Optimization Research On Fault Diagnosis Method Of Chiller

With the increasing energy consumption of the society,the problem of building energy conservation has become increasingly prominent,and the HVAC system,as the most energy-consuming part of the building,bears important energy-saving tasks.As the main energy-consuming equipment in the HVAC system,the chiller maintains the normal operation of the equipment and reduces the energy loss caused by the fault.However,the abnormal working conditions caused by the chiller failure in the building HVAC system frequently occur,so timely and accurate detection and diagnosis after the fault has become a hot research topic,and timely stopping the energy loss caused by the fault helps to save energy.This paper focuses on the research of fault location diagnosis of chillers,analyzes the causal relationship between fault causes and fault phenomena,and introduces fault related information as additional evidence for fault diagnosis.A fault diagnosis method for chillers based on Bayesian network is proposed.To some extent,it solved the difficulties caused by the complexity of faults and signs,and conducted in-depth research on the following contents:(1)The working principle of the centrifugal chiller is described in detail,and the typical fault types are analyzed in depth.(2)Based on FMEA data form and expert knowledge,network structure learning,based on Delphi expert scoring and historical data analysis,network parameter learning,establish FDD model.(3)Introduce equipment-related information as additional evidence for diagnosis,solve the complication between the cause and symptom of chiller failure,and establish a composite Bayesian network structure model based on ICF(Information-Cause-Phenomenon)to improve Reliability of diagnostic results.(4)The experimental results of ASHRAE-1043 were used as the data source.After the steady-state screening,as the learning and verification source of this model,the final inference judgment was implemented in the Ge Nle software by using the group combination tree method.(5)The CF model and the ICF model are respectively verified by fault reasoning,and the diagnostic accuracy rate is analyzed.The ICF model is fault-inferred under the condition of incomplete information,and the diagnostic accuracy rate is analyzed.The results show that the model can effectively improve the correct rate of fault diagnosis of chillers.