Full metadata record
|dc.contributor||Department of Building Services Engineering||en_US|
|dc.publisher||Hong Kong Polytechnic University||-|
|dc.rights||All rights reserved||en_US|
|dc.title||A systematic fault diagnosis strategy for building HVAC systems||en_US|
|dcterms.abstract||The building energy use accounts for a large portion of energy end use of commercial sectors. The performance of HVAC systems is very important in terms of energy saving and energy efficiency. However, the HVAC systems may suffer various faults, such as fouling, pipe clog and improper control, etc. A great deal of research can be found on the performance evaluation and fault detection and diagnosis (FDD) for HVAC components. However, few researchers paid attention to the system-level FDD and the comprehensive structure of the building system diagnosis strategy. The thesis presents a three-level FDD strategy for the building HVAC systems, i.e. the building load estimation/forecast scheme for overall building performance, the system-level FDD scheme for the HVAC systems and the component-level scheme for the chiller. The building-level diagnosis scheme adopts a simplified building load estimation model as the benchmark to characterize the overall performance of the entire building system. The basic model of the building load estimation/forecast scheme is the thermal network model in thermal resistance and capacitance pattern. The building envelopes including exterior walls and roofs are represented by a thermal network model with three resistances and two capacitances. The internal mass is represented by a thermal network model with two resistances and two capacitances. The building load estimation scheme using monitoring weather information (e.g. solar radiation, outdoor air temperature/relative humidity) via building management systems (BMS) as the input of the building thermal network model is applied in the building-level diagnosis. The building load forecast scheme using the weather forecast information from the observatory as the input of the building thermal network model is applied in the optimal control strategies. In the forecast scheme, two weather prediction modules are incorporated into the thermal network model. One is the outdoor air temperature/relative humidity prediction module based on the grey dynamic model. The other is the solar radiation prediction module based on the cloud amount and temperature forecast from the observatory. Both weather prediction modules and the building load estimation/forecast scheme are validated using the field data. The system-level FDD scheme for the HVAC systems has two steps. The first step is to detect, diagnose the sensor, and to estimate the fault (i.e. sensor fault detection, diagnosis and bias estimation (FDD&E)) prior to the use of the system FDD method. The second step is to diagnose the system (i.e., system FDD) by using the sensor FDD&E as the guarantee of measurement health. Principal component analysis (PCA) as the basis of the sensor FDD&E is capable of capturing the variance of a number of correlated sensor measurements based on the first law. Using the normal or corrected sensor measurements, one or more performance indices (Pis) are obtained to characterize the performance of the HVAC systems. An on-line adaptive threshold of the PI residuals, determined by the training data and measured data, is used to define the normal range. The sensitivity analysis of the sensor FDD&E to system faults and the validation of the system-level FDD scheme are conducted using simulation data. As chillers take the largest part of the power consumption in HVAC systems, the component-level FDD scheme for the chiller is developed using fuzzy modeling and artificial neural network (ANN). Based on the sensitivity analysis tests, performance indices (PI) are selected to characterize the health status of the chiller. PI residual is defined as the difference between the PI model benchmark and PI measurement. All the PI residuals are fuzzified into a series of standardized quantitative Pis (SQPs) using membership functions. A SQP is an interval covering an operation range of PI residuals. All the SQPs cover the full operation range of PI residuals in the tests. SQP is very effective to distinguish the faults, even to the faults having the same qualitative rule patterns. Then, ANN is used to identify the chiller fault by matching the SQPs with the fault category. The scheme is validated using the laboratory data provided by the ASHRAE RP 1043. The three-level building HVAC system diagnosis strategy is developed into a software package implemented on IBmanager, which is an open integration and management platform for intelligent building systems based on the middleware technologies. As a function module of IBmanager, the software package of the building HVAC system diagnosis strategy is supposed to report alarms, generate the diagnosis results and recommend improvements through an Intelligent Control and Diagnosis System for a commercial building. This system is a platform working in the foreground of the working station as an interface between IBmanager and end-users.||en_US|
|dcterms.extent||xx, 188 leaves : ill. ; 30 cm.||en_US|
|dcterms.isPartOf||PolyU Electronic Theses||en_US|
|dcterms.LCSH||Hong Kong Polytechnic University -- Dissertations.||en_US|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: