SUMMARYThe objective of this study is to detect faults due to multiple element failures in HVAC systems occurringconcurrently. To classify and detect single as well as multiple faults, measurements were made of supply airtemperature, OA-damper position, supply fan pressure, indoor temperature and airflow rate in a variableair volume heating ventilating and air conditioning test facility. Experimental results show that three typesof patterns emerge in the analysis of multi-fault problems. To solve the multi-fault problem, a new strategybased on pattern classification and the use of residual ratios is presented. It is shown that the residual ratiocan be used to diagnose and accurately identify and detect multiple-faults occurring in HVAC systems.Copyright # 2005 John Wiley & Sons, Ltd.KEY WORDS: FDD (fault detection and diagnosis); VAV (variable air volume); neural network; HVACsystem; single-fault; multiple-fault1. INTRODUCTIONHVAC systems in buildings have become increasingly larger and complex. 35322
As such they requireclose monitoring to evaluate their performance with respect to energy efficiency and equipmentoperation. The increasing complexity has made it very difficult to detect the causes, locationsand effects of faults in various parts of building systems. Faults whether they occur in HVACequipment, instrumentation or control loops contribute to higher energy consumption, couldshorten equipment life, and degrade indoor environment. Therefore, a fault detectiontechnology is necessary that detects performance deterioration quickly and accurately so thata faster remedial action could be taken to improve indoor environment and ensure reliabilityand safety of the system.Many initial studies on HVAC systems have applied rule based diagnosis methods (Liu andKelly, 1998) to detect faults. Some other studies have used rule-based and statistical methods (Anderson et al., 1998). In Chen and Braun (2000) a fault diagnosis method based on optimalcontrol theory is presented. Norford proposed a fault diagnosis technique based on the usage ofelectric power. Chen and Braun (2001) applied an FDD technique to develop a method fordiagnosing and detecting faults in packaged air-conditioners.The use of knowledge-based methods to detect faults in HVAC systems, which areapplied to actual buildings, has grown in recent years such as the work done by Katipamula et al.(1999), House et al. (1999). Also, Lee et al. (1996) used a prediction based model for detectingAHU faults. He trained a neural network to recognize the fault patterns.
Recently, morereliable fault detection methods have been developed. These methods use residuals thatare insensitive to outer disturbances and have high fidelity in terms of identifying the faultpatterns.In this study, we are interested in exploring a pattern diagnosing method for detecting singleelement fault or combined faults involving multiple elements both of which occur in HVACsystems. To this end, first a single fault replication was done in which signals from outdoor airdamper, indoor supply temperature sensor, indoor temperature sensor, airflow rate sensor andsupply blower were used. Also multiple faults were replicated by applying faults in two elementsat a time. The following element combinations were used in multiple fault experiments: supplytemperature sensor and outdoor air damper; supply temperature sensor and supply flow rate;and supply temperature sensor and supply fan.2. EXPERIMENTAL TEST FACILITYExperiments were conducted in a test room housed in an environmental chamber (EC) testfacility (Figure 1).
The EC is a multi-purpose research and test facility. It is used to conductcomprehensive and controlled experiments such as evaluation of HVAC control strategies,energy efficiency, thermal performance of building envelopes and fault detection and diagnosticstudies. The test facility consists of a main chamber and a three floor level attached space. Onthese three floors several test rooms are built. The overall dimensions of the test facility are givenin Table I. The test facility can simulate different outdoor weather conditions for conductingdifferent tests irrespective of time of the year or seasons. For example, any temperature between 25 to 508C can be created using a pre-selected rate of temperature increase/or decrease tosimulate typical outdoor weather conditions. A complete list of thermal parameter specificationsof the EC are given in Table II.Several different test rooms were built in the EC on the first and second floors. The layout oftest rooms is depicted in Figure 2 and the overall dimensions of the test rooms are summarizedin Table III. As shown in Figure 2 the test facility consists of four different rooms. Theseinclude: an experimental room for performance monitoring and control of radiant floor heatingsystems (ONDOL in Figure 2); an experimental room for evaluating energy efficiency ofheating/cooling systems; an indoor environment and thermal comfort room; and an HVAC testroom for the study of FDD and control strategies. The HVAC test room can also be used toevaluate performance of under-floor air conditioning (UFAC) systems, variable air volume(VAV) systems and monitoring of indoor air quality (IAQ).The results presented in this paper are based on the tests conducted in the HVAC testroom shown in Figure 2. A schematic diagram of the variable air volume air handling Unit(AHU) and the location of sensors/actuators and controllers used in the fault detection and diagnosis experiments are shown in Figure 3. The AHU is a variable air volume unitwhich can vary the airflow rates in the system by modulating the fan speed via a pressurecontroller. Similarly, a return air fan is made to track the supply air fan while maintaining aconstant air flow difference. The two other control loops used included a discharge airtemperature control loop and an outdoor, return and exhaust damper control loop. 暖通空调系统故障检测英文文献和中文翻译:http://www.youerw.com/fanyi/lunwen_33211.html