for air quality evaluation, carried on the study of the smallest evaluation factors of the air quality evaluation. Through the combination of rough set and neural network, the way improved the information process system which has better fault-tolerant and anti-jamming ability. It provides a powerful way for process magnanimous indefinite and incomplete data.The neural network was optimized based on the genetic-ant (GAAA) algorithm. The way enhanced the network to approach the precision, conquered the disadvantages of the information process model for air quality evaluation, then reduced computing quantity and made the information process system for air quality evaluation to conform reality. Through the simulating experiment, the experiment result indicated that the intelligent information process technology which was put forward has intelligent, expansibility and retractility than other exist information process technologies. The evaluation result is impersonality and precise. The method not only may carry on the air quality evaluation scientific, but also may put in the water quality synthetic evaluation and so on, and may carry on the air quality evaluation graduation through four factors, moreover may complete the air quality synthetic evaluation deal with mult-evaluatioin factors, without the limit of contamination sorts and amounts.更多还原