水蓄冷性能监测、测评与分析

发布时间：2018-03-29 23:06

本文选题：时间序列　切入点：水蓄冷　出处：《中国地质大学(北京)》2017年硕士论文

【摘要】：水蓄冷作为缓解电力系统压力的一种有效手段,对于削峰填谷有着至关重要的作用。目前我国水蓄冷空调技术虽然不断发展普及,但对水蓄冷监测数据处理的研究还尚未开始。水蓄冷运行过程中,短时间内就会积累大量的监测数据。如何利用这些监测数据对水蓄冷空调的状态进行高效、准确的评估,并及时准确的预测水蓄冷蓄能设备的冷量负荷,已成为新的研究课题。论文以时间序列分析为基础,依次对水蓄冷运行数据做预处理、异常检测和趋势预测分析,三个过程相互衔接配合共同构成了水蓄冷运行监测模型。本文首先对水蓄冷蓄能罐的运行原理简单介绍。其次对水蓄冷运行监测模型深入研究:预处理部分,引入基于数据相对变化率的孤立点识别和基于最相似近邻法的空缺值填补,为异常检测与预测提供数据基础;异常检测部分,利用时间序列的相似性,在局部异常因子(LOF)的基础上,采用改进的局部异常系数(LOC)来衡量异常,对水蓄冷监测数据进行异常检测分析,提高算法性能;负荷预测部分,应用类比合成算法,从历史数据集中查询出最为相似的类比序列,从而生成预测。最后,通过整合所有算法,以Matlab为平台搭建了基于时间序列分析的水蓄冷运行监测系统,并使用北京某软件园实时监测数据对模型进行验证,结论与实际情况相符,验证了时间序列分析在水蓄冷运行监测的应用切实可行。此外,基于以上算法可以构建监测功能界面,实时监测并显示水蓄冷的运行状态,实现异常报警功能,形成集监测与预测于一体的水蓄冷分析系统。最后对该系统夏季运行情况做经济性分析,得出比常规系统节省21万元的结论。时间序列在水蓄冷运行监测数据方面的应用为水蓄冷空调系统的运行监测另辟蹊径,克服人工检测的低效性。本文构建的水蓄冷运行监测分析方法对于水蓄冷性能监测是一种新的尝试与探索。
[Abstract]:As an effective means to relieve the pressure of power system, water storage plays an important role in cutting the peak and filling the valley. At present, although the technology of water storage and air conditioning is developing and popularizing in our country, However, the research on the data processing of water storage monitoring has not yet begun. During the operation of water storage, a large amount of monitoring data will be accumulated in a short period of time. How to use these monitoring data to evaluate the state of water storage air conditioning efficiently and accurately, It has become a new research topic to predict the cooling load of water storage equipment in time and accurately. Based on the time series analysis, the paper makes preprocessing, anomaly detection and trend prediction analysis of the water storage operation data in turn. The three processes are connected with each other to form the monitoring model of water storage operation. Firstly, this paper introduces the operation principle of water storage storage tank. Secondly, the monitoring model of water storage operation is deeply studied. In order to provide data basis for anomaly detection and prediction, outlier recognition based on relative change rate of data and vacancy value based on most similar nearest neighbor method are introduced. Based on the local anomaly factor (LOF), the improved local anomaly coefficient (LOC) is used to measure the anomaly, and the abnormal detection and analysis of the monitoring data of water storage are carried out to improve the performance of the algorithm. By querying the most similar analogical sequence from the historical data set, the prediction is generated. Finally, by integrating all the algorithms, a monitoring system of water storage and cold storage operation based on time series analysis is built on the platform of Matlab. The model is verified by using the real-time monitoring data of a software park in Beijing. The conclusion is in accordance with the actual situation, and the application of time series analysis in the monitoring of water storage operation is proved to be feasible. Based on the above algorithm, the monitoring function interface can be constructed, the running state of water storage can be monitored and displayed in real time, and the abnormal alarm function can be realized. A water storage analysis system is formed, which integrates monitoring and prediction. Finally, the economic analysis of the system in summer is made. The conclusion is that 210000 yuan is saved compared with the conventional system. The application of time series in the monitoring data of water storage operation is a new way to monitor the operation of water storage air conditioning system. In order to overcome the inefficiency of manual detection, the method of monitoring and analyzing the operation of water storage is a new attempt and exploration for monitoring the performance of water storage.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM73;TB657.2

【相似文献】