Data-driven modeling, fault diagnosis and optimal sensor selection for HVAC chillers

Chillers constitute a significant portion of energy consumption equipment in heating, ventilating and air-conditioning (HVAC) systems. The growing complexity of building systems has become a major challenge for field technicians to troubleshoot the problems manually; this calls for automated "smart-service systems" for performing fault detection and diagnosis (FDD). The focus of this paper is to develop a generic FDD scheme for centrifugal chillers and also to develop a nominal data-driven ("black-box") model of the chiller that can predict the system response under new loading conditions. In this vein, support vector machines, principal component analysis, and partial least squares are the candidate fault classification techniques in our approach. We present a genetic algorithm-based approach to select a sensor suite for maximum diagnosabilty and also evaluated the performance of selected classification procedures with the optimized sensor suite. The responses of these selected sensors are predicted under new loading conditions using the nominal model developed via the black-box modeling approach. We used the benchmark data on a 90-t real centrifugal chiller test equipment, provided by the American Society of Heating, Refrigerating and Air-Conditioning Engineers, to demonstrate and validate our proposed diagnostic procedure. The database consists of data from sixty four monitored variables of the chiller under 27 different modes of operation during nominal and eight faulty conditions with different severities. Note to Practitioners-Heating, ventilating and air-conditioning (HVAC) systems constitute largest portion of energy consumption equipment. Even though safety is not a critical issue in HVAC industry, the complexity of modern HVAC systems, the operational and maintenance costs associated with the equipment are calling for sophisticated automatic fault diagnosis tools. Among the HVAC components chillers are primarily known for significant energy consumption. Presently, very little of the existing research on fault diagnosis of HVAC systems is aimed at chillers and suffers from certain drawbacks. The primary goal of this paper is to address these issues and develop a generic fault diagnosis tool applicable to any HVAC component. In this vein, we proposed a data-driven approach based on neural network and statistical tools for fault diagnosis, and nominal model development. Since chillers are an example of a data-rich environment we adopted a genetic algorithm based approach for optimal sensor selection for maximizing the diagnosibility. The approach is also validated on an experimental data from 90-t centrifugal chiller provided by the American Society of Heating, Refrigerating and Air-Conditioning Engineers and also has a potential for practical application.