Lahouari Cheded

Unified approach to detection and isolation of parametric faults using a kalman filter residual-based approach

A unified method to detection and isolation of parametric faults in a physical system resulting from variations in the parameters of its constituting subsystems, termed herein as diagnostic parameters, uses Kalman filter residuals. Rather than using the feature vector made of the coefficients of the numerator and denominator of the system transfer function, which is known to be a non-linear function of the diagnostic parameter variations, the method first shows and then exploits, for fault detection purposes, the fact that the Kalman filter residual is a multi-linear function of the deviations in the diagnostic parameters, i.e. the residual is separately linear in each parameter. A fault is then isolated using a Bayesian multiple composite hypotheses testing approach. A reliable map relating the diagnostic parameters to the residual is obtained off-line using fault emulators. The unified fault detection and isolation method is successfully evaluated on both simulated data as well as on real data obtained from a benchmarked laboratory-scale coupled-tank system used to exemplify an industrial two-tank process.

Channel equalization using simplified least mean-fourth algorithm

The steady flow of new research results and developments in the field of adaptive equalization that was witnessed for at least the last four decades is clearly evidenced by the many footprints of success it left behind and shows no sign of ending. The thrust of research and implementation in this field is mainly powered by the use of the well-known mean-square cost function upon which relies the ubiquitous least-mean square (LMS) algorithm. However, such an algorithm is well-known to lead to sub-optimal solutions in the real world that is largely dominated by non-Gaussian interference signals. The use of a non-mean-square cost function would successfully tackle these types of interference signals but would invariably involve a higher computational cost. To address these important practical issues, this paper proposes a new adaptive equalization technique that combines both the least-mean-fourth (LMF) algorithm, which is governed by a non-mean-square cost function, with a power-of-two quantizer (PTQ) in the coefficient update process, which greatly reduces the computational cost involved and which therefore makes the proposed technique applicable to time-varying environments. This paper not only elaborates on the basic idea behind the proposed technique but also defines the necessary assumptions and provides a thorough statistical performance analysis (including a study of the convergence behavior) of the combined algorithm LMF-PTQ that is at the core of the proposed technique. An extensive simulation work was carried out and showed that the theoretical predictions are very well substantiated.

The exact impact of amplitude quantization on multi-dimensional, high-order moments estimation

Abstract In this paper, we study the exact impact of amplitude quantization on the process of estimating a multi-dimensional ( m -D) higher-order moment using quantized signal processing schemes. First, we derive the relationship between an m -D, finite-order unquantized moment and its quantized counterpart and express it as the sum of two key terms, one of which is wanted as it embodies the desired information, namely the m -D unquantized moment to be estimated, in addition to some other quantity that gives an exact measure of the unavoidable component of the impact of amplitude quantization. As for the second term, it wholly represents the other unwanted component of the impact under consideration that can, at least in theory, be exactly computed given the input statistics or avoided if certain specific conditions are satisfied by the input statistics. The issue of avoidance of this unwanted component will not be discussed here as it is reported elsewhere as referenced in the text. The m -D results relevant to the impact of amplitude quantization in the 1-D and 2-D cases are then derived and a complete statistical characterization of the quantization noise in these 2 cases is then obtained using simple and intuitive arguments. The theoretical results obtained in these 2 cases are applied to the important class of Gaussian signals, leading to a new interpretation of a Gaussian-driven quantized channel. Finally, numerous simulation tests have been carried out and have resulted in a very good support of the theoretical predictions.

An exact FFT recovery theory: a nonsubtractive dither quantization approach with applications

Fourier transform is undoubtedly one of the cornerstones of digital signal processing (DSP). The introduction of the now famous FFT algorithm has not only breathed a new lease of life into an otherwise latent classical DFT algorithm, but also led to an explosion in applications that have now far transcended the confines of the DSP field. For a good accuracy, the digital implementation of the FFT requires that the input and/or the 2 basis functions be finely quantized. This paper exploits the use of coarse quantization of the FFT signals with a view to further improving the FFT computational efficiency while preserving its computational accuracy and simplifying its architecture. In order to resolve this apparent conflict between preserving an excellent computational accuracy while using a quantization scheme as coarse as can be desired, this paper advances new theoretical results which form the basis for two new and practically attractive FFT estimators that rely on the principle of 1 bit nonsubtractive dithered quantization (NSDQ). The proposed theory is very well substantiated by the extensive simulation work carried out in both noise-free and noisy environments. This makes the prospect of implementing the 2 proposed 1 bit FFT estimators on a chip both practically attractive and rewarding and certainly worthy of a further pursuit.

Measurement error sensitivity analysis for detecting and locating leak in pipeline using ANN and SVM

This paper presents an approach for detecting, locating and estimating the size of leak in a pipeline using pressure sensors, differential pressure sensors and flow-rate sensors. To overcome the problem with existing approaches we use differential pressure sensors that detect small change in pressure in order to detect small change in leak size. The pipeline system is modeled and simulated in EPANET software, and the input-output data acquired from it (i.e. sensor measurements and the leak locations and sizes) are used in MATLAB and DTREG software to develop Artificial Neural Network (ANN) and Support Vector Machines (SVM) models. Comparison of results shows that SVM is less sensitive and more stable to noise increment than ANN. However the performance of ANN is better with very small noises.

Non-stationary vibration signal analysis of rotating machinery via time-frequency and wavelet techniques

This paper introduces an efficient and powerful approach to fault detection in rotating machinery using time-frequency analysis based on both Fourier and wavelet transforms of the monitored vibration signal. Time-frequency techniques are powerful tools for analyzing transient information in vibration signature for both condition monitoring and fault detection purposes. Our work on fault detection reported in this paper is two-fold: (1) application of the short-time Fourier transforms (STFT) and the exploitation of the spectrogram-based time-frequency distribution to detect various mechanical faults during the start-up & coast down phases in rotating machinery and (2) application of a novel wavelet-based technique combining both the continuous wavelet and the wavelet packet transforms. This novel technique exploits the use of the modulus of the local maxima lines in the wavelet domain, to detect impulsive mechanical faults such as impact blade-to-stator rubbing in turbo machinery. Both the analysis and the extensive simulation work carried out here show in particular the superiority of our proposed combined wavelet-based approach over the traditional Fourier Transform (FFT) method, in reliably diagnosing impulsive mechanical faults and start-up and cost down signals.

A novel energy-aware approach for locating leaks in water pipeline using a wireless sensor network and noisy pressure sensor data

We propose a novel energy-aware approach to detect a leak and estimate its size and location in a noisy water pipeline using least-squares and various pressure measurements in the pipeline network. The novelty in our work hinges on the fusion of the duty-cycling (DC) and data-driven (DD) strategies, both well-known techniques for energy reduction in a wireless sensor network (WSN). To maximize the information gain and minimize the energy consumed by the WSN, we first study the effects of (a) various levels of sensor measurement uncertainty and (b) the use of the smallest possible number of pressure sensors on the overall accuracy of our approach. Using the DD strategy only, a noisy environment, and a small number of sensors, the performance of our scheme shows that, for small leak sizes, the estimation error in both leak location and size becomes unacceptably high. Next, using as few sensors as possible for an acceptable accuracy, we fused the DD strategy with the DC one to minimize the sensing, processing, and communication energies. The fusion approach yielded a better performance with significant energy saving, even in noisy environments. EPANET was used to model the pipeline network and leak and MATLAB to implement, analyze, and evaluate our fusion approach.

Vibration monitoring and faults detection using wavelet techniques

This paper introduces an efficient approach for fault detection in rotating machinery by analyzing its vibration signals using wavelet techniques. Specifically our approach uses the wavelet packet transform (WPT) to decompose the vibration signals in the wavelet packet space, in order to reveal the transient information in these signals. Faults are efficiently detected by exploiting the mean values of the energy in the detail signals. The wavelet-based approach is also compared with the traditional Fourier-based one. Both analysis and an extensive simulation of the two approaches clearly show the superiority of the WPT-based approach over the Fourier-based one, in efficiently diagnosing faults from vibration signals.

Robust Control of a Closed-Loop Identified System with Parametric/Model Uncertainties and External Disturbances

This paper deals with the closed-loop identification of a two-tank process used in industry. The identified model is then utilized to develop robust controllers i.e. H E; and sliding mode controllers. It is shown that these controllers guarantee a satisfactory performance in the face of both model/parametric uncertainties and external disturbances. The designed controllers have been successfully tested through extensive simulation. In addition, this paper shows that the designed robust controllers far outperform traditional controllers such as P, PI, and PID, in the face of parametric model uncertainties and the effects of external disturbances. The successful use of the designed robust controllers encourages their extension to other physical systems.

Adaptive channel equalization: A simplified approach using the quantized-LMF algorithm

In digital communication, adaptive channel equalization techniques underpin the successful provision of high speed and reliable data transmission over severely-dispersive channels, e.g. wireless and mobile ones. In a real world that is largely dominated by non- Gaussian interference signals, adaptive equalizers relying heavily on the LMS are bound to yield suboptimal performances. This work addresses this sub-optimality issue by proposing a new adaptive equalizer which judiciously combines the power of the least- mean fourth (IMF) algorithm to better tackle non-Gaussian environments, and the capability of the power-of-two quantizer (PTQ) to greatly reduce the IMFs high computational load. This combination endows the proposed algorithm with a capability of tracking fast-changing channels. A performance analysis of the proposed adaptive channel equalizer, based on a new linear approximation of the PTQ, is also presented. Extensive simulation testing of the proposed adaptive equalizer corroborates very well the theoretical findings predicted by the analysis of the linearized LMF- PTQ algorithm.