Data-driven Fault Detection Research Articles

Data-driven design of robust fault detection system for wind turbines

In this paper, a robust data-driven fault detection approach is proposed with application to a wind turbine benchmark. The main challenges of the wind turbine fault detection lie in its nonlinearity, unknown disturbances as well as significant measurement noise. To overcome these difficulties, a data-driven fault detection scheme is proposed with robust residual generators directly constructed from available process data. A performance index and an optimization criterion are proposed to achieve the robustness of the residual signals related to the disturbances. For the residual evaluation, a proper evaluation approach as well as a suitable decision logic is given to make a correct final decision. The effectiveness of the proposed approach is finally illustrated by simulations on the wind turbine benchmark model.

Cold Start Approach for Data-Driven Fault Detection

A typical assumption in supervised fault detection is that abundant historical data are available prior to model learning, where all types of faults have already been observed at least once. This assumption is likely to be violated in practical settings as new fault types can emerge over time. In this paper we study this often overlooked cold start learning problem in data-driven fault detection, where in the beginning only normal operation data are available and faulty operation data become available as the faults occur. We explored how to leverage strengths of unsupervised and supervised approaches to build a model capable of detecting faults even if none are still observed, and of improving over time, as new fault types are observed. The proposed framework was evaluated on the benchmark Tennessee Eastman Process data. The proposed fusion model performed better on both unseen and seen faults than the stand-alone unsupervised and supervised models.

Electronic Returnless Fuel System Fault Diagnosis and Isolation: A Data-Driven Approach


 
 
 The Electronic Return-less Fuel System (ERFS) manages the delivery of fuel from the fuel tank to the engine. The pressure in the fuel line is electronically controlled by the fuel system control module by speeding up or slowing down the fuel pump. This allows the system to efficiently control the amount of fuel provided to the engine when compared to vehicles equipped with a standard fuel system wherein the fuel pump continuously runs at full speed. A failure in the fuel system that impacts the ability to deliver fuel to the engine will have an immediate effect on system performance. Consequently, improved reliability and availability, and reduction in the number of walk-home situations require efficient fault detection, isolation and prognosis of the ERFS system. This paper develops and implements data-driven fault detection, isolation and severity estimation algorithms for the ERFS. The HIL Fuel System Rig and a Chevrolet Silverado truck were used to collect and analyze the fuel system behavior under different fault conditions. Several data-driven classifiers, such as support vector machines, K- nearest Neighbor, Discriminant analysis, Bayes classifier, Partial- least squares, Quadratic and Linear classifiers, were implemented on a limited set of data for both training and testing. Regression techniques, such as Partial least squares regression and Principle component regression, are used to estimate the severity of faults. The resulting solution approach has the potential to be applicable to a wide variety of systems, ranging from automobiles to aerospace systems.
 
 

Design of an active fault tolerant control system for a simulated industrial steam turbine

An active fault tolerant control (FTC) scheme is proposed in this paper to accommodate for an industrial steam turbine faults based on integration of a data-driven fault detection and diagnosis (FDD) module and an adaptive generalized predictive control (GPC) approach. The FDD module uses a fusion-based methodology to incorporate a multi-attribute feature via a support vector machine (SVM) and adaptive neuro-fuzzy inference system (ANFIS) classifiers. In the GPC formulation, an adaptive configuration of its internal model has been devised to capture the faulty model for the set of internal steam turbine faults. To handle the most challenging faults, however, the GPC control configuration is modified via its weighting factors to demand for satisfactory control recovery with less vigorous control actions. The proposed FTC scheme is hence able to systematically maintain early FDD with efficient fault accommodation against faults jeopardizing the steam turbine availability. Extensive simulation tests are conducted to explore the effectiveness of the proposed FTC performances in response to different categories of steam turbine fault scenarios.

Model-Based and Data-Driven Fault Detection Performance for a Small UAV

Fault detection and identification algorithms may rely on knowledge of underlying system dynamics while some eschew this modeling in favor of data-driven anomaly detection. This paper considers model-based residual generation and data-driven anomaly detection for a small, low-cost unmanned aerial vehicle using both types of approaches and applies those algorithms to experimental faulted and unfaulted flight-test data. The model-based fault detection strategy uses robust linear filtering methods to reject exogenous disturbances, e.g., wind, and provide robustness to model errors. The data-driven algorithm is developed to operate exclusively on raw flight-test data without detailed system knowledge. The detection performance of these complementary, but different, methods is compared.

Data‐driven subspace‐based adaptive fault detection for solar power generation systems

Data-driven fault detection has emerged as one of the most prevalent topics in the fault diagnosis. In this study, a novel data-driven subspace-based fault-detection scheme is proposed to handle the problem of fault detection with system uncertainties in solar power generation systems. A data-driven subspace-based predictor is developed by using the input–output measurements. The residual signal is generated from the predictive error of the predictor and a fault-detection filter that is designed to diminish the influence of system uncertainties. An adaptive algorithm is developed for updating the fault-detection filter. Faults can be detected by comparing the evaluated residual signal with a threshold. The reliability of the designed fault-detection scheme is verified in three cases in a solar power generation system.

Robust Fault Detection with missing data via Sparse Decomposition

Data-driven fault diagnosis doesn't depend upon the precise model of an industrial process, however, the quality of the process data set has an important effect on its result. As a quite common phenomenon, the process data sets with missing data usually reduce the performance of data based algorithms. In this paper, a novel data-driven fault detection method based on sparse decomposition is proposed to deal with the issue of missing data. In our approach, the K-SVD and OMP algorithms are used to learn the sparse representation model of the training data and to conduct the sparse decomposition of the testing data, respectively. Compared to the traditional fault detection method, PC A, our sparse decomposition based method outperforms PCA with missing data. At last, simulation experiments verify our result.

Multi-sensor information fusion for fault detection in aircraft gas turbine engines

The article addresses data-driven fault detection in commercial aircraft gas turbine engines in the framework of multi-sensor information fusion and symbolic dynamic filtering. The hierarchical decision and control structure, adopted in this article, involves construction of composite patterns, namely, atomic patterns extracted from single sensors, and relational patterns representing cross-dependence between a pair of sensors. While the underlying theories are presented along with necessary assumptions, the proposed method is validated on the NASA C-MAPSS simulation test bed of aircraft gas turbine engines; both single-fault and multiple-fault scenarios have been investigated. Since aircraft engines undergo natural degradation during the course of their normal operation, the issue of distinguishing between a fault and natural degradation is also addressed.

Survey on data-driven industrial process monitoring and diagnosis

This paper provides a state-of-the-art review of the methods and applications of data-driven fault detection and diagnosis that have been developed over the last two decades. The scope of the problem is described with reference to the scale and complexity of industrial process operations, where multi-level hierarchical optimization and control are necessary for efficient operation, but are also prone to hard failure and soft operational faults that lead to economic losses. Commonly used multivariate statistical tools are introduced to characterize normal variations and detect abnormal changes. Further, diagnosis methods are surveyed and analyzed, with fault detectability and fault identifiability for rigorous analysis. Challenges, opportunities, and extensions are summarized with the intent to draw attention from the systems and control community and the process control community.

Data-Driven Fault Detection in Aircraft Engines With Noisy Sensor Measurements

An inherent difficulty in sensor-data-driven fault detection is that the detection performance could be drastically reduced under sensor degradation (e.g., drift and noise). Complementary to traditional model-based techniques for fault detection, this paper proposes symbolic dynamic filtering by optimally partitioning the time series data of sensor observation. The objective here is to mask the effects of sensor noise level variation and magnify the system fault signatures. In this regard, the concepts of feature extraction and pattern classification are used for fault detection in aircraft gas turbine engines. The proposed methodology of data-driven fault detection is tested and validated on the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) test-bed developed by NASA for noisy (i.e., increased variance) sensor signals.

Data Driven Fault Detection and Isolation of a Wind Turbine Benchmark

AbstractThis paper investigates data-driven fault detection and isolation (FDI) designs for a wind turbine benchmark problem. The benchmark is described by a SimuLink model, which contains nonlinear lookup tables and unknown wind disturbances. Based on classical FDI design methods, a linearization of the SimuLink model into a standard state-space form and describing the linearization errors as perturbations may be necessary. To avoid these difficult modeling procedures, this paper applies a data-driven design method, which produces FDI filters directly based on the simulated data from the benchmark SimuLink model. The fixed-value sensor faults therein are especially targeted. Moreover, we develop in this paper a new data-driven fault isolation scheme, via exploiting hardware redundancy in a plant. Based on this, a bank of robust data-driven detection filters are designed for the benchmark and implemented in parallel. The simulation results show the effectiveness of the applied data-driven scheme.

Total PLS Based Contribution Plots for Fault Diagnosis

摘要: 多变量统计过程监控对于复杂工业过程是一种有效的故障检测和诊断技术. 最小二乘(或称潜空间投影)模型是多变量统计过程监控中常用的一种投影模型, 能够同时对过程数据和质量数据进行建模. 讨论了一种新的基于全潜空间投影模型的故障诊断技术. 全潜空间投影模型中有4个检测统计量. 提出了一种新的T2贡献图计算方法, 对于所有检测统计量, 得到了相应的贡献图算法. 为了确定一个变量是否发生了故障, 计算所有变量贡献图的控制限. 该技术可以将辨识到的故障变量分为与Y有关和与Y无关的两类. 基于Tennessee Eastman过程的案例研究表明了该技术的有效性. 关键词: 数据驱动 / 全潜空间投影 / 贡献图 / 故障诊断

Total PLS Based Contribution Plots for Fault Diagnosis

Multivariate statistical process monitoring (MSPM) is an efficient data-driven fault detection and diagnosis approach for complex industrial processes. Partial least squares or projection to latent structures (PLS) is one of the latent projection structures used in MSPM, which uses process data X and quality data Y together. In this paper, we discuss a new fault diagnosis approach based on total projection to latent structures (T-PLS). Four kinds of monitoring statistics are used in T-PLS, and a new definition of variable contributions to T 2 of PLS is proposed. Then, definitions of variable contributions to all statistics are derived to identify the faults. Control limits for contribution plots are calculated to identify whether a variable is in abnormal situation or not. Further, the proposed method separates the identified variables into faulty variables related to Y and unrelated to Y more clearly than conventional method. A case study on Tennessee Eastman process (TEP) indicates the efficiency of the proposed approach.

Data-driven Fault Detection and Diagnosis for Complex Industrial Processes

The objective of this paper is to provide a status report of the methods and applications of data-driven fault detection and diagnosis to the IFAC Safeprocess community. The scope of the problem is described with reference to the scale and complexity of industrial processes, where multi-level hierarchical optimization and control are a must for efficient operation, but are also prone to hard or soft faults that lead to economic losses. Commonly used multivariate statistical tools are introduced to characterize normal variations and detect abnormal changes. Further diagnosis methods are discussed next, with additional topics for rigorous analysis. An application case is presented, and future challenges and opportunities are speculated with the hope to draw more attention from the Safeprocess community.

Observer and Data-Driven-Model-Based Fault Detection in Power Plant Coal Mills

This paper presents and compares model-based and data-driven fault detection approaches for coal mill systems. The first approach detects faults with an optimal unknown input observer developed from a simplified energy balance model. Due to the time-consuming effort in developing a first principles model with motor power as the controlled variable, data-driven methods for fault detection are also investigated. Regression models that represent normal operating conditions (NOCs) are developed with both static and dynamic principal component analysis and partial least squares methods. The residual between process measurement and the NOC model prediction is used for fault detection. A hybrid approach, where a data-driven model is employed to derive an optimal unknown input observer, is also implemented. The three methods are evaluated with case studies on coal mill data, which includes a fault caused by a blocked inlet pipe. All three approaches detect the fault as it emerges. The optimal unknown input observer approach is most robust, in that, it has no false positives. On the other hand, the data-driven approaches are more straightforward to implement, since they just require the selection of appropriate confidence limit to avoid false detection. The proposed hybrid approach is promising for systems where a first principles model is cumbersome to obtain.

Simulation of coal pyrolysis in plasma jet by CPD model

Data-driven fault detection technique has exhibited its wide applications in industrial process monitoring. However, how to extract the local and non-Gaussian features effectively is still an open problem. In this paper, statistics locality preserving projections (SLPP) is proposed to extract the local and non-Gaussian features. Firstly, statistics pattern analysis (SPA) is applied to construct process statistics and grasp the non-Gaussian statistical property using high order statistics. Then, locality preserving projections (LPP) method is used to discover local manifold structure of the statistics. In essence, LPP tries to map the close points in the original space to close in the low-dimensional space. Lastly, T2 and squared prediction error (SPE) charts of SLPP model are used to detect process faults. One simple simulated system and the Tennessee Eastman process show that the proposed SLPP method is more effective than principal component analysis, LPP and statistics principal component analysis in fault detection performance.