基于软测量的空调系统检测技术研究及其应用

发布时间：2018-01-04 00:03

本文关键词：基于软测量的空调系统检测技术研究及其应用　出处：《中国建筑科学研究院》2014年硕士论文　论文类型：学位论文

【摘要】：随着科学技术的发展,测试技术得到了飞速的进步,测试手段的发展日新月异,由传统的模拟仪器、数字化仪器、智能化仪器到现今的虚拟仪器时代。然而,在实际测试过程中仍存在许多无法或难以直接用传感器或过程检测仪表进行测量的重要过程参数,软测量技术是解决此类测试问题的有效途径。本文主要从冷水机组冷凝器污垢热阻和冷水机组综合部分性能系数IPLV两个方面展开研究,采用软测量技术中的基于机理的传统建模与基于数据驱动的建模两种方法,实现冷水机组冷凝器污垢热阻清洗报警在线监测及综合部分负荷性能系数(IPLV)检测。首先从传热学基本原理出发,建立污垢热阻计算理论模型,找出污垢热阻的影响因素,并分析了其对污垢系数的影响度。在此基础上,对理论模型进行改进,建立了基于不同对数平均温差定义的冷水机组污垢热阻测试模型。对两种模型的适用范围及各影响因素之间的关系展开讨论,并对模型误差进行分析。重点讨论了冷凝器传热系数、负荷率、冷凝器端差、污垢热阻之间的关系,为冷凝器污垢热阻的在线监测提供理论支持。在以上讨论的基础上,建立的基于制冷量衰减幅度冷凝器污垢热阻清洗报警策略,该模型根据实际运行工况的不同设定不同的报警限值限定,对冷水机组污垢热阻实现分类报警,避免了由于机组本身性能及运行条件的改变导致冷凝器端差变化的影响,适用于不同负荷率、不同机组的冷凝器污垢热阻实时在线监测。在LabVIEW2012语言平台下编写“冷水机组冷凝器污垢热阻清洗报警系统”程序,实现研究成果软件应用的转化。对建立的清洗报警系统进行不确定度分析。在10%以内的相对不确定度的情况下,反推各测量仪器的允许不确定度,从而对相关传感器准确度进行限定,保证监测的结果可靠性。最后对实际工程监测数据进行处理,比较四种BP模型训练周期,最终确定本文采用自适应学习率调整的BP建模方法,建立了结构为5×11×l的IPLV现场检测模型。通过4个模型的对比校核及实际机组性能曲线的比较,证明所建立的模型具有较高的预测精度,为现场获取机组的IPLV提供了便捷的途径。
[Abstract]:With the development of science and technology, testing technology has made rapid progress, the development of test methods by simulation instrument, change rapidly, the traditional digital instrument, intelligent instrument to the era of virtual instrument nowadays. However, there are still many important process parameters are unable or difficult to be directly with the sensor or process measurement instrument were measured in the actual testing process in the soft measurement technology is the effective way to solve such test problems.
This paper mainly launches the research from two aspects of water chiller condenser fouling resistance and water chiller performance comprehensive coefficient IPLV, the soft measurement technology based on traditional modeling mechanism and modeling method based on data driven two, realize the condenser fouling resistance chiller cleaning alarm monitoring and integrated part load value (IPLV) detection.
First, from the basic principle of heat transfer, the establishment of fouling resistance calculation model, find out the factors influencing the fouling resistance, and analyzed its effect on fouling coefficient. On this basis, the theoretical model was improved, established the chiller fouling test model based on the definition of different logarithmic mean temperature difference. The relationship between factors of scope of the two models and the influence is discussed, and the model error is analyzed. The paper discusses the condenser heat transfer coefficient, load rate, condenser terminal difference, the relationship between fouling resistance, provide theoretical support for the on-line monitoring of condenser fouling resistance.
Based on the above discussion, based on the cooling capacity decay rate of fouling in condenser cleaning alarm strategy, the model according to the different setting of the actual operating conditions of different alarm limits defined for the water chiller fouling resistance classification alarm, to avoid the impact caused the condenser end difference unit performance and changes in operating conditions. That is suitable for different loading rate, different units of the real-time online monitoring of fouling in condenser.
Chiller condenser fouling cleaning alarm system "program" in the LabVIEW2012 language platform, realize the transformation of research achievements of software application. Uncertainty analysis is performed to establish the cleaning alarm system. Within 10% relative uncertainty under the condition of the measuring instruments allow estimation uncertainty, and the related sensor the accuracy was limited, ensure the monitoring results of reliability.
At the end of the actual monitoring data processing, comparison of four kinds of BP model training cycle, finally determined by using the adaptive learning rate adjustment method of BP modeling, establish the structure for detection model of 5 * 11 * l IPLV site. By comparing the performance curves of the 4 models and checking the actual units, demonstrate that the prediction accuracy the model has high, provides a convenient way for the field acquisition unit IPLV.

【学位授予单位】：中国建筑科学研究院
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU831

【参考文献】