جستجوی مقالات مرتبط با کلیدواژه « classification » در نشریات گروه « برق »

The enhanced power transfer capability is possible with the six-phase transmission system but it did not gain popularity due to the lack of a proper protection scheme to secure the line from 120 types of different possible short circuit faults. This work presents a protection scheme with discrete wavelet transform (db4 mother wavelet) and an artificial neural network (ANN). The Levenberg-Marquardt algorithm is used for training the ANNs. This protection scheme requires only the pre-processed current information of the sending end bus. For fault detection and classification of all 120 fault types, a single ANN module is implemented with six inputs and six outputs. For fault location estimation in each phase, 11 ANN modules with six outputs are implemented, one for each of the 11 types of combination of faults. The MATLAB/ SIMULINK simulation results of the proposed protection technique implemented on the six-phase Allegheny power transmission system show that it is effective and efficient in detecting and classifying all the faults with varying fault parameters with an accuracy of 99.76%. It is found that the performance of the fault location estimation modules is better with the training data and moderate with the testing data.

Trust is the main concern of the Bank's customers regarding electronic and Internet services. The trust of both customers is logically and experimentally important to each other, and banks need to take more steps as service providers to maintain their customers. It is necessary to increase the factors affecting the satisfaction and reliability of customers in banks using data mining. In this paper, we examine the factors affecting the increase of customers' confidence in banking and Internet banking services and the impact of any perceived credit factor by public and private banks, service providers, and infrastructure providers in electronic banking. The presented method is based on scientific data mining algorithms such as clustering and classification of the decision tree J48 and the neural network, as well as a quick and practical application of the miner. Data are analyzed using a questionnaire with the bank customers of 25 Tejart bank branches in Tehran. The experimental results demonstrate that the accuracy of the decision tree classification algorithm is 84.04 and the neural network is 72.3%.

Concepts and laws of physics have been a valuable source of inspiration for engineers to overcome human challenges and problems. Classification is an important example of such problems that plays a major role in various fields of engineering sciences. It is shown that discriminative classifiers tend to outperform their generative counterparts, especially in the presence of sufficient labeled training data. In this paper, we present a new physics-inspired discriminative classification method using minimum potential line. To do this, we first consider two groups of fixed point charges (as two classes of data) and a movable classifier line between them. Then, we find a stable position for the classifier line by minimizing the total potential integral on the classifier line due to the two groups of point charges. Surprisingly, it will be shown that the obtained classifier is actually an uncertainty-based classifier that minimizes the total uncertainty of the classifier line. The effectiveness of the proposed method is validated by some experiments on both synthetic and real datasets. First, two synthetic datasets are constructed to visually demonstrate the efficiency of the proposed method. Then, some real-world benchmark datasets are selected (from the well-known UCI machine learning repository) to compare the classification performance of the proposed method with three well-known classification methods.

Automatic modulation classification is used in various applications, including satellite communication systems, military communication, and submarine communications. In this paper, the automatic classification of modulation types is done using a two-stage method that combines a short-time Fourier transform (STFT) and a hybrid deep neural network (HDNN). At the first stage, using the STFT as a data source, the time-frequency information is retrieved from the modulated signals. A hybrid deep neural network feed two-dimensional (2D) images as inputs. In the second stage, the HDNN feeds the 2D time-frequency data to classify the various modulation types. Six various types of modulation schemes, including amplitude-shift keying, frequency-shift keying, phase-shift keying, quadrature amplitude-shift keying, quadrature frequency-shift keying, and quadrature phase-shift keying, are recognized automatically in the SNR range of 0 to 25 dB. An exhaustive computer simulation has been performed to evaluate the performance of the proposed digital modulation classification method. The simulation results show that, in comparison with the existing methods, our proposed method performs well and significantly reduces the processing time.

Head and neck cancer (HNC) recurrence is ever increasing among Ghanaian men and women. Because not all machine learning classifiers are equally created, even if multiple of them suite very well for a given task, it may be very difficult to find one which performs optimally given different distributions. The stacking learns how to best combine weak classifier models to form a strong model. As a prognostic model for classifying HNSCC recurrence patterns, this study tried to identify the best stacked ensemble classifier model when the same ML classifiers for feature selection and stacked ensemble learning are used. Four stacked ensemble models; in which first one used two base classifiers: gradient boosting machine (GBM) and distributed random forest (DRF); second one used three base classifiers: GBM, DRF, and deep neural network (DNN); third one used four base classifiers: GBM, DRF, DNN, and generalized linear model (GLM); and fourth one used five base classifiers: GBM, DRF, DNN, GLM, and Naïve bayes (NB) were developed, using GBM meta-classifier in each case. The results showed that implementing stacked ensemble technique consisting of five base classifiers on gradient boosted features achieved better performance than achieved on other feature subsets, and implementing this stacked ensemble technique on gradient boosted features achieved better performance compared to other stacked ensemble techniques implemented on gradient boosted features and other feature subsets used. Learning stacked ensemble technique having five base classifiers on GBM features is clinically appropriate as a prognostic model for classifying and predicting HNSCC patients’ recurrence data.

The transportation sector accounts for a significant portion of global energy consumption and, gasoline is a major fuel consumed in road transport. On the other hand, the excessive consumption of gasoline can lead to an increase in unnecessary trips and road accidents. This study aims to determine the impact of macroscale factors on gasoline consumption. In this regard, we investigated the effect of gasoline price, oil reserves, and income level on gasoline consumption per capita in about 90 countries, including Iran, over a period of 17 years. Also, Rail and Air travel per capita and membership in the Organization for Economic Co-operation and Development (OECD) were considered as control variables. For this purpose, one of the classification techniques utilized in the area of data mining, Classification, and Regression Tree (CART), was employed. The variable importance measure (VIM) was calculated to quantify the association of each independent variable with the target variable. The results indicated that oil reserves, gasoline prices, and average income have a normalized significance of 100, 58.5, and 30.3 % respectively. Other variables do not have significant importance. So, higher per-capita gasoline consumption is exclusively involved in oil-rich countries such as Iran. Therefore, considering national oil reserves should be prioritized when comparing fuel consumption across world countries. Also, more expensive gasoline would relatively diminish its use. However, this effect of the gasoline price is mostly confirmed for countries with lower national oil reserves which often have higher prices than their counterparts.

To mitigate COVID-19’s overwhelming burden, a rapid and efficient early screening scheme for COVID-19 in the first-line is required. Much research has utilized laboratory tests, CT scans, and X-ray data, which are obstacles to agile and real-time screening. In this study, we propose a user-friendly and low-cost COVID-19 detection model based on self-reportable data at home. The most exhausted input features were identified and included in the demographic, symptoms, semi-clinical, and past/present disease data categories. We employed Grid search to identify the optimal combination of hyperparameter settings that yields the most accurate prediction. Next, we apply the proposed model with tuned hyperparameters to 11 classic state-of-the-art classifiers. The results show that the XGBoost classifier provides the highest accuracy of 73.3%, but statistical analysis shows that there is no significant difference between the accuracy performance of XGBoost and AdaBoost, although it proved the superiority of these two methods over other methods. Furthermore, the most important features obtained using SHapely Adaptive explanations were analyzed. “Contact with infected people,” “cough,” “muscle pain,” “fever,” “age,” “Cardiovascular commodities,” “PO2,” and “respiratory distress” are the most important variables. Among these variables, the first three have a relatively large positive impact on the target variable. Whereas, “age,” “PO2”, and “respiratory distress” are highly negatively correlated with the target variable. Finally, we built a clinically operable, visible, and easy-to-interpret decision tree model to predict COVID-19 infection.

In spite of the numerous benefits over the traditional power distribution system, protection of the microgrid is a challenging and complex task. The varying fault resistances due to dissimilar grounding conditions can affect the performance of the protection scheme. Under such conditions, the magnitude of the fault current can vary from lower to higher level. In addition to the above, the dissimilar magnitude of fault current during grid connected and islanded mode demands a protection scheme that can easily discriminate the mode of operation. The magnitude of fault current in grid-connected and islanded modes needs a robust protection scheme. In this regard, an ensemble of subspace kNN based robust protection scheme has been proposed to detect the faulty conditions of the microgrid. The tasks of the mode detection, fault detection/classification as well as faulty line identification has been carried out in the proposed work. In the proposed protection scheme, discrete wavelet transform (DWT) has been used for processing of the data. After recording the voltage and current signals at bus-1, the protection scheme has been validated. The validation of the protection scheme in Section 6 reveals that the protection scheme is efficiently working.

This paper reports support vector machine (SVM) based fault detection and classification in microgrid while considering distortions in voltages and currents, time and frequency series parameters, and differential parameters. For SVM-based fault classification, the data set is formed by analysing the operation of the standard IEC microgrid model, with and without grid interconnection, under different fault and non-fault scenarios. Fault scenarios also include different locations, resistances, and incident angles of fault. Whereas, for non-fault scenarios, the variation in load is considered. Voltages and currents from both ends of the distribution line (DL) are sampled at 1920 Hz. The time and frequency series parameters, total harmonic distortion (THD) in current and voltage, and differential parameters are determined. The SVM algorithm uses these parameters to detect and classify faults. The performance of this developed SVM based algorithm is compared with that of different machine learning algorithms. This comparative analysis reveals that SVM detects and classifies the faults on the microgrid with an accuracy of over 99.99%. The performance of the proposed method is also tested with 30 dB, 35 dB, and 40 dB noise in the generated data, which represent measurement errors.

With the development of electrical network infrastructure and the emergence of concepts such as demand response and using electric vehicles for purposes other than transportation, knowing the behavioral patterns of network technical specifications to manage electrical systems has become very important optimally. One of the critical parameters in the electrical system management is the distribution network imbalance. There are several ways to improve and control network imbalances. One of these ways is to detect the behavior of bus imbalance profiles in the network using data analysis. In the past, data analysis was performed for large environments such as states and countries. However, after the emergence of smart grids, behavioral study and recognition of these patterns in small-scale environments has found a fundamental and essential role in the deep management of these networks. One of the appropriate methods in identifying behavioral patterns is data mining. This paper uses the concepts of hierarchical and k-means clustering methods to identify the behavioral pattern of the imbalance index in an unbalanced distribution network. For this purpose, first, in an unbalanced network without the electric vehicle parking, the imbalance profile for all busses is estimated. Then, by applying the penetration coefficient of 25% and 75% for electric vehicles in the network, charging\discharging effects on the imbalance profile is determined. Then, by determining the target cluster and using demand response, the imbalance index is improved. This method reduces the number of busses competing in demand response programs. Next, using the concept of classification, a decision tree is constructed to minimize metering time.

The spread of internet and smartphones in recent years has led to the popularity and easy accessibility of social networks among users. Despite the benefits of these networks, such as ease of interpersonal communication and providing a space for free expression of opinions, they also provide the opportunity for destructive activities such as spreading false information or using fake accounts for fraud intentions. Fake accounts are mainly managed by bots. So, identifying bots and suspending them could very much help to increase the popularity and favorability of social networks. In this paper, we try to identify Persian bots on Twitter. This seems to be a challenging task in view of the problems pertinent to processing colloquial Persian. To this end, a set of features based on user account information and activity of users added to content features of tweets to classify users by several machine learning algorithms like Random Forest, Logistic Regression and SVM. The results of experiments on a dataset of Persian-language users show the proper performance of the proposed methods. It turns out that, achieving a balanced-accuracy of 93.86%, Random Forest is the most accurate classifier among those mentioned above.

As the number of features in practical applications grows, the feature selection process becomes increasingly important. Since the feature selection problem is NP-hard, no exact algorithm can determine the optimal subset over a reasonable period of time. However, traditional feature selection methods are time-consuming and tend to get stuck in local optima. Unlike traditional search techniques, metaheuristics are more effective at exploring and exploiting the search domain because they use several operators. Besides these behaviors of metaheuristics, we present an improved Equilibrium Optimizer algorithm using the density of population and entropy operator, which has proven a good exploration ability to provide a promising candidate solution. After that, a new feature selection model is developed based on the improved Equilibrium Optimizer algorithm. K-nearest neighbors is used as an evaluator for the new solutions. In order to test the performance of the proposed algorithm, simulation experiments are conducted on a set of 14 standard test functions containing both unimodal and multimodal functions. To evaluate the effectiveness of the proposed algorithm, 15 UCI benchmark datasets and five metaheuristics, GA, CS, GSA, RDA, and BBA  are applied. The experimental results revealed the effectiveness of our approach in terms of accuracy performance for the feature selection process.