가변속 냉동시스템의 고장 진단

Abstract

A refrigerating machine is basically consisted of an evaporator, a condenser, a compressor, an expansion valve, and pipes connected to each part. It is mainly applied to the fields of heating, ventilating, and air-conditioning(HVAC). Lately, the refrigerating machine has become more complicated owing to system integration of mechanical machines and electrical machines. Thus efficiency and reliability of the system have been improved. However, this tendency makes it more expensive to deal with the fault when the system is failed. Moreover, even though a simple fault of the system is happened, it can cause huge damage by inducing whole system shut-down. Therefore, an efficient fault diagnosis is very important. We need experts who have deep knowledge and skilled experiences about the complicated target system for fault treatment. Unfortunately, it is not easy to keep the expert in a same position so long time because of frequent job transfer. Therefore, the construction of fault diagnosis system that rudimentary engineers can diagnose problems fast and exactly like experts is essential to ensure the reliability and progress cost-efficiency. We suggest the off-line fault diagnosis based on the fault case data, the backward reasoning using T-invariance of Petri-net(PN), and the inter-node search manner. At first, a PN model is built from complicated fault case data to express cause and effect of each fault. Then, the PN model is analyzed using the backward reasoning of transition-invariance to retrieve out causes of target fault about a refrigerating machine. The inter-node search manner is utilized to reduce the search area of fault causes in this step. Especially, new fault knowledge about the refrigeration system can be reflected very easily in the proposed method. Next, the Man Machine Interface(MMI) is designed to diagnose for beginners easily and exactly like experts. As all arithmetic of MMI for fault diagnosis is performed by matrix information, we can reduce computer memory for the fault diagnosis. Finally, the validity of the proposed method is successfully confirmed through an application example about a compressor in a refrigerating machine. We suggest on-line fault detection and diagnosis(FDD) for a variable speed refrigeration system(VSRS) based on current information analysis of a compressor. Off-line fault diagnosis based on rule based reasoning is necessary to combine with on-line fault diagnosis to conduct the fault diagnosis about various and complicate symptoms from the refrigeration system. We note especially that the inverter driven compressor motors can be unavoidably exposed to hostile environment such as ripple in electrical torque or frequent speed change comparing to the motors under conventional control schemes. Moreover some faults based on only temperature and pressure will not be detected because of its inherent long time constant of the system and compensation of the control variables by control loop. Hence it is shown that the current information can play an important role in the more precise and quick FDD of the refrigeration system. Therefore a novel on-line FDD scheme using not only temperatures and pressures but also analysis of the harmonics and amplitude of the current from compressors is suggested at first. Then we set up the experimental variable speed system for FDD. Finally, the proposed method is applied to a FDD of a refrigeration system driven by the induction motor with variable voltage variable frequency(VVVF) inverter to confirm the validity of the proposed system.