Data-based Diagnosis Of Abnormal Energy Consumption In Public Buildings

In the context of social energy conservation and emission reduction requirements,the issue of building energy consumption has attracted wide attention,and the high energy consumption of public buildings is undoubtedly crucial to the study of building energy conservation.Energy consumption diagnosis of public buildings is conducive to the formulation of energy planning and energy saving measures,which can effectively reduce energy waste.Taking an office building as the research object,this thesis establishes energy diagnosis models for central air conditioning system and lighting socket system to diagnose abnormal energy consumption.The main research contents and results are as follows :(1)Construction of energy evaluation system for central air conditioning system in office buildings.In order to better understand the energy consumption level of central air conditioning system and help to find out the abnormal energy consumption of central air conditioning system,an energy consumption evaluation method for single office building air conditioning system is studied.The abnormal value identification,missing value filling and normalization method of central air conditioning system in office buildings are introduced in detail.The grey correlation analysis method is used to determine the outdoor temperature and date attributes have great influence on the energy consumption of air conditioning system.Four energy consumption patterns of air conditioning systems were found by K-Means clustering algorithm.The energy consumption evaluation strategy is established for the working-day energy consumption model with the average energy consumption as the energy consumption benchmark,the one-quarter digit of energy consumption as the guiding value,and the three-quarter digit of energy consumption as the constraint value.(2)Construction of abnormal energy use diagnosis model for central air conditioning system in office building.In order to improve the efficiency of Apriori algorithm in finding association rules between equipment operation parameters and energy consumption of central air conditioning system,the algorithm is optimized according to the inherent attribute characteristics of air conditioning system data.The improved algorithm maps data to a new data storage structure,which only needs to scan the database once.It increases the intensity of transaction pruning.Reduces the generation of redundant candidate itemsets.The improved Apriori algorithm is applied to the data mining of air conditioning system,which improves the efficiency of data analysis.Through the evaluation and analysis of the obtained rules by domain knowledge,it is successfully found that the temperature difference between the chilled water supply and return water in the air conditioning system is too small,the non-working time energy consumption of the air conditioning system equipment,the cold machine load rate is too low in some time,and the meter fault is four kinds of abnormal energy consumption,which provides guidance for the energy saving management and operation of the building.(3)Energy consumption diagnosis of lighting socket system.Firstly,the energy consumption prediction model of lighting socket system based on LSTM neural network is established,and the prediction effect of the model under three different time lengths of 24 h,48h and 72 h is studied.Among them,the prediction effect of 24 h is the best,and the average relative error is 3.62%.The energy consumption diagnosis method based on energy consumption prediction is established and applied to the actual building lighting socket system.The abnormal energy consumption situation is found,and the reasons that may lead to abnormal energy consumption are given,including part of the lighting lamps are not closed and the equipment is standby.