Research On Key Technologies Of Performance Prediction And Anomaly Detection For Secondary Supply Networks Of District Heating Systems

With the sustained development of China's economy and the acceleration of urbanization,the district heating system is also booming,which provides a comfortable living environment at the same time,is gradually intensifying the current contradiction between energy supply and demand.At the same time,the energy efficiency of central heating in China is only about 1/3 of that in developed countries,so it is imperative to study the energy saving of central heating.Aiming at the key components and key problems in the secondary heat supply network of central heating at present,this paper analyzes the thermodynamic operation state of plate heat exchanger in the secondary heat supply network under various complex working conditions,heat load prediction and analysis for different building characteristic parameters and climate parameters,and the detection of abnormal heat consumption on the user side.The main research work is as follows:A simple and accurate real-time control model of two-phase flow plate heat exchanger is presented.The basic characteristics of heat and mass transfer of plate heat exchanger are studied theoretically and experimentally.Aiming at the problem that the steam side of two-phase flow plate heat exchanger contains phase change process,the calculation of heat transfer is relatively complicated,and the parameters of each working condition are difficult to be measured or acquired,a plate heat exchanger model for two-phase flow is proposed.Firstly,the heat transfer process is mathematically described by mechanism analysis and heat and mass transfer equation,and the parameters are lumped.Linear least squares method and nonlinear least squares method are used to identify the unknown parameters in order to reduce the prediction error in application,and finally the relationship between controllable or predictable parameters is obtained,which satisfies the real-time requirement.The application requirements of control and optimization can improve the quality of heating.This paper presents an accurate heat load forecasting model for central heating x users considering building characteristic parameters and climate parameters.Based on the practical application and theoretical calculation,the model uses the advantages of physical and empirical modeling methods to describe the real-time heat transfer process of a room from the classical engineering equations of thermodynamics law,and derives the basic control equations.At the same time,a simple and accurate hybrid model structure is deduced by lumped and identified variables.The validity and accuracy of the model are studied and validated on the basis of the actual residential heat consumption data of the selected end users.The proposed modeling method takes into account the user's building parameters,outdoor temperature,wind speed,height and other factors affecting the thermal load of the building,and the quantitative effects of the above factors on the thermal load of the building are studied through experiments.The model provides guidance and suggestions for heat balance,on-demand distribution and network design between secondary heating networks.An anomaly detection method of user side heating for two heating network is put forward.Based on the proposed thermal load forecasting method,the original data in low-dimensional space are projected into high-dimensional space for clustering or classification calculation by using Kernel Gaussian Mixture Cluster(KGMC)and Kernel Gaussian Mixture Inferential Index(KGMMII).In order to achieve.Compared with K-Means and Gaussian Mixed Mode(GMM),the detection rate and false alarm rate are improved to some extent.On the basis of the proposed anomaly detection algorithm,the data characteristics of the anomaly heat,such as openning windows,installation of pressurized pumps,etc,for different building types are judged.At the same time,the accuracy of the proposed detection algorithm is measured by comparing with the household measurement data.The anomaly detection method will help heating companies to detect anomalies actively,do a good job of heating regulation,improve energy efficiency and thermal comfort.A user thermal load data processing platform based on large data,i.e.industrial large data acquisition and analysis platform,is built.Yarn is used as the underlying resource management platform of Spark system,and yarn-cluster mode is adopted.The related parts and structure of the platform are as follows:data source,data acquisition layer,data analysis layer,data storage layer.Data processing flow;heating data is directly imported to HDFS or through Sqoop into the data analysis layer,through iterative calculation,KGMC clustering and KGMMII classification are implemented,the final results are stored in HDFS,for specific application analysis.