Sensor Fault Diagnosis and Compensation for Chilled Water Unit Based on Soft Sensor
The chilled water unit is closely correlated with the temperature control within a building,and malfunctions in its internal sensors can directly impede its proper operation,consequently leading to a decline in energy utilization efficiency.Therefore,research on fault diagnosis and compensation strategies for sensor networks is of paramount significance to ensuring the reliable and stable operation of chilled water units and optimizing indoor comfort.This paper proposed a fault diagnosis and compensation method for chilled water unit sensing systems based on soft sensors.Firstly,a temperature monitor based on a fault diagnosis soft sensor was established,and integrated with the measured temperature sequence to detect sensor faults by residual error.Meanwhile,a novel fault threshold construction method,considering both the false alarm rate and fault detection rate,was proposed to achieve accurate sensor fault diagnosis.On this basis,faulty sensors were located through mutual checking among multiple sensors.Secondly,for detected faulty sensors,soft sensors for compensation were constructed,using the historical normal values of faulty sensors to estimate the current value and fulfill data recovery.Finally,the effectiveness of the proposed fault diagnosis and compensation method was validated through experiments.The results indicated that the average false alarm rate of the fault diagnosis soft sensor was 1.29％,and the average fault detection rate was 97.18％.The mean absolute error and root-mean-square error for the fault compensation soft sensor were 0.030 6 and 0.048 0,respectively.
soft sensorchilled water unitfault diagnosis and compensationsensordeep learning
NETL
NSTL
万方数据
