Energy Consumption Monitoring And Prediction Research Of The Office Buildings In Fuzhou

Building energy consumption accounts for about one-third of the total energy consumption of national economy,building energy conservation is an important aspect of energy conservation work.There is huge energy saving space in the large public buildings energy system which is complexity with numerous users and large per unit area energy consumption.Efficient building energy consumption monitoring system provides real-time power consumption of each energy systems,so the weak link of the energy can be found out and the basis for the management and operation of energy-saving buildings was provided.The domestic and international study of buildings energy consumption statistics,related application software research and development,the building energy consumption prediction was summarized.Building energy consumption was usually predicted by gray system,artificial neural networks or linear regression algorithm,but poor ability for solving nonlinear problems and over-fitting problems existed in these algorithms.Building energy consumption monitoring system of the office building in Fuzhou region was established.The partition and operation and the energy systems of the building were studied.The monitoring system consists of data acquisition platform,data transmission platform and data management platform.According to the characteristics of energy equipment and the necessary of the energy management,the energy sub-items is designed combining the low-voltage distribution systems tree diagram.The design ideas such as "directly monitoring key energy consuming devices ","directly monitoring few circuit devices,indirect monitoring multi-circuit equipment" and "directly monitoring large power fluctuation devices" are proposed.For the confusion of the distribution line in the building,"addition principle" and"subtraction principles" are applied in sub-metering for the distribution circuit with few mixed branch and "proportional split method" is applied in sub-metering for the distribution circuit with mixed multi-branch to simplify the system.Demand for monitoring system database of the building energy consumption is analyzed and the conceptual structure and logical structure of the database are determined.Then,the building energy database is established and the databases have been debugged.The building energy monitoring system has the functions of data acquisition,data processing,data analysis,data display and alarm.The building energy consumption monitoring system has been running very well for more than one year.Based on a lot of basic data and analytical processing from energy consumption monitoring system,peak energy consumption of day and night and energy consumption of working and rest days were analyzed and compared by"waveform eigenvalue method".The daily status of energy saving management is showed and the weakness of energy saving in equipment operation is revealed.Based on GM and LSSVM algorithm,GM-LSSVM combination method is applied in the forecast of building energy consumption for the first time.Gray correlation analysis is applied to determine the main factors affecting building energy consumption.Energy consumption samples collected by GM methods are accumulated.Then the processed data are taken as training samples to establish LSSVM input-output model.The selection of the model parameters were optimized by particle group method.Less sample data is required by this method combining gray system which has advantages for solving nonlinear problems by LSSVM method.MATLAB simulation is applied to validate the model by calculating the objective evaluation target value of MRE,MAE and RMSE.It is proved that the model has stronger generalization ability and higher prediction accuracy,which gives data support to the early warning value setting of energy comsuption.