Research On Energy-saving Diagnosis And Optimization Of Central Air-conditioning Based On Data Mining

Energy consumption of central air-conditioning accounts for more than 40% of the total building energy consumption,so the efficient operation of central air-conditioning under different working conditions have positive significance in reducing total energy consumption.Central air-conditioning is a non-linear and complex system composed of multiple sub-system,whose actual operation efficiency is influenced by the matching characteristic of each sub-system and the usage pattern.And meanwhile it is a dynamic system affected by multiple factors.Therefore,it is crucial to improve the matching degree of the output cooling capacity of air conditioning and actual load demand in the filed of air-conditioning energy saving.The operating data of central air-conditioning is a carrier reflecting the operating characteristics of system directly.And the data highlight the features of strong coupling correlation,multi-dimensional,huge amount,complexity and so on.With the help of pattern recognition,prediction and classification and decision support of data mining,the pre-processing of historical operating data,the feature parameter extraction based on factor analysis,the research on pattern recognition based on space-time coupling and the energy-saving diagnosis optimization based on decision tree were developed,fully excavating the potential information of operation parameters and providing macroscopic direction for the research of central air conditioning energy saving optimization.The details are as following.Firstly,the pre-processing of the operating data was studied,and 30.3% invalid data records were removed.Next,the study of the operation characteristics indicates that its operation has the characteristics of long-time,high-load and large-tolerance.Then,the redundancy problem of the operating data set was settled by the feature of parameter extraction based on factor analysis.The basic operating parameters of the central air conditioning(include 10 parameters)were synthesized into refrigeration factor,transport factor and load factor,and the three are independent of each other.Subsequently,the three operating modes of the system were concluded based on the characteristic factor and K-means clustering algorithm,which include low-frequency small temperature difference large flow in non-air conditioning season mode,high-frequency middle temperature difference low flow in air conditioning season mode,high-frequency high temperature difference small flow in air conditioning season mode.And meanwhile,the Integrated Partial Load Value(IPLV)and the Seasonal Energy Efficiency Ratio(SEER)of the system were analyzed.The results show the IPLV value(4.88 and 5.51)and the SEER value(3.21)of the system are unqualified with the related restricted standard value(6.0 and 3.5),which further indicates that the system has a certain space for energy saving optimization.Finally,based on the mutual interaction of each subsystem of central air conditioning and the structural characteristics of energy consumption/energy efficiency,the step-by-step energy efficiency diagnosis and optimization model was developed with the help of the logic relation demonstration ability and decision support ability of decision tree algorithm.It is validated that the proposed step-by-step energy efficiency diagnosis and optimization strategy has strong universality ability and good application reliability and economy.Its confidence is above 96.5%,and the maximum operating energy consumption can be saved by 32% and the annual operating cost can be saved by 55,000 yuan.The strategy takes the matching characteristics of each subsystem as the result orientation,and establishes the research framework of central air conditioning energy saving diagnosis and optimization based on the system level,ultimately providing a macro research direction of central air-conditioning energy saving.