Content uploaded by Vasso Reppa
Author content
All content in this area was uploaded by Vasso Reppa on Nov 15, 2018
Content may be subject to copyright.
A preview of the PDF is not available
A Distributed Architecture for HVAC Sensor Fault Detection and Isolation
Abstract and Figures
This paper presents a design and analysis methodology for detecting and isolating multiple sensor faults in heating, ventilation, and air-conditioning (HVAC) systems. The proposed methodology is developed in a distributed framework, considering a multizone HVAC system as a set of interconnected nonlinear subsystems. A dedicated local sensor fault diagnosis (LSFD) agent is designed for each subsystem, while it may exchange information with other LSFD agents. Distributed sensor fault detection is conducted using robust analytical redundancy relations of estimation-based residuals and adaptive thresholds. The distributed sensor fault isolation procedure is carried out by combining the decisions of the LSFD agents and applying a reasoning-based decision logic. The performance of the proposed methodology is analyzed with respect to robustness, sensor fault detectability, and isolability. Simulation results are used for illustrating the effectiveness of the proposed methodology applied to an eight-zone HVAC system.
Figures - uploaded by Vasso Reppa
Author content
All figure content in this area was uploaded by Vasso Reppa
Content may be subject to copyright.
... 3) The fact that the MVFP model, provided in the Appendix, can be easily reconfigured in both parameters and dynamics to host more subsystems associated with the ICE allows the generalization of the results and enhances the applicability of the method. 4) This article considers a system described by nonlinear DAEs, with the relevant formulation given in Section II, while most papers in SFDI literature deal with systems described by ordinary differential equations (ODEs) [25], [26], [27], [28]. 5) Furthermore, the present article also includes performance metrics regarding the propagation of sensor faults between different monitoring modules, in Section VI-A, in addition to the local ones found extensively in literature [18], [29]. ...
... Using TU Delft's Blue supercomputer, the simulation was run 100 times with a noise level of 5% affecting all system sensors and a varying seed. Based on the acquired results, implementing the metrics defined in (25) and (26) yields P U G = (51.16 ± 0.81)% and P E G = (57.83 ...
This article proposes a distributed model-based methodology for the diagnosis of faults affecting multiple sensors used for condition monitoring and control of marine internal combustion engines (ICEs). To handle the complexity of the ICE, we consider it as a set of interconnected physical subsystems that constitute the physical layer. For every subsystem, the detection of sensor faults relies on the design of cyber agents, where every agent monitors one subsystem. To handle the heterogeneous dynamics of each subsystem in the fault detection decision-making process, each agent uses differential and algebraic residuals alongside adaptive bounds. For isolation purposes, a combinatorial decision logic is employed, realized in two cyber levels: the local and the global decision logic. The first aims at the recognition of all sensor fault patterns that might have affected the engine based on the local agent fault signatures and certain binary decision matrices. The latter is used to capture the propagation of sensor faults between the different monitoring agents. Simulation results are used to showcase the proposed methodology’s efficiency in tackling the problem and its applicability.
... appears on Ma and Wang [2]. In Reppa et al. [3] the sensor faults do have a time varying quality, but the fault severity does not reflect a realistic case, as it is needed in our study. Padilla et al. [4] and Wang et al. [5] present a case where sensor fault does behave in a severity that is of the correct magnitude, but the bias applied to the system is fixed. ...
... The lower limit is set to −0.5 to take estimation uncertainties into consideration. In such cases the supervisor will ignore the bias compensation and directly receive the sensor measurement signals, i.e. 3 . Upon receiving the fault flag, the model parameters will be reinitialized followed by a new training process. ...
... One approach to anomaly detection in HVAC systems is the application of computational intelligence (CI) techniques, followed by optimizing climate satisfaction and environmental well-being [4]. Another approach to anomaly detection in HVAC systems is using fault detection and isolation (FDI) techniques to detect and isolate sensor faults in HVAC systems [5]. Anomaly detection techniques, such as outlier detection, have also been applied to HVAC systems. ...
HVAC systems are important in buildings due to their significant energy consumption, impact on indoor air quality, and role in occupant comfort. Optimizing the operation and control of these systems is crucial for improving energy efficiency and reducing costs. Anomaly detection in HVAC systems aims to optimize energy consumption, improve thermal comfort and indoor air quality, detect and isolate sensor faults, and, more importantly, detect cyber-attacks. By analyzing system data for unusual patterns or unauthorized access attempts, anomaly detection can play a vital role in safeguarding HVAC systems against cyber threats. Detecting and isolating potential cyber-attacks can prevent disruptions in building operations, protect sensitive data, and ensure the continued functionality of HVAC systems securely and reliably. In this study, Gradient Boosting Regressor is used to improve the anomaly detection capabilities of HVAC systems. Traditional anomaly detection methods often struggle to adapt to the dynamic nature of HVAC systems and may generate false alarms or miss critical issues. To address these challenges, we propose the application of Gradient Boosting Regressor, a powerful machine learning technique, to enhance anomaly detection accuracy and reliability. We evaluate the model's performance using real-world HVAC data, comparing it with existing anomaly detection methods. The results demonstrate significant improvements in the system's ability to identify anomalies accurately while minimizing false alarms. This research advances HVAC system security by providing a more robust and adaptive anomaly detection solution. Integrating Gradient Boosting Regressor into the cybersecurity framework of HVAC systems offers improved protection against cyber threats, thereby enhancing the resilience and reliability of critical infrastructures.
... Similarly, the safety of miners might be jeopardized if a sensor reading in a wireless underground sensor network, which is employed to monitor underground mines, is manipulated. Hence, it is crucial to consistently verify the state of a sensor and promptly identify any deviations to prevent such occurrences (Reppa, V., 2014). ...
Wireless sensor networks (WSNs) consist of many sensor nodes that are densely deployed throughout a randomized geographical area to monitor, detect, and analyze various physical phenomena. The primary obstacle encountered in WSNs pertains to the significant reliance of sensor nodes on finite battery power for wireless data transfer. Sensors as a crucial element inside Cyber-Physical Systems (CPS) renders them vulnerable to failures arising from intricate surroundings, substandard manufacturing, and the passage of time. Various anomalies can appear within WSNs, mostly attributed to defects such as hardware and software malfunctions and anomalies and assaults initiated by unauthorized individuals. These anomalies significantly impact the overall integrity and completeness of the data gathered by the networks. Therefore, it is imperative to provide a critical mechanism for the early detection of faults, even in the presence of constraints imposed by the sensor nodes. Machine Learning (ML) techniques encompass a range of approaches that may be employed to identify and diagnose sensor node faults inside a network. This paper presents a novel Energy-aware and Context-aware fault detection framework (ECFDF) that utilizes the Extra-Trees algorithm (ETA) for fault detection in WSNs. To assess the effectiveness of the suggested methodology for identifying context-aware faults (CAF), a simulated WSN scenario is created. This scenario consists of data from humidity and temperature sensors and is designed to emulate severe low-intensity problems. This study examines six often-seen categories of sensor fault, including drift, hard-over/bias, spike, erratic/precision, stuck, and data loss. The ECFDF approach utilizes an Energy-Efficient Fuzzy Logic Adaptive Clustering Hierarchy (EE-FLACH) algorithm to select a Super Cluster Head (SCH) within WSNs. The SCH is responsible for achieving optimal energy consumption within the network, and this selection process facilitates the early detection of faults. The results of the simulation indicate that the ECFDF technique has superior performance in terms of Fault Detection Accuracy (FDA), False-Positive Rate (FPR), and Mean Residual Energy (MRE) when compared to other detection and classification methods.
... In general, SFDI methods emerged for the building HVAC system can be classified into model-based and data-driven approaches [3]. The model-based SFDI techniques use the analytical model of the process [4][5][6] and hence, they require an adequate and precise understanding of the HVAC system mathematical model. The HVAC system is generally composed of mechanical and electrical systems, including cold and heat source systems, air handling systems, Probabilistic PCA -✓ ✓ [12] PCA & pattern matching Operational ✓ ✓ [13] PCA & clustering analysis Simulation ✓ ✓ ✓ ✓ [14] Satizky-Golay PCA Operational ✓ ✓ ✓ ✓ [15] PCA Simulation (TRANSYS) ✓ ✓ ✓ ✓ [16] PCA & AFT Simulation (Modelica) ✓ ✓ ✓ ✓ [17] PCA ...
Part II of this article will be published in Volume 11, Number 2, April 2005. Poorly maintained, degraded, and improperly controlled equipment wastes an estimated 15% to 30% of energy used in commercial buildings. Much of this waste could be prevented with widespread adoption of automated condition-based maintenance. Automated fault detection and diagnostics (FDD) along with prognostics provide a cornerstone for condition-based maintenance of engineered systems. Although FDD has been an active area of research in other fields for more than a decade, applications for heating, ventilating, air conditioning, and refrigeration (HVAC&R) and other building systems have lagged those in other industries. Nonetheless, over the last decade there has been considerable research and development targeted toward developing FDD methods for HVAC&R equipment. Despite this research, there are still only a handful of FDD tools that are deployed in the field. This paper is the first of a two-part review of methods for automated FDD and prognostics whose intent is to increase awareness of the HVAC&R research and development community to the body of FDD and prognostics developments in other fields as well as advancements in the field of HVAC&R. This first part of the review focuses on generic FDD and prognostics, providing a framework for categorizing methods, describing them, and identifying their primary strengths and weaknesses. The second paper in this review, to be published in the April 2005 International Journal of HVAC&R Research, will address research and applications specific to the fields of HVAC&R
This paper presents a distributed methodology for detecting and isolating multiple sensor faults in interconnected nonlinear systems. For each of the interconnected subsystems, a corresponding Local Sensor Fault Diagnosis (LSFD) agent is designed, which is allowed to exchange information with neighboring LSFD agents. The decisions of the LSFD agents are integrated and processed by a Global Sensor Fault Diagnosis (GSFD) agent in order to isolate multiple sensor faults propagating between neighboring LSFD agents through information exchange. The combined local and global isolation decision logic is designed based on diagnostic reasoning, taking into account the effects of sensor faults in the local sensor set and the transmitted faulty sensor information on the agents' decisions.
This paper introduces a robust method for performing active actuator fault detection and diagnostics (FDD) in heating ventilation and air conditioning (HVAC) systems. The proposed actuator FDD strategy, for testing whether an actuator is stuck in a given position, is designed on using an invariant hypothesis testing approach and is an improvement of a previous strategy that employed an adaptive detection strategy. The parameter-invariant detector is formulated to provide a constant detection performance, invariant to unknown building parameters, and it is described how this approach can replace the adaptive detector in the previous work. A closed-loop experimental HVAC testbed at the KTH Royal Institute of Technology campus in Stockholm, Sweden is introduced and employed to evaluate the parameter-invariant detector.
This paper focuses on the development of a suitable Fault Detection and Isolation (FDI) strategy for application to a system of inter-connected and distributed systems, as a basis for a fault-tolerant Network Control System (NCS) problem. The work follows a recent study showing that a hierarchical decentralized control system architecture may be suitable for fault-tolerant control (FTC) of a network of distributed and interacting subsystems. The main idea is to use robust FDI methods to facilitate the discrimination between faults acting within one subsystem and faults acting in other areas of the network, so that a powerful form of active FTC of the NCS can be implemented, through an autonomous network coordinator. By using a robust form of the Unknown Input Observer (UIO), fault effects in each subsystem are de-coupled from the other subsystems, thus facilitating a powerful way to achieve local FDI in each subsystem under autonomous system coordination. Whilst the autonomous distributed control system provides active FTC under learning control, the FDI-based Reconfiguration Task enhances the network fault-tolerance, so that more significant subsystem faults can be accommodated in order to achieve a suitable standard of Quality of Performance (QoP) of the NCS.
This paper presents a passive robust fault detection and isolation approach using non-linear interval observers. In industrial complex systems there is usually some uncertainty on model parameters that can be bounded using intervals. A model with parameters bounded in interval is known as an "interval model". Intervals observers propagate parameter uncertainty to the residual generating an adaptive threshold that allow to robust detect system faults. In order to isolate faults, a bank of those observers with a specified fault signature is required. Finally, this approach will be applied to detect and isolate some of the faults proposed in an industrial actuator used as an FDI benchmark in the European project DAMADICS.
This work considers the problem of isolating actuator and sensor faults in nonlinear process systems. The key idea of the proposed method is to exploit the analytical redundancy in the system through state observer design. To this end, we consider subsets of faults, and design state observers that use information of inputs and outputs only subject to faults in each subset. We then design residuals using the process model and state estimates such that each residual is only sensitive to the corresponding subset of faults. The occurrence of faults in a subset is detected if the corresponding residual breaches its threshold. With the ability of detecting the occurrence of faults in a subset, faults can be isolated using a bank of residuals and a logic rule. The proposed method enables differentiation between and isolation of actuator and sensor faults while explicitly accounting for system nonlinearity. The effectiveness of the fault isolation design subject to plant-model mismatch and measurement noise is illustrated using a chemical reactor example.