Sensor Fault Diagnosis for Air Handling Unit of Heating Ventilation and Air Conditioning Based on Voting Mechanism
Existing fault diagnosis methods developed for Air Handling Unit(AHU)of Heating Ventilation and Air Conditioning(HVAC)tend to be centralized.The few distributed methods usually require solving a large number of time-consuming optimization problems,making it impossible to complete fault diagnosis in a timely manner.In response to the above challenges,a distributed fault diagnosis method based on a novel voting mechanism is proposed.In this method,a novel voting mechanism is proposed to establish a Boltzmann machine to describe the sensor network,determine the edge weights of the Boltzmann machine through mutual voting among sensors,and iterate over the state of the Boltzmann machine,which is also the state of the sensors,based on the edge weights to locate the sensor faults.Moreover,a novel voting strategy based on Euclidean distance is designed to determine the voting values.Additionally,a method is developed to reset the Boltzmann machine's weight matrix by adding a node to the Boltzmann machine,which maintains the original voting relationship among the sensors while symmetrizing the Boltzmann machine to ensure convergence of the iteration of the Boltzmann machine state.This method does not need solving many optimization problems,leading to lower computational requirements compared to existing distributed methods.The proposed method is validated using actual data provided by ASHRAE Project RP-1312.The experimental results show that the proposed method can accurately and efficiently diagnose bias and drift faults in AHU sensors.
万方数据
维普
