Critical infrastructures such as power grids, communication and energy networks; potentially dangerous industrial processes such as nuclear or chemical plants; transportation systems or autonomous robots. These are examples of systems for which safety and resiliency should be an integral part of their design. The occurrence of faults in these cases can lead to unacceptable losses, or simply make their operation uneconomical. For instance, in the offshore wind energy sector Operation&Maintenance costs can already reach up to 30% of the total lifetime cost (May et al., 2015).
Detecting and accommodating faults before they lead to extreme consequences is thus a key requirement.
Model-based techniques constitute a powerful way to detect faults, by comparing the predictions of a mathematical model (or digital twin as often referred to in the industry) of the system to real-time measurements coming from sensors. The central problem of fault diagnosis then becomes detecting when differences between the two are caused by physiological uncertainties present in the model and the sensors, and when they are due to an actual, pathological fault.
We contributed by extending centralized model-based methods in order to handle large-scale, uncertain nonlinear systems (Ferrari et al., 2012). Then we focused on the central problem of getting better diagnosis performances in terms of the so-called False Alarm Rates and Missed Detection Rates. Our key to reaching this objective is moving from a deterministic to a probabilistic approach: in (Rostampour et al., 2017), (Rostampour et al., 2018), (Rostampour et al., 2020) we used a scenario approach to derive set-based thresholds with user-desired probabilistic robustness. Our research efforts are now focusing on deriving fast, general and efficient methods for quantifying and propagating uncertainties through arbitrary nonlinear systems.
Our work on fault diagnosis found applications in wind energy, cyber-physical systems security,
automotive and aerospace industries and in robotics.
Joint work with (mostly): Thomas Parisini, Marios M. Polycarpou, Francesca Boem, Vahab Rostampour, Yichao Liu, Ping Wu.
Publications
A Sliding Mode Observer Approach to Oscillatory Fault Detection in Commercial Aircraft
Keijzer, Twan,
Engelbrecht, Japie A. A.,
Goupil, Philippe,
and Ferrari, Riccardo M.G.
Control Engineering Practice
2023
Threshold Design for Fault Detection with First Order Sliding Mode Observers
Keijzer, Twan,
and Ferrari, Riccardo M.G.
Automatica
2022
Sliding Mode Observer (SMO) based methods have been extensively used for Fault Estimation (FE). However, the fault detection (FD) problem for these SMO based FE methods has not been completely solved. In this paper a robust threshold on the so-called Equivalent Output Injection (EOI) is presented which enables FD for systems with measurement noise and unmatched uncertainties. This threshold is applicable to a large class of existing SMO based FE methods, and its applicability can easily be verified. Theoretical guarantees on the detection performance of this threshold are provided, and further demonstrated via a simulation study.
Enabling Co-Innovation for a Successful Digital Transformation in Wind Energy Using a New Digital Ecosystem and a Fault Detection Case Study
Barber, Sarah,
Lima, Luiz Andre Moyses,
Sakagami, Yoshiaki,
Quick, Julian,
Latiffianti, Effi,
Liu, Yichao,
Ferrari, Riccardo M.G.,
Letzgus, Simon,
Zhang, Xujie,
and Hammer, Florian
Energies
2022
In the next decade, further digitalisation of the entire wind energy project lifecycle is expected to be a major driver for reducing project costs and risks. In this paper, a literature review on the challenges related to implementation of digitalisation in the wind energy industry is first carried out, showing that there is a strong need for new solutions that enable co-innovation within and between organisations. Therefore, a new collaboration method based on a digital ecosystem is developed and demonstrated. The method is centred around specific “challenges”, which are defined by “challenge providers” within a topical “space” and made available to participants via a digital platform. The data required in order to solve a particular “challenge” are provided by the “challenge providers” under the confidentiality conditions they specify. The method is demonstrated via a case study, the EDP Wind Turbine Fault Detection Challenge. Six submitted solutions using diverse approaches are evaluated. Two of the solutions perform significantly better than EDP’s existing solution in terms of Total Prediction Costs (saving up to €120,000). The digital ecosystem is found to be a promising solution for enabling co-innovation in wind energy in general, providing a number of tangible benefits for both challenge and solution providers.
Floating offshore wind turbine fault diagnosis via regularized dynamic canonical correlation and Fisher discriminant analysis
Wu, Ping,
Liu, Yichao,
Ferrari, Riccardo M.G.,
and Wingerden, Jan-Willem
IET Renewable Power Generation
2021
Abstract Over the past decades, Floating Offshore Wind Turbine (FOWT) has gained increasing attention in wind engineering due to the rapidly growing energy demands. However, difficulties in turbine maintenance will increase due to the harsh operational conditions. Fault diagnosis techniques play a crucial role to enhance the reliability of FOWTs and reduce the cost of offshore wind energy. In this paper, a novel data-driven fault diagnosis method using regularized dynamic canonical correlation analysis (RDCCA) and Fisher discriminant analysis (FDA) is proposed for FOWTs. Specifically, to overcome the collinearity problem that exists in measured process data, dynamic canonical correlation analysis with a regularization scheme, is developed to exploit the relationship between input and output signals. Then, the residual signals are generated from the established RDCCA model for fault detection. To further classify the fault type, an FDA model is trained from the residual signals of different training faulty data sets. Simulations on a FOWT baseline model based on the widely used National Renewable Energy Laboratory FAST simulator are carried out to demonstrate the feasibility and efficacy of the proposed fault detection and classification method. Results have shown many salient features of the proposed method with potential applications in FOWTs.
Fault-Tolerant Individual Pitch Control of Floating Offshore Wind Turbines via Subspace Predictive Repetitive Control
Liu, Yichao,
Frederik, Joeri,
Ferrari, Riccardo M.G.,
Wu, Ping,
Li, Sunwei,
and Wingerden, Jan-Willem
Wind Energy
2021
Fault diagnosis of the 10MW Floating Offshore Wind Turbine Benchmark: A mixed model and signal-based approach
Liu, Yichao,
Ferrari, Riccardo M.G.,
Wu, Ping,
Jiang, Xiaoli,
Li, Sunwei,
and Wingerden, Jan-Willem
Renewable Energy
2021
Privatized distributed anomaly detection for large-scale nonlinear uncertain systems
Rostampour, Vahab,
Ferrari, Riccardo M.G.,
Teixeira, André M.H.,
and Keviczky, Tamas
IEEE Transactions on Automatic Control
2020
Data Driven Incipient Fault Detection via Canonical Variate Dissimilarity and Mixed Kernel Principal Component Analysis
Wu, Ping,
Liu, Yichao,
and Ferrari, Riccardo M.G.
IEEE Transactions on Industrial Informatics
2021
Distributed fault diagnosis for continuous-time nonlinear systems: The input–output case
Boem, Francesca,
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
Annual Reviews in Control
2013
Abstract In this paper, some new results on distributed fault diagnosis of continuous-time nonlinear systems with partial state measurements are proposed. By exploiting an overlapping decomposition framework, the dynamics of a nonlinear uncertain large-scale dynamical system is described as the interconnections of several subsystems. Each subsystem is monitored by a Local Fault Diagnoser: a set of local estimators, based on the nominal local dynamic model and on an adaptive approximation of the interconnection and of the fault function, allows to derive a local fault decision. A consensus-based protocol is used in order to improve the detectability and the isolability of faults affecting variables shared among different subsystems because of the overlapping decomposition. A sufficient condition ensuring the convergence of the estimation errors is derived. Finally, possibly non-conservative time-varying threshold functions guaranteeing no false-positive alarms and theoretical results dealing with detectability and isolability sufficient conditions are presented.
Distributed Fault Detection and Isolation of Large-Scale Discrete-Time Nonlinear Systems: An Adaptive Approximation Approach
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
IEEE Transactions on Automatic Control
2012
This paper deals with the problem of designing a distributed fault detection and isolation methodology for nonlinear uncertain large-scale discrete-time dynamical systems. As a divide et impera approach is used to overcome the scalability issues of a centralized implementation, the large scale system being monitored is modelled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state components. For each subsystem, a Local Fault Diagnoser is designed, based on the measured local state of the subsystem as well as the transmitted variables of neighboring states that define the subsystem interconnections. The local diagnostic decision is made on the basis of the knowledge of the local subsystem dynamic model and of an adaptive approximation of the interconnection with neighboring subsystems. The use of a specially-designed consensus-based estimator is proposed in order to improve the detectability and isolability of faults affecting variables shared among overlapping subsystems. Theoretical results are provided to characterize the detection and isolation capabilities of the proposed distributed scheme. Finally, simulation results are reported showing the effectiveness of the proposed methodology.
Distributed Fault Diagnosis With Overlapping Decompositions: An Adaptive Approximation Approach
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
IEEE Transactions on Automatic Control
2009
This technical note deals with the problem of designing a distributed fault detection methodology for distributed (and possibly large-scale) nonlinear dynamical systems that are modelled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state components. For each subsystem, a local fault detector is designed, based on the measured local state of the subsystem as well as the transmitted variables of neighboring states that define the subsystem interconnections. The local detection decision is made on the basis of the knowledge of the local subsystem dynamic model and of an adaptive approximation of the interconnection with neighboring subsystems. The use of a specially-designed consensus-based estimator is proposed in order to improve the detectability of faults affecting variables shared among different subsystems. Simulation results provide an evidence of the effectiveness of the proposed distributed fault detection scheme.
Beta Residuals: Improving Fault-Tolerant Control for Sensory Faults via Bayesian Inference and Precision Learning
Baioumy, Mohamed,
Hartemink, William,
Ferrari, Riccardo M.G.,
and Hawes, Nick
In IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes
2022
Model-based fault-tolerant control (FTC) often consists of two distinct steps: fault detection & isolation (FDI), and fault accommodation. In this work we investigate posing fault- tolerant control as a single Bayesian inference problem. Previous work showed that precision learning allows for stochastic FTC without an explicit fault detection step. While this leads to implicit fault recovery, information on sensor faults is not provided, which may be essential for triggering other impact-mitigation actions. In this paper, we introduce a precision-learning based Bayesian FTC approach and a novel beta residual for fault detection. Simulation results are presented, supporting the use of beta residual against competing approaches.
Towards Stochastic Fault-tolerant Control using Precision Learning and Active Inference
Baioumy, Mohamed,
Pezzato, Corrado,
Corbato, Carlos Hernández,
Hawes, Nick,
and Ferrari, Riccardo M.G.
In International Workshop on Active Inference
2021
Fault-tolerant Control of Robotic Systems with Sensory Faults using Unbiased Active Inference
Baioumy, Mohamed,
Pezzato, Corrado,
Ferrari, Riccardo M.G.,
Corbato, Carlos Hernández,
and Hawes, Nick
In European Control Conference
2021
Fault Detection of the Mooring system in Floating Offshore Wind Turbines based on the Wave-excited Linear Model
Liu, Yichao,
Fontanella, Alessandro,
Wu, Ping,
Ferrari, Riccardo M.G.,
and Wingerden, Jan-Willem
In Science of Making Torque from Wind Conference
2020
Active inference for fault tolerant control of robot manipulators with sensory faults
Pezzato, Corrado,
Baioumy, Mohamed,
Corbato, Carlos Hernández,
Hawes, Nick,
Wisse, Martijn,
and Ferrari, Riccardo M.G.
In International Workshop on Active Inference
2020
Fast Adaptive Fault Accommodation in Floating Offshore Wind Turbines via Model-Based Fault Diagnosis and Subspace Predictive Repetitive Control
Liu, Yichao,
Wu, Ping,
Ferrari, Riccardo M.G.,
and Wingerden, Jan-Willem
In IFAC World Congress
2020
Adaptive fault accommodation of pitch actuator stuck type of fault in floating offshore wind turbines: a subspace predictive repetitive control approach
Liu, Yichao,
Frederik, Joeri,
Fontanella, Alessandro,
Ferrari, Riccardo M.G.,
and Wingerden, Jan-Willem
In American Control Conference
2020
Differentially-Private Distributed Fault Diagnosis for Large-Scale Nonlinear Uncertain Systems
Rostampour, Vahab,
Ferrari, Riccardo M.G.,
Teixeira, André M.H.,
and Keviczky, Tamas
In IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes
2018
A set based probabilistic approach to threshold design for optimal fault detection
Rostampour, Vahab,
Ferrari, Riccardo M.G.,
and Keviczky, Tamas
In American Control Conference
2017
Traditional deterministic robust fault detection threshold designs, such as the norm-based or limit-checking method, are plagued by high conservativeness, which leads to poor fault detection performance. On one side they are ill-suited at tightly bounding the healthy residuals of uncertain nonlinear systems, as such residuals can take values in arbitrarily shaped, possibly non-convex regions. On the other hand, they must be robust even to worst-case, rare values of the modeling and measurement uncertainties. In order to maximize performance of detection, we propose two innovative ideas. First, we introduce threshold sets, parametrized in a way to bound arbitrarily well the residuals produced in healthy condition by an observer-based residual generator. Secondly, we formulate a chance-constrained cascade optimization problem to determine such a set, leading to optimal detection performance of a given class of faults, while guaranteeing robustness in a probabilistic sense. We then provide a computationally tractable framework by using randomization techniques, and a simulation analysis where a well-known three-tank benchmark system is considered.
Optimal Topology for Distributed Fault Detection of Large-scale Systems
Boem, Francesca,
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes
2015
Abstract The paper deals with the problem of defining the optimal topology for a distributed fault detection architecture for non-linear large-scale systems. A stochastic modelbased framework for diagnosis is formulated. The system structural graph is decomposed into subsystems and each subsystem is monitored by one local diagnoser. It is shown that overlapping of subsystems allows to improve the detectability properties of the monitoring architecture. Based on this theoretical result, an optimal decomposition design method is proposed, able to define the minimum number of detection units needed to guarantee the detectability of certain faults while minimizing the communication costs subject to some computation cost constraints. An algorithmic procedure is presented to solve the proposed optimal decomposition problem. Preliminary simulation results show the potential of the proposed approach.
An algebraic approach for robust fault detection of input-output elastodynamic distributed parameter systems
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In European Control Conference
2013
This paper deals with the problem of designing a robust fault detection methodology for a class of input-output, uncertain dynamical distributed parameter systems, namely mechanical elastodynamic systems, which are representative of a whole class of problems related to on-line health monitoring of mechanical and civil engineering structures. The proposed approach does not require full state measurements and is robust to measuring, modeling and numerical errors, thanks to a time varying detection threshold. In order to avoid the problems associated with classical discretization techniques for distributed parameter systems, which can lead to numerical errors difficult to bound a priori, and thus higher thresholds, a suitable structure-preserving algebraic approach, called Cell Method, will be employed. This method consists in writing the equations of a distributed parameter system directly in discrete form, avoiding the usual discretization process and leading to a symplectic, that is energy preserving, numerical scheme.
Distributed fault detection for uncertain nonlinear systems: A network delay compensation strategy
Boem, Francesca,
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In American Control Conference
2013
This paper proposes a delay compensation strategy for a distributed fault detection architecture, allowing to manage delays and packet losses in the communication network between the Local Fault Diagnosers. A novel consensus-based estimator with time-varying weights is introduced, permitting to improve detectability in the case of variables shared among more than one subsystem. In the consensus protocol, at each step each agent uses only the information given by the agent and the communication link which are more reliable at that time. The convergence of the proposed estimator is demonstrated and analytical conditions for detectability are derived.
Distributed fault diagnosis for input-output continuous-time nonlinear systems
Boem, Francesca,
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes
2012
In this paper, new results on distributed fault diagnosis of continuous–time nonlinear systems with partial state measurements are proposed. Following an overlapping decomposition framework, the dynamics of a nonlinear uncertain large-scale dynamical systems is described as the interconnection of several subsystems. Each subsystem is monitored by its own Local Fault Diagnoser, based on a set of local estimators. A consensus-based protocol is used to improve the detectability and the isolability of faults affecting variables shared among different subsystems because of the overlapping decomposition. A sufficient condition assuring the convergence of the estimation errors is derived. Time-varying threshold functions guaranteeing no false-positive alarms and theoretical results containing detectability and isolability conditions
Fault Detection and Isolation of the Wind Turbine Benchmark: an Estimation-based Approach
Zhang, Xiaodong,
Zhang, Qi,
Zhao, Songling,
Ferrari, Riccardo M.G.,
Polycarpou, Marios M.,
and Parisini, Thomas
In IFAC World Congress
2011
Distributed fault diagnosis of large-scale discrete-time nonlinear systems: New results on the isolation problem
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In Conference on Decision and Control
2010
This paper deals with the problem of designing a distributed fault detection and isolation methodology for nonlinear discrete-time dynamical systems that are modelled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state components. For each subsystem, a Local Fault Diagnoser is designed, based on the measured local state of the subsystem as well as the transmitted variables of neighboring states that define the subsystem interconnections. The local diagnostic decision is made on the basis of the knowledge of the local subsystem dynamic model and of an adaptive approximation of the interconnection with neighboring subsystems. A Global Fault Diagnoser is introduced to isolate distributed faults that cannot be isolated locally. The use of a specially-designed consensus-based estimator is proposed in order to improve the detectability and isolability of faults affecting variables shared among different subsystems. Simulation results are utilized to illustrate of the effectiveness of the proposed distributed fault diagnosis methodology.
A robust fault detection and isolation scheme for a class of uncertain input-output discrete-time nonlinear systems
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In American Control Conference
2008
This paper extends very recent results on discrete- time nonlinear fault detection and isolation to the case of discrete-time nonlinear systems with unstructured modeling uncertainty and partial state measurement. The fault diagnosis architecture consists of a fault detection and approximation estimator and a bank of fault isolation estimators, each corresponding to a particular type of fault. A time-varying threshold that guarantees no false-positive alarms and fault detectability conditions are derived analytically. For the fault isolation scheme, we design adaptive residual thresholds associated with each isolation estimator and obtain sufficient conditions for fault isolability. To illustrate the theoretical results, a simulation example based on a input-output discrete-time version of the three-tank benchmark problem is presented.
A fault detection and isolation scheme for nonlinear uncertain discrete-time sytems
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In Conference on Decision and Control
2007
This paper presents a fault detection and isolation scheme for abrupt and incipient faults in nonlinear uncertain discrete-time systems. The proposed fault diagnosis architecture consists of the fault detection and approximation estimator and a bank of fault isolation estimators, each corresponding to a particular type of fault. A time-varying threshold that guarantees no false-positive alarms and fault detectability conditions is derived analytically. For the fault isolation scheme, we design adaptive residual thresholds associated with each isolation estimator and obtain sufficient conditions for fault isolability. To illustrate the theoretical results, a simulation example based on a discrete-time version of the three-tank problem is presented.
Distributed fault diagnosis with overlapping decompositions and consensus filters
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In American Control Conference
2007
This paper deals with the problem of building a distributed fault detection architecture for large-scale dynamical systems that are modelled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state components. For each subsystem a local fault detector is built, such that it can measure the local state of its subsystem as well as receive through communication links a measure of the neighboring states. The local detection decision is made on the basis of the knowledge of the local subsystem dynamic model and of an adaptively learned approximation of the interconnection with neighbouring subsystems. The use of a specially-designed consensus filter is proposed in order to improve the capability of the diagnosers to detect faults affecting variables shared among different subsystems. Simulation results give an evidence of the effectiveness of the proposed fault detection scheme and of the use of overlapping decompositions and consensus filters.
A fault detection scheme for distributed nonlinear uncertain systems
Ferrari, Riccardo M.G.,
Parisini, Thomas,
and Polycarpou, Marios M.
In IEEE International Symposium on Intelligent Control
2006
This paper considers the problem of designing a fault detection scheme for a distributed nonlinear dynamic system. A network of distributed estimators is constructed where an adaptive estimator based on an on-line neural approximation model is embedded into each estimation agent. The local detection decision is made on the basis of the knowledge of the local dynamic model and on an on-line-learned approximation of the dynamic influence of the neighboring sub-systems. The stability of the adaptive estimation scheme is rigorously investigated and sufficient fault detectability conditions are also proposed. Simulation results are finally provided to demonstrate the effectiveness of the proposed architecture and methodology
Additional References
Economic analysis of condition monitoring systems for offshore wind turbine sub-systems
May, Allan,
McMillan, David,
and Thöns, Sebastian
IET Renewable Power Generation
2015