Research On Data Driven Fault Detection And Diagnosis Of Air-conditioning System

With the development of social development,the number of buildings in China is increasing,and the energy consumption is also constantly increasing.The energy of HVAC systems is one of the most important parts of the energy.Research work on energy saving of HVAC is imminent.At the same time,as one of the most important places for human activities,its function and environment has great significance to production and life.With the improvement of people’s life quality,the stability of HVAC,indoor environment comfort and so on have also become the core concerns in the field of HVAC research.HVAC fault is one of the main factors that affect building stability,comfort,and energy saving.Therefore,research on HVAC fault is of great significance and value.Since improvement of building monitoring systems,the research based on monitoring data has become an inevitable direction for future development in this field.Based on the discussion,this paper grasped the points of the fault research development,and conducted a detailed investigation of the status of fault diagnosis,considered the shortcomings and advantages of the methods,based on the monitoring data,using data mining models to achieve automatic fault diagnosis.This paper carried out the work on three aspects: "diagnostic method system","fault diagnosis model" and "diagnostic model application".In the part of diagnosis method system,based on the existing theoretical achievements,this paper makes an indepth analysis of the air conditioning system and typical equipment characteristics,combined with engineering experience,taking common air conditioning system as the research object,and sorts out the typical faults in the system in detail.Study and clarify the characteristics of fault characteristics,judgment conditions,fault impact and other characteristics,establish AHU,terminal,and cold source fault rule bases respectively,and use the fishbone diagram based on this to logically sort out typical fault characteristics and form 8 fault diagram.This paper focuses on the system’s stability,effectual and energy-saving.Based on their fault characteristics,Bayesian networks,decision trees,and prediction algorithms are used for data mining analysis methods.To build three fault detection and diagnosis models.The three types of diagnostic models established in this research are deployed on the cloud platform to form a "fault detection and diagnosis tool" and applied to projects.