Browsing by Author "Aiordachioaie, Dorel"

Time-frequency image processing is considered in the context of change detection and diagnosis purposes based on signal processing paradigm. A method for selection and extraction of features from time-frequency is considered and evaluated. New images are obtained by applying a criterion based on the contours generated by the main components of the analyzed time-frequency image. The transformed images are less complex, and could be white and black only. Features based on statistical moments are considered, selected and used to define discriminant functions, in order to improve the results of the classification. The features include the number of the contours, the average area defined by the contours, the variance of the areas and the Renyi entropies. As case study, signals coming from vibration generated by faults in bearings are considered.

The research domain of the paper is the time-frequency image processing. Firstly, a comparison among the features selection and extraction methods from time-frequency images of vibration signals in bearings with faults is made. Both time and frequency methods are considered and discussed, from classification performance point of view. A new method of feature selection based on the decomposition of time-frequency images in sub-bands is introduced. For each sub-band, discrete cosine transform is applied. The coefficients of the transform are features which must be processed for classification task. Distance based classifier is considered with vectors of features of various lengths. Computer based experiments are conducted with real data from a benchmark database with vibration signals generated by various type and size of faults. The results are very encouraging and show the feasibility of the method.

This exploratory work looks on time series complexity, for direct applications in change detection of the structure and properties of the signals. The objective of the paper was to compare the results of the evaluation of the complexity of time series data, obtained with measures based on entropy and data complexity indices. As test signals, determinist, random and chaotic signals are considered, in an independent and mixed probabilistic approach. Complexity descriptors are based on entropies, as Renyi, Tsallis, Multiscale technique, and two data complexity indices, as Lempel-Ziv complexity and Lyapunov exponent. High values of complexity measures are expected for all cases where random or chaotic components are dominant, i.e. greater amplitudes than the determinist components. The complexity measures are evaluated in terms of monotony, sensitivity at the length of time series, and change detection capability of the structure of the analyzed signal. The results of the experiments based on computer-based simulations are presented with fuzzy labels and show a variety of results, i.e. good results for some cases and small for others. Such results suggest an aggregate criterion for change detection, with at least two terms, one based on entropy and another one on complexity indices.

Change detection and diagnosis are important research directions and activities in the field of system engineering and conditional maintenance of equipments and industrial processes. The paper promotes a new method for change detection and optimal segmentation of vibrating data obtained during operation of rolling element bearings (REB). After a description of the bearing faults and dynamic simulation of REB, the paper makes a review of the change detection and segmentation approaches, that could be used in REB fault detection and diagnosis. A new approach for change detection and optimal segmentation of vibrating signals, aiming to determine the change points in signals generated by the faults, produced during REB operating, is presented; the efficiency of the segmentation method is proven using Monte Carlo simulations for different signal models, including models with changes in the mean, in FIR, and AR model parameters, frequently used in processing vibrating signals. In the final part, the paper analyses some experimental results obtained using this approach and data from the Case Western Reserve University Bearing Data Center.

The objective of the paper refers to change detection in the structure of time series data, with direct application in fault detection of industrial processes and their components. The detection criterion is based on a combination of multiscale analysis technique with Renyi entropy on each scale, followed by a cumulative sum evaluation to estimate the point of change. Two case studies are considered. One considers a synthetized mixed signal, by mixing a deterministic signal with one having random components. The other one uses real vibration signals generated by some faults in the bearings of rotating machines. The method generates different values of the entropy for different structures of the time series and allows change detection of the incipient faults in bearings, i.e. small amplitude of the vibration signals, and changes in the structure and complexity of these signals. From fault detection point of view, the results are at the same level of qualitative performance with other approaches based on multiscale analysis and other type of entropies, e.g. Multi Scale Entropy (MSE). From change detection in the structure of the signals, the proposed method is superior, being more sensitive to random components and smoother. The proposed method could be adapted to other categories of difficult signals or processes, as those from medical area.

This work is an exploratory and preliminary study on change detection in the structure of time series based on structural entropy processing, with direct application to fault detection in production processes. Starting with some theoretical elements related to entropy, various types of estimators are presented, and some are discussed, as the approximate entropy (ApEn), the sample entropy (SampEn), and the multiscale entropies (MSE). The case studies contain synthetized signals as white Gaussian noise, 1/f noise, and random signals with a probabilistic structure, ended with real vibration signals generated by faults in bearings. Finally, a change detection criterion is promoted based on the average of multiscale entropy and cumulative sum technique. The study is useful for a better understanding of the entropy-based methods for change detection in the structure of the time-series and it prepares the next level of the research, i.e. methods based on Renyi entropy with recursive estimators - when possible - and their extension to chaotic and stochastic systems, especially from the structure change detection point of view.

The paper presents the vibration analysis problem with application in predictive maintenance of Rolling Elements Bearings (REB). After an overview of the maintenance approach, the condition monitoring in predictive maintenance is presented. A general view on change detection problem, with application in vibration monitoring, precedes some experimental results obtained in REB operating, for multiple faults and faults which gradually occur, with the conceptual description of the algorithm used. The approach proved to offer more robust detection of faults in REB, able to assure proactive actions in predictive maintenance.

Condition monitoring, change detection, and diagnosis are important activities and research directions in the field of system engineering and maintenance of the equipment and industrial processes. Starting from a relative difficulty to implement new advanced algorithms in industry, the paper puts forward a development system for building and testing new methods and algorithms in the field of change detection and diagnosis. The proposed system has specialized sub-systems/modules: VIBROGEN to generate vibration signals, as effects of the faults, under controlled conditions, i.e., time, size, and working loads; VIBROSIG for signal pre-processing and acquisition tasks. The VIBROCHANGE is the main sub-system and it works on two levels. The upper one is VIBROTOOL, which runs on high-level programming languages with complex libraries and algorithms, suitable more for algorithm development. The lower one is VIBROMOD, and it works in real conditions of the monitored process, with restricted hardware and software, such as type and library. It implements and runs the converted algorithms from the high-level language, and accepted by to the industrial electronic equipment. The system provides robust results on both levels. The development stage of the algorithms is considered finished only when the results on both levels are close enough one to the other. As example of usage, some results of change detection algorithms, running on the highest-level, are presented and discussed. The processed signals are coming from vibrations generated by bearings with various type and size of faults. The obtained results recommend that the system should be used in the development and testing activities of models, methods, and algorithms for condition monitoring.