An improved PCA method with application to boiler leak detection

Principal component analysis (PCA) is a popular fault detection technique. It has been widely used in process industries, especially in the chemical industry. In industrial applications, achieving a sensitive system capable of detecting incipient faults, which maintains the false alarm rate to a minimum, is a crucial issue. Although a lot of research has been focused on these issues for PCA-based fault detection and diagnosis methods, sensitivity of the fault detection scheme versus false alarm rate continues to be an important issue. In this paper, an improved PCA method is proposed to address this problem. In this method, a new data preprocessing scheme and a new fault detection scheme designed for Hotelling's T2 as well as the squared prediction error are developed. A dynamic PCA model is also developed for boiler leak detection. This new method is applied to boiler water/steam leak detection with real data from Syncrude Canada's utility plant in Fort McMurray, Canada. Our results demonstrate that the proposed method can effectively reduce false alarm rate, provide effective and correct leak alarms, and give early warning to operators.     

Sliding mode observer based incipient sensor fault detection with application to high-speed railway traction device

This paper considers incipient sensor fault detection issue for a class of nonlinear systems with “observer unmatched” uncertainties. A particular fault detection sliding mode observer is designed for the augmented system formed by the original system and incipient sensor faults. The designed parameters are obtained using LMI and line filter techniques to guarantee that the generated residuals are robust to uncertainties and that sliding motion is not destroyed by faults. Then, three levels of novel adaptive thresholds are proposed based on the reduced order sliding mode dynamics, which effectively improve incipient sensor faults detectability. Case study of on the traction system in China Railway High-speed is presented to demonstrate the effectiveness of the proposed incipient senor faults detection schemes.     

Adaptive PCA based fault diagnosis scheme in imperial smelting process

In this paper, an adaptive fault detection scheme based on a recursive principal component analysis (PCA) is proposed to deal with the problem of false alarm due to normal process changes in real process. Our further study is also dedicated to develop a fault isolation approach based on Generalized Likelihood Ratio (GLR) test and Singular Value Decomposition (SVD) which is one of general techniques of PCA, on which the off-set and scaling fault can be easily isolated with explicit off-set fault direction and scaling fault classification. The identification of off-set and scaling fault is also applied. The complete scheme of PCA-based fault diagnosis procedure is proposed. The proposed scheme is first applied to Imperial Smelting Process, and the results show that the proposed strategies can be able to mitigate false alarms and isolate faults efficiently.     

基于支持向量分类器PCA方法及其在过程监控和故障检测中的应用

提出了基于支持向量分类器对过程进行性能监控和故障检测的改进 PCA方法 ,该方法避免了多元统计过程控制(MSPC)假设主元必须服从正态分布的前提。此外 ,通过对苯 -甲苯两组分精馏分离过程的仿真研究表明 ,该方法是有效的 ,并具有比传统多元统计过程控制更为优越的性能。     

An improved simplex-based adaptive evolutionary digital filter and its application for fault detection of rolling element bearings

The de-noising performance and convergence behavior of the adaptive evolutionary digital filter (EDF) are restricted by the factors of constant evolutionary coefficients and taking the reciprocal of average energy of residual signal as the fitness function. In this paper, an improved adaptive evolutionary digital filter based on the simplex method (EDF-SM) is proposed to overcome the shortcomings of the original EDF. A new evolutionary rule was constructed by introducing the simplex-based mutating method and by then combining this with the original cloning and mating methods. The reciprocal of sample entropy was taken as the fitness function and variable evolutionary coefficients were employed. Numerical examples show that the proposed EDF-SM exhibits a higher convergence rate and a better de-noising behavior than the other EDFs. The effectiveness of the proposed method in discovering fault characteristics and detecting faults of rolling element bearings is supported using an experimental test.     

An adaptive SK technique and its application for fault detection of rolling element bearings

In this paper, we propose an adaptive spectral kurtosis (SK) technique for the fault detection of rolling element bearings. The primary contribution is adaptive determination of the bandwidth and center frequency. This is implemented with successive attempts to right-expand a given window along the frequency axis by merging it with its subsequent neighboring windows. Influence of the parameters such as the initial window function, bandwidth and window overlap on the merged windows as well as how to choose those parameters in practical applications are explored. Based on simulated experiments, it can be found that the proposed technique can further enhance the SK-based method as compared to the kurtogram approach. The effectiveness of the proposed method in fault detection of the rolling element bearings is validated using experimental signals.     

Isolability of faults in sensor fault diagnosis

A major concern with fault detection and isolation (FDI) methods is their robustness with respect to noise and modeling uncertainties. With this in mind, several approaches have been proposed to minimize the vulnerability of FDI methods to these uncertainties. But, apart from the algorithm used, there is a theoretical limit on the minimum effect of noise on detectability and isolability. This limit has been quantified in this paper for the problem of sensor fault diagnosis based on direct redundancies. In this study, first a geometric approach to sensor fault detection is proposed. The sensor fault is isolated based on the direction of residuals found from a residual generator. This residual generator can be constructed from an input-output or a Principal Component Analysis (PCA) based model. The simplicity of this technique, compared to the existing methods of sensor fault diagnosis, allows for more rational formulation of the isolability concepts in linear systems. Using this residual generator and the assumption of Gaussian noise, the effect of noise on isolability is studied, and the minimum magnitude of isolable fault in each sensor is found based on the distribution of noise in the measurement system. Finally, some numerical examples are presented to clarify this approach.     

Experimental analysis on fault detection for a direct coupled rotor-bearing system

Rotating machinery is becoming faster and lightweight due to the advanced technologies made in engineering and materials sciences. It is required them to run for longer periods of time. All of these factors mean that the detection, location and analysis of faults play a vital role in highly reliable operations. Using vibration analysis, the condition of a machine can be periodically monitored. In this study, dynamic behavior of a direct coupled rotor-bearing system is investigated. Experimental vibration analyses in the vertical direction of the system are implemented. Vibration monitoring with trend analysis and spectrum graphs are employed to diagnose the excessive vibration source(s). It is seen that the rotating machineries can have one or more vibration sources. The vibration values obtained from each bearing show that the main excessive vibration sources in the system stem from mechanical looseness and misalignment.     

使用PCA及SVM算法进行滚动轴承故障检测

本文使用PCA（主元分析）算法对滚动轴承振动信号数据进行预处理,这可降低数据维数和提取数据特征信息;将预处理后数据作为SVM（支持向量机）算法的输入,通过SVM算法来检测轴承故障。     

Nonstationary multiscale turbulence simulation based on local PCA

Turbulence simulation methods are of fundamental importance for evaluating the performance of control strategies for Adaptive Optics (AO) systems. In order to obtain a reliable evaluation of the performance a statistically accurate turbulence simulation method has to be used. This work generalizes a previously proposed method for turbulence simulation based on the use of a multiscale stochastic model. The main contributions of this work are: first, a multiresolution local PCA representation is considered. In typical operating conditions, the computational load for turbulence simulation is reduced approximately by a factor of 4, with respect to the previously proposed method, by means of this PCA representation. Second, thanks to a different low resolution method, based on a moving average model, the wind velocity can be in any direction (not necessarily that of the spatial axes). Finally, this paper extends the simulation procedure to generate, if needed, turbulence samples by using a more general model than that of the frozen flow hypothesis.     

Hierarchical adaptive deep convolution neural network and its application to bearing fault diagnosis

Traditional artificial methods and intelligence-based methods of classifying and diagnosing various mechanical faults with high accuracy by extracting effective features from vibration data, such as support vector machines and back propagation neural networks, have been widely investigated. However, the problems of extracting features automatically without significantly increasing the demand for machinery expertise and maximizing accuracy without overcomplicating machine structure have to date remained unsolved. Therefore, a novel hierarchical learning rate adaptive deep convolution neural network based on an improved algorithm was proposed in this study, and its use to diagnose bearing faults and determine their severity was investigated. To test the effectiveness of the proposed method, an experiment was conducted with bearing-fault data samples obtained from a test rig. The method achieved a satisfactory performance in terms of both fault-pattern recognition and fault-size evaluation. In addition, comparison revealed that the improved algorithm is well suited to the fault-diagnosis model, and that the proposed method is superior to other existing methods.     

Additive fault detection in nonlinear dynamic systems with saturation

This paper describes the effects of input saturation on the performance of a model-based fault detection method based on the input-output parity equation approach. For this purpose, the level control of a chemical reactor has been chosen as the control process to be analyzed, where the saturation of the dynamic process is due to the inflow control valve, and only additive faults have been considered. This study has been carried out in two ways: first by simulation techniques and second on a real industrial system. In the simulated case, the decrease in the fault detectability due to the saturation effects is shown, and some ways of achieving higher fault detectability are explored. The results obtained in the industrial case complement those obtained in the simulated case, and also reveal the existence of a relation between the control strategy used in the process and additive fault detectability, in the sense that increases in fault detectability are obtained due to the use of faster control strategies.     

A new adaptive cascaded stochastic resonance method for impact features extraction in gear fault diagnosis

Gearboxes are widely used in engineering machinery, but tough operation environments often make them subject to failure. And the emergence of periodic impact components is generally associated with gear failure in vibration analysis. However, effective extraction of weak impact features submerged in strong noise has remained a major challenge. Therefore, the paper presents a new adaptive cascaded stochastic resonance (SR) method for impact features extraction in gear fault diagnosis. Through the multi-filtered procession of cascaded SR, the weak impact features can be further enhanced to be more evident in the time domain. By analyzing the characteristics of non-dimensional index for impact signal detection, new measurement indexes are constructed, and can further promote the extraction capability of SR for impact features by combining the data segmentation algorithm via sliding window. Simulation and application have confirmed the effectiveness and superiority of the proposed method in gear fault diagnosis.     

Islanding detection scheme based on adaptive identifier signal estimation method

This paper proposes a novel, passive-based anti-islanding method for both inverter and synchronous machine-based distributed generation (DG) units. Unfortunately, when the active/reactive power mismatches are near to zero, majority of the passive anti-islanding methods cannot detect the islanding situation, correctly. This study introduces a new islanding detection method based on exponentially damped signal estimation method. The proposed method uses adaptive identifier method for estimating of the frequency deviation of the point of common coupling (PCC) link as a target signal that can detect the islanding condition with near-zero active power imbalance. Main advantage of the adaptive identifier method over other signal estimation methods is its small sampling window. In this paper, the adaptive identifier based islanding detection method introduces a new detection index entitled decision signal by estimating of oscillation frequency of the PCC frequency and can detect islanding conditions, properly. In islanding conditions, oscillations frequency of PCC frequency reach to zero, thus threshold setting for decision signal is not a tedious job. The non-islanding transient events, which can cause a significant deviation in the PCC frequency are considered in simulations. These events include different types of faults, load changes, capacitor bank switching, and motor starting. Further, for islanding events, the capability of the proposed islanding detection method is verified by near-to-zero active power mismatches.     

Robust fault detection of turbofan engines subject to adaptive controllers via a Total Measurable Fault Information Residual (ToMFIR) technique

This paper provides a new design of robust fault detection for turbofan engines with adaptive controllers. The critical issue is that the adaptive controllers can depress the faulty effects such that the actual system outputs remain the pre-specified values, making it difficult to detect faults/failures. To solve this problem, a Total Measurable Fault Information Residual (ToMFIR) technique with the aid of system transformation is adopted to detect faults in turbofan engines with adaptive controllers. This design is a ToMFIR-redundancy-based robust fault detection. The ToMFIR is first introduced and existing results are also summarized. The Detailed design process of the ToMFIRs is presented and a turbofan engine model is simulated to verify the effectiveness of the proposed ToMFIR-based fault-detection strategy.     

An adaptive stochastic resonance method based on grey wolf optimizer algorithm and its application to machinery fault diagnosis

Stochastic resonance (SR) is widely used as an enhanced signal detection method in machinery fault diagnosis. However, the system parameters have significant effects on the output results, which makes it difficult for SR method to achieve satisfactory analysis results. To solve this problem and improve the performance of SR method, this paper proposes an adaptive SR method based on grey wolf optimizer (GWO) algorithm for machinery fault diagnosis. Firstly, the SR system parameters are optimized by the GWO algorithm using a redefined signal-to-noise ratio (SNR) as optimization objective function. Then, the optimal SR output matching the input signal can be adaptively obtained using the optimized parameters. The proposed method is validated on a simulated signal detection and a rolling element bearing test bench, and then applied to the gear fault diagnosis of electric locomotive. Compared with the conventional fixed-parameter SR method, the adaptive SR method based on genetic algorithm (GA-SR) as well as the well-known fast kurtogram method, the proposed method can achieve a greater accuracy. The results indicated that the proposed method has great practical values in engineering.     

A data mining approach for fault diagnosis: An application of anomaly detection algorithm

Rolling-element bearing failures are the most frequent problems in rotating machinery, which can be catastrophic and cause major downtime. Hence, providing advance failure warning and precise fault detection in such components are pivotal and cost-effective. The vast majority of past research has focused on signal processing and spectral analysis for fault diagnostics in rotating components. In this study, a data mining approach using a machine learning technique called anomaly detection (AD) is presented. This method employs classification techniques to discriminate between defect examples. Two features, kurtosis and Non-Gaussianity Score (NGS), are extracted to develop anomaly detection algorithms. The performance of the developed algorithms was examined through real data from a test to failure bearing. Finally, the application of anomaly detection is compared with one of the popular methods called Support Vector Machine (SVM) to investigate the sensitivity and accuracy of this approach and its ability to detect the anomalies in early stages.     

An advanced Park's vectors approach for bearing fault detection

Induction motors vibrations, caused by bearing defects, result in the modulation of the stator current. In this research, a method based on Park's vector approach for bearing fault detection using three-phase stator current analysis is presented. In order to evaluate the ability of the proposed method several experiments are performed, and sets of data are gathered before and after using defective bearings. Both localized and distributed defects are evaluated using this method. The experimental results from our study suggest that the proposed method provides a powerful and general approach to incipient fault detection.