• Nuclear Engineering and Technology
• /
• v.53 no.12
• /
• pp.4072-4079
• /
• 2021

This article presents a study on the state-of-the-art methods for automated radioactive material detection and identification, using gamma-ray spectra and modern machine learning methods. The recent developments inspired this in deep learning algorithms, and the proposed method provided better performance than the current state-of-the-art models. Machine learning models such as: fully connected, recurrent, convolutional, and gradient boosted decision trees, are applied under a wide variety of testing conditions, and their advantage and disadvantage are discussed. Furthermore, a hybrid model is developed by combining the fully-connected and convolutional neural network, which shows the best performance among the different machine learning models. These improvements are represented by the model's test performance metric (i.e., F1 score) of 93.33% with an improvement of 2%-12% than the state-of-the-art model at various conditions. The experimental results show that fusion of classical neural networks and modern deep learning architecture is a suitable choice for interpreting gamma spectra data where real-time and remote detection is necessary.

• International Journal of Computer Science & Network Security
• /
• v.21 no.12spc
• /
• pp.556-564
• /
• 2021

Numerous spatial and transform-domain-based conventional denoising algorithms struggle to keep critical and minute structural features of the image, especially at high noise levels. Although neural network approaches are effective, they are not always reliable since they demand a large quantity of training data, are computationally complicated, and take a long time to construct the model. A new framework of enhanced hybrid filtering is developed for denoising color images tainted by additive white Gaussian Noise with the goal of reducing algorithmic complexity and improving performance. In the first stage of the proposed approach, the noisy image is refined using a high-dimensional non-local means filter based on Principal Component Analysis, followed by the extraction of the method noise. The wavelet transform and SURE Shrink techniques are used to further culture this method noise. The final denoised image is created by combining the results of these two steps. Experiments were carried out on a set of standard color images corrupted by Gaussian noise with multiple standard deviations. Comparative analysis of empirical outcome indicates that the proposed method outperforms leading-edge denoising strategies in terms of consistency and performance while maintaining the visual quality. This algorithm ensures homogeneous noise reduction, which is almost independent of noise variations. The power of both the spatial and transform domains is harnessed in this multi realm consolidation technique. Rather than processing individual colors, it works directly on the multispectral image. Uses minimal resources and produces superior quality output in the optimal execution time.

• Journal of Power Electronics
• /
• v.19 no.4
• /
• pp.907-921
• /
• 2019

This paper presents two advanced hybrid pulse-width modulation (PWM) strategies for multilevel inverters (MLIs) that provide both common-mode voltage (CMV) elimination and current ripple reduction. The first PWM utilizes sequences that apply one switching state at the double ends of a half-carrier cycle. The second PWM combines the advantages of the former and an existing four-state PWM. Analyses of the harmonic characteristics of the two groups of switching sequences based on a general switching voltage model are carried out, and algorithms to optimize the current ripple are proposed. These methods are simple and can be implemented online for general n-level inverters. Using a three-level NPC inverter and a five-level CHB inverter, good performances in terms of the root mean square current ripple are obtained with the proposed PWM schemes as indicated through improved harmonic distortion factors when compared to existing schemes in almost the entire region of the modulation index. This also leads to a significant reduction in the current total harmonic distortion. Simulation and experimental results are provided to verify the effectiveness of the proposed PWM methods.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.05a
• /
• pp.610-613
• /
• 2013

Vehicular Ad Hoc NETwork (VANET), a subclass of Mobile Ah Hoc NETwork (MANET) has been a tech-buzz for the last couple of decades. VANET, yet not deployed, promises the ease, comfort, and safety to both drivers and passengers once deployed. The by far most important factor in successful VANET application is the data dissemination scheme. Such data includes scheduled beacons that contain whereabouts information of vehicles. In this paper, we aim at regularly broadcasted beacons and devise an algorithm to disseminate the beacon information up to a maximum distance and alleviate the broadcast storm problem at the same time. According to the proposed scheme, a vehicle before re-broadcasting a beacon, takes into account the current vehicular density in its neighborhood. The re-broadcasters are chosen away from the source of the beacon and among the candidate re-broadcasters, if the density in the neighborhood is high, then the candidate rebroadcaster re-broadcasts the beacon with high probability and with low probability, otherwise. We also performed thorough simulations of our algorithms and the results are sound according to the expectations.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.335-341
• /
• 2022

In this paper, we adopt a transformer that shows remarkable performance in natural language processing as an acoustic model of hybrid speech recognition. The transformer acoustic model uses attention structures to process sequential data and shows high performance with low computational cost. This paper proposes a method to improve the performance of transformer AM by applying each of the four algorithms of sequence discriminative training, a weighted finite-state transducer (wFST)-based learning used in the existing DNN-HMM model. In addition, compared to the Cross Entropy (CE) learning method, sequence discriminative method shows 5 % of the relative Word Error Rate (WER).

• Computers and Concrete
• /
• v.28 no.6
• /
• pp.635-642
• /
• 2021

In this paper, soft learning techniques are used to predict the ultimate torsional capacity of Reinforced Concrete beams strengthened with Fiber Reinforced Polymer. Soft computing techniques, namely Artificial Neural Network, trained by various back propagation algorithms, and Particle Swarm Optimization (PSO) algorithm, have been used to model and predict the torsional strength of Reinforced Concrete beams strengthened with Fiber Reinforced Polymer. The performance of each model has been evaluated by using statistical parameters such as coefficient of determination (R²), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The hybrid PSO NN model resulted in an R² of 0.9292 with an RMSE of 5.35 for training and an R² of 0.9328 with an RMSE of 4.57 for testing. Another model, ANN BP, produced an R² of 0.9125 with an RMSE of 6.17 for training and an R² of 0.8951 with an RMSE of 5.79 for testing. The results of the PSO NN model were in close agreement with the experimental values. Thus, the PSO NN model can be used to predict the ultimate torsional capacity of RC beams strengthened with FRP with greater acceptable accuracy.

• Computers and Concrete
• /
• v.32 no.3
• /
• pp.263-286
• /
• 2023

Experimenting with concrete to determine its compressive and tensile strengths is a laborious and time-consuming operation that requires a lot of attention to detail. Researchers from all around the world have spent the better part of the last several decades attempting to use machine learning algorithms to make accurate predictions about the technical qualities of various kinds of concrete. The research that is currently available on estimating the strength of concrete draws attention to the applicability and precision of the various machine learning techniques. This article provides a summary of the research that has previously been conducted on estimating the strength of concrete by making use of a variety of different machine learning methods. In this work, a classification of the existing body of research literature is presented, with the classification being based on the machine learning technique used by the researchers. The present review work will open the horizon for the researchers working on the machine learning based prediction of the compressive strength of concrete by providing the recommendations and benefits and drawbacks associated with each model as determining the compressive strength of concrete practically is a laborious and time-consuming task.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.4
• /
• pp.1276-1295
• /
• 2023

Sensor networks are now an essential aspect of wireless communication, especially with the introduction of new gadgets and protocols. Their ability to be deployed anywhere, especially where human presence is undesirable, makes them perfect choices for remote observation and control. Despite their vast range of applications from home to hostile territory monitoring, limited battery power remains a limiting factor in their efficacy. To analyze and transmit data, it requires intelligent use of available battery power. Several studies have established effective routing algorithms based on clustering. However, choosing optimal cluster heads and similarity measures for clustering significantly increases computing time and cost. This work proposes and implements a simple two-phase technique of route creation and maintenance to ensure route reliability by employing nature-inspired ant colony optimization followed by the fuzzy decision engine (FDE). Benchmark methods such as PSO, ACO and GWO are compared with the proposed HRCM's performance. The objective has been focused towards establishing the superiority of proposed work amongst existing optimization methods in a standalone configuration. An average of 15% improvement in energy consumption followed by 12% improvement in latency reduction is observed in proposed hybrid model over standalone optimization methods.

• Smart Structures and Systems
• /
• v.32 no.6
• /
• pp.393-402
• /
• 2023

Recently, hybrid and electric vehicles have been actively developed to replace internal combustion engine (ICE) vehicles. However, their vibrations and noise with complex spectra cause discomfort to drivers. To reduce the vibrations transmitted through primary excitation sources such as powertrains, structural changes have been introduced. However, the interference among different parts is a limitation. Thus, active mounting systems based on smart materials have been actively investigated to overcome these limitations. This study focuses on diminishing the source movement when a structure with two active mounting systems is excited to a single sinusoidal and a multi-frequency signal, which were investigated for source movement reduction. The overall structure was modeled based on the lumped parameter method. Active vibration control was implemented based on the modeled structure, and a multi-normalization least mean square (NLMS) algorithm was used to obtain the control input for the active mounting system. Furthermore, the performance of the NLMS algorithm was compared with that of the quantification method to demonstrate the performance of active vibration control. The results demonstrate that the vibration attenuation performance of the source component was improved.

• International Journal of Advanced Culture Technology
• /
• v.11 no.4
• /
• pp.393-405
• /
• 2023

Accurate prediction of wind speed and power is vital for enhancing the efficiency of wind energy systems. Numerous solutions have been implemented to date, demonstrating their potential to improve forecasting. Among these, deep learning is perceived as a revolutionary approach in the field. However, despite their effectiveness, the noise present in the collected data remains a significant challenge. This noise has the potential to diminish the performance of these algorithms, leading to inaccurate predictions. In response to this, this study explores a novel feature engineering approach. This approach involves altering the data input shape in both Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) and Autoregressive models for various forecasting horizons. The results reveal substantial enhancements in model resilience against noise resulting from step increases in data. The approach could achieve an impressive 83% accuracy in predicting unseen data up to the 24th steps. Furthermore, this method consistently provides high accuracy for short, mid, and long-term forecasts, outperforming the performance of individual models. These findings pave the way for further research on noise reduction strategies at different forecasting horizons through shape-wise feature engineering.