The wavelet transform has been successfully used in the area of power quality analysis. There are many published papers with methods for power quality disturbance classification or harmonics measurement, which use wavelet transform. However, the properties of the wavelet transform can drastically vary from the choice of the wavelet. In this paper we analyze the influence of the choice of the wavelet to the accuracy of the power quality classification method and to high frequency harmonics measurements. Additionally to the well known wavelet filters we introduce near perfect reconstruction filter banks. The simulation results indicate that these filter banks are a good choice for classification of power quality disturbances, especially in the presence of noise and for high frequency harmonics measurements
The research was aimed at examining the impact of the petrographic composition of coal from the Janina mine on the gasification process and petrographic composition of the resulting char. The coal was subjected to fluidized bed gasification at a temperature below 1000°C in oxygen and CO2 atmosphere. The rank of coal is borderline subbituminous to bituminous coal. The petrographic composition is as follows: macerals from the vitrinite (61.0% vol.); liptinite (4.8% vol.) and inertinite groups (29.0% vol.). The petrofactor in coal from the Janina deposit is 6.9. The high content of macerals of the inertinite group, which can be considered inert during the gasification, naturally affects the process. The content of non-reactive macerals is around 27% vol. The petrographic analysis of char was carried out based on the classification of International Committee for Coal and Organic Petrology. Both inertoid (34.7% vol.) and crassinetwork (25.1% vol.) have a dominant share in chars resulting from the above-mentioned process. In addition, the examined char contained 3.1% vol. of mineroids and 4.3% vol. of fusinoids and solids. The calculated aromaticity factor increases from 0.75 in coal to 0.98 in char. The carbon conversion is 30.3%. Approximately 40% vol. of the low porosity components in the residues after the gasification process indicate a low degree of carbon conversion. The ash content in coal amounted to 13.8% and increased to 24.10% in char. Based on the petrographic composition of the starting coal and the degree of conversion of macerals in the char, it can be stated that the coal from the Janina deposit is moderately suitable for the gasification process.
The primary evaluation of the economic losses caused by water pollution in Shanghai in the year 2009 is made by classification approach in order to provide basis for decision of the relative water management policy. The result shows that the portion of water pollution losses in GDP of Shanghai was 2.7%, which was still lower than the average level of whole China despite of the local high population density and the scale of industry, suggesting to some extent the continuous attention in water protection paid by Shanghai government.
This work is focused on the automatic recognition of environmental noise sources that affect humans’ health and quality of life, namely industrial, aircraft, railway and road traffic. However, the recognition of the latter, which have the largest influence on citizens’ daily lives, is still an open issue. Therefore, although considering all the aforementioned noise sources, this paper especially focuses on improving the recognition of road noise events by taking advantage of the perceived noise differences along the road vehicle pass-by (which may be divided into different phases: approaching, passing and receding). To that effect, a hierarchical classification scheme that considers these phases independently has been implemented. The proposed classification scheme yields an averaged classification accuracy of 92.5%, which is, in absolute terms, 3% higher than the baseline (a traditional flat classification scheme without hierarchical structure). In particular, it outperforms the baseline in the classification of light and heavy vehicles, yielding a classification accuracy 7% and 4% higher, respectively. Finally, listening tests are performed to compare the system performance with human recognition ability. The results reveal that, although an expert human listener can achieve higher recognition accuracy than the proposed system, the latter outperforms the non-trained listener in 10% in average.
Population density varies sharply from place to place on the whole territory of Poland. The largest number of people per 1 km2 is 21,531, while uninhabited areas account for about 48% of the country. Such uneven, non-Gaussian distribution of the data causes some difficulty in choosing the classification method in geometric choropleth maps. A thorough evaluation of a geometric choropleth map of population data is not possible using only traditional indicators such as the Tabular Accuracy Index (TAI). That is why the aim of the article is to develop an innovative index based on distance analysis and neighbour analysis of grid cells. Two indexes have been suggested in this paper: the Spatial Distance Index (SDI) and the Spatial Contiguity Index (SCI). The paper discusses the use of five classification methods to evaluate choropleth maps of population data, like head-tail breaks, natural breaks, equal intervals, quantile, and geometrical intervals. A comprehensive assessment of such geometric choropleth maps is also done. The research was conducted for the whole territory of Poland, using data from the 2011 National Census of Population and Housing. Population data are presented in the 1km grid. The results of the analysis are shown on thematic maps. A compatibility of the choropleth maps with urban-rural typology of the OECD (Organisation for Economic Co-operation and Development) was also checked.
Classification techniques have been widely used in different remote sensing applications and correct classification of mixed pixels is a tedious task. Traditional approaches adopt various statistical parameters, however does not facilitate effective visualisation. Data mining tools are proving very helpful in the classification process. We propose a visual mining based frame work for accuracy assessment of classification techniques using open source tools such as WEKA and PREFUSE. These tools in integration can provide an efficient approach for getting information about improvements in the classification accuracy and helps in refining training data set. We have illustrated framework for investigating the effects of various resampling methods on classification accuracy and found that bilinear (BL) is best suited for preserving radiometric characteristics. We have also investigated the optimal number of folds required for effective analysis of LISS-IV images.
In this paper the authors propose a decision support system for automatic blood smear analysis based on microscopic images. The images are pre-processed in order to remove irrelevant elements and to enhance the most important ones – the healthy blood cells (erythrocytes) and the pathologic ones (echinocytes). The separated blood cells are analysed in terms of their most important features by the eigenfaces method. The features are the basis for designing the neural network classifier, learned to distinguish between erythrocytes and echinocytes. As the result, the proposed system is able to analyse the smear blood images in a fully automatic way and to deliver information on the number and statistics of the red blood cells, both healthy and pathologic. The system was examined in two case studies, involving the canine and human blood, and then consulted with the experienced medicine specialists. The accuracy of classification of red blood cells into erythrocytes and echinocytes reaches 96%.
Power electronic circuits (PECs) are prone to various failures, whose classification is of paramount importance. This paper presents a data-driven based fault diagnosis technique, which employs a support vector data description (SVDD) method to perform fault classification of PECs. In the presented method, fault signals (e.g. currents, voltages, etc.) are collected from accessible nodes of circuits, and then signal processing techniques (e.g. Fourier analysis, wavelet transform, etc.) are adopted to extract feature samples, which are subsequently used to perform offline machine learning. Finally, the SVDD classifier is used to implement fault classification task. However, in some cases, the conventional SVDD cannot achieve good classification performance, because this classifier may generate some so-called refusal areas (RAs), and in our design these RAs are resolved with the one-against-one support vector machine (SVM) classifier. The obtained experiment results from simulated and actual circuits demonstrate that the improved SVDD has a classification performance close to the conventional one-against-one SVM, and can be applied to fault classification of PECs in practice.
Abstract. In this paper we present a new class of neuro-fuzzy systems designed for system modelling and pattern classi.cation. Our approach is characterized by automatic determination of fuzzy inference in the process of learning. Moreover, we introduce several .exibility concepts in the design of neuro-fuzzy systems. The method presented in the paper is characterized by high accuracy which outperforms previous techniques applied for system modelling and pattern classi.cation.
Based on recent advances in non-linear analysis, the surface electromyography (sEMG) signal has been studied from the viewpoints of self-affinity and complexity. In this study, we examine usage of critical exponent analysis (CE) method, a fractal dimension (FD) estimator, to study properties of the sEMG signal and to deploy these properties to characterize different movements for gesture recognition. SEMG signals were recorded from thirty subjects with seven hand movements and eight muscle channels. Mean values and coefficient of variations of the CE from all experiments show that there are larger variations between hand movement types but there is small variation within the same type. It also shows that the CE feature related to the self-affine property for the sEMG signal extracted from different activities is in the range of 1.855~2.754. These results have also been evaluated by analysis-of-variance (p-value). Results show that the CE feature is more suitable to use as a learning parameter for a classifier compared with other representative features including root mean square, median frequency and Higuchi's method. Most p-values of the CE feature were less than 0.0001. Thus the FD that is computed by the CE method can be applied to be used as a feature for a wide variety of sEMG applications.
The paper deals with the new method of automatic vehicle classification called ALT (ALTernative). Its characteristic feature is versatility resulting from its open structure, moreover a user can adjust the number of vehicles and their category according to individual requirements. It uses an algorithm for automatic vehicle recognition employing data fusion methods and fuzzy sets. High effectiveness of classification while retaining high selectivity of division was proved by test results. The effectiveness of classification of all vehicles at the level of 95% and goods trucks of 100% is more than satisfactory.
This paper presents a novel strategy of fault classification for the analog circuit under test (CUT). The proposed classification strategy is implemented with the one-against-one Support Vector Machines Classifier (SVC), which is improved by employing a fault dictionary to accelerate the testing procedure. In our investigations, the support vectors and other relevant parameters are obtained by training the standard binary support vector machines. In addition, a technique of radial-basis-function (RBF) kernel parameter evaluation and selection is invented. This technique can find a good and proper kernel parameter for the SVC prior to the machine learning. Two typical analog circuits are demonstrated to validate the effectiveness of the proposed method.
At present, the speed of production and its complexity increases with each passing year due to the shorter product life cycle and competition in the global market. This trend is also observed in the machine-building industry, therefore, in order to ensure the competitiveness of enterprises and reduce the cost of production, it is necessary to intensify production. This is especially true in the machining of complex parts that require a great number of setups, and technological equipment. The problem-oriented analysis of complex parts was carried out, the parts classification was structured and developed according to the design and technological features. This made it possible to offer advanced manufacturing processes for complex parts like levers, forks, and connecting rods. The flexible fixtures for specified complex parts were developed. The effectiveness of the proposed manufacturing processes,
In order to make the analog fault classification more accurate, we present a method based on the Support Vector Machines Classifier (SVC) with wavelet packet decomposition (WPD) as a preprocessor. In this paper, the conventional one-against-rest SVC is resorted to perform a multi-class classification task because this classifier is simple in terms of training and testing. However, this SVC needs all decision functions to classify the query sample. In our study, this classifier is improved to make the fault classification task more fast and efficient. Also, in order to reduce the size of the feature samples, the wavelet packet analysis is employed. In our investigations, the wavelet analysis can be used as a tool of feature extractor or noise filter and this preprocessor can improve the fault classification resolution of the analog circuits. Moreover, our investigation illustrates that the SVC can be applicable to the domain of analog fault classification and this novel classifier can be viewed as an alternative for the back-propagation (BP) neural network classifier.
The article describes the problem of selection of heat treatment parameters to obtain the required mechanical properties in heat- treated bronzes. A methodology for the construction of a classification model based on rough set theory is presented. A model of this type allows the construction of inference rules also in the case when our knowledge of the existing phenomena is incomplete, and this is situation commonly encountered when new materials enter the market. In the case of new test materials, such as the grade of bronze described in this article, we still lack full knowledge and the choice of heat treatment parameters is based on a fragmentary knowledge resulting from experimental studies. The measurement results can be useful in building of a model, this model, however, cannot be deterministic, but can only approximate the stochastic nature of phenomena. The use of rough set theory allows for efficient inference also in areas that are not yet fully explored.
An automatic analysis of product reviews requires deep understanding of the natural language text by machine. The limitation of bag-of-words (BoW) model is that a large amount of word relation information from the original sentence is lost and the word order is ignored. Higher-order-N-grams also fail to capture the long-range dependency relations and word order information. To address these issues, syntactic features extracted from the dependency relations can be used for machine learning based document-level sentiment classification. Generalization of syntactic dependency features and negation handling is used to achieve more accurate classification. Further to reduce the huge dimensionality of the feature space, feature selection methods based on information gain (IG) and weighted frequency and odds (WFO) are used. A supervised feature weighting scheme called delta term frequency-inverse document frequency (TF-IDF) is also employed to boost the importance of discriminative features using the observed uneven distribution of features between the two classes. Experimental results show the effectiveness of generalized syntactic dependency features over standard features for sentiment classification using Boolean multinomial naive Bayes (BMNB) classiﬁer.
In recent years, deep learning and especially deep neural networks (DNN) have obtained amazing performance on a variety of problems, in particular in classification or pattern recognition. Among many kinds of DNNs, the convolutional neural networks (CNN) are most commonly used. However, due to their complexity, there are many problems related but not limited to optimizing network parameters, avoiding overfitting and ensuring good generalization abilities. Therefore, a number of methods have been proposed by the researchers to deal with these problems. In this paper, we present the results of applying different, recently developed methods to improve deep neural network training and operating. We decided to focus on the most popular CNN structures, namely on VGG based neural networks: VGG16, VGG11 and proposed by us VGG8. The tests were conducted on a real and very important problem of skin cancer detection. A publicly available dataset of skin lesions was used as a benchmark. We analyzed the influence of applying: dropout, batch normalization, model ensembling, and transfer learning. Moreover, the influence of the type of activation function was checked. In order to increase the objectivity of the results, each of the tested models was trained 6 times and their results were averaged. In addition, in order to mitigate the impact of the selection of learning, test and validation sets, k-fold validation was applied.
In this article, we present a comprehensive measurement system to determine the level of user emotional arousal by the analysis of electrodermal activity (EDA). A number of EDA measurements were collected, while emotions were elicited using specially selected movie sequences. Data collected from 16 participants of the experiment, in conjunction with those from personal questionnaires, were used to determine a large number of 20 features of the EDA, to assess the emotional state of a user. Feature selection was performed using signal processing and analysis methods, while considering user declarations. The suitability of the designed system for detecting the level of emotional arousal was fully confirmed, throughout the number of experiments. The average classification accuracy for two classes of the least and the most stimulating movies varies within the range of 61‒72%.
Qualitative and quantitative results of high terrain elevation effect on spectral radiance of optical satellite image which affect the accuracy in retrieving of land surface cover changes is given. The paper includes two main parts: correction model of spectral radiance of satellite image affected by high terrain elevation and assessment of impacts and variation of land cover changes before and after correcting influence of high terrain elevation to the spectral radiance of the image. Study has been carried out with SPOT 5 in Hoa Binh mountain area of two periods: 2007 and 2010. Results showed that appropriate correction model is the Meyer’s one. The impacts of correction spectral radiance to 7 classes of classified images fluctuate from 15% to 400%. The varying changes before and after correction of image radiation fluctuate over 7 classes from 5% to 100%.
Self-aligning roller bearings are an integral part of the industrial machinery. The proper analysis and prediction of the various faults that may happen to the bearing beforehand contributes to an increase in the working life of the bearing. This study aims at developing a novel method for the analysis of the various faults in self-aligning bearings as well as the automatic classification of faults using artificial neural network (ANN) and deep neural network (DNN). The vibration data is collected for six different faults as well as for the healthy bearing. Empirical mode decomposition (EMD) followed by Hilbert Huang transform is used to extract instantaneous frequency peaks which are used for fault analysis. Time domain and time-frequency domain features are then extracted which are used to implement the neural networks through the pattern recognition tool in MATLAB. A comparative study of the outputs from the two neural networks is also performed. From the confusion matrix, the efficiency of the ANN has been found to be 95.7% and using DNN has been found to be 100%.
EEG signal-based sleep stage classification facilitates an initial diagnosis of sleep disorders. The aim of this study was to compare the efficiency of three methods for feature extraction: power spectral density (PSD), discrete wavelet transform (DWT) and empirical mode decomposition (EMD) in the automatic classification of sleep stages by an artificial neural network (ANN). 13650 30-second EEG epochs from the PhysioNet database, representing five sleep stages (W, N1-N3 and REM), were transformed into feature vectors using the aforementioned methods and principal component analysis (PCA). Three feed-forward ANNs with the same optimal structure (12 input neurons, 23 + 22 neurons in two hidden layers and 5 output neurons) were trained using three sets of features, obtained with one of the compared methods each. Calculating PSD from EEG epochs in frequency sub-bands corresponding to the brain waves (81.1% accuracy for the testing set, comparing with 74.2% for DWT and 57.6% for EMD) appeared to be the most effective feature extraction method in the analysed problem.
Purpose: to demonstrate the possibility of finding features reliable for more precise distinguishing between normal and abnormal Pattern Electroretinogram (PERG) recordings, in Continuous Wavelet Transform (CWT) coefficients domain. To determine characteristic features of the PERG and Pattern Visual Evoked Potential (PVEP) waveforms important in the task of precise classification and assessment of these recordings. Material and methods: 60 normal PERG waveforms and 60 PVEPs as well as 47 PERGs and 27 PVEPs obtained in some retinal and optic nerve diseases were studied in the two age groups (<= 50 years, > 50 years). All these signals were recorded in accordance with the guidelines of ISCEV in the Laboratory of Electrophysiology of the Retina and Visual Pathway and Static Perimetry, at the Department and Clinic of Ophthalmology of the Pomeranian Medical University. Continuous Wavelet Transform (CWT) was used for the time-frequency analysis and modelling of the PERG signal. Discriminant analysis and logistic regression were performed in statistical analysis of the PERG and PVEP signals. Obtained mathematical models were optimized using Fisher F(n1; n2) test. For preliminary evaluation of the obtained classification methods and algorithms in clinical practice, 22 PERGs and 55 PVEPs were chosen with respect to especially difficult discrimination problems (borderline recordings). Results: comparison between the method using CWT and standard time-domain based analysis showed that determining the maxima and minima of the PERG waves was achieved with better accuracy. This improvement was especially evident in waveforms with unclear peaks as well as in noisy signals. Predictive, quantitative models for PERGs and PVEPs binary classification were obtained based on characteristic features of the waveform morphology. Simple calculations algorithms for clinical applications were elaborated. They proved effective in distinguishing between normal and abnormal recordings. Conclusions: CWT based method is efficient in more precise assessment of the latencies of the PERG waveforms, improving separation between normal and abnormal waveforms. Filtering of the PERG signal may be optimized based on the results of the CWT analysis. Classification of the PERG and PVEP waveforms based on statistical methods is useful in preliminary interpretation of the recordings as well as in supporting more accurate assessment of clinical data.
This article presents a study on music genre classification based on music separation into harmonic and drum components. For this purpose, audio signal separation is executed to extend the overall vector of parameters by new descriptors extracted from harmonic and/or drum music content. The study is performed using the ISMIS database of music files represented by vectors of parameters containing music features. The Support Vector Machine (SVM) classifier and co-training method adapted for the standard SVM are involved in genre classification. Also, some additional experiments are performed using reduced feature vectors, which improved the overall result. Finally, results and conclusions drawn from the study are presented, and suggestions for further work are outlined.
This article presents the peculiarities and methodical principles for designing the technologies and forms of organization of the construction liquidation cycle for typical unified series of residential buildings. The systematic approach for developing the necessary settings and indicators of the structure of a complex technological process for disassembling, destructing and demolishing of structural elements and buildings in general is given. The multigraph is created for the closed walk model of correlation of the parameters of the organizational and technological solutions of the construction liquidation cycle.