• The Journal of the Acoustical Society of Korea
• /
• v.25 no.4
• /
• pp.159-163
• /
• 2006

In this paper a data-driven method to improve the performance of the Jacobian adaptation (JA) for the noisy speech recognition is proposed. In stead of constructing the reference HMM by using the model composition method like the parallel model combination (PMC), we propose to train the reference HMM directly with the noisy speech. This was motivated from the idea that the directly trained reference HMM will model the acoustical variations due to the noise better than the composite HMM. For the estimation of the Jacobian matrices, the Baum-Welch algorithm is employed during the training. The recognition experiments have been done to show the improved performance of the proposed method over the Jacobian adaptation as well as other model compensation methods.

• Journal of Korea Multimedia Society
• /
• v.20 no.9
• /
• pp.1559-1566
• /
• 2017

Recently in the IoT(Internet of Things) environment, a data collection in real-time through device's sensor has increased with an emergence of various devices. Collected data from IoT environment shows a large scale, non-uniform generation cycle and atypical. For this reason, the distributed processing technique is required to analyze the IoT sensor data. However if you do not consider the optimal scheduling for data and the processor of IoT in a distributed processing environment complexity increase the amount in assigning a task, the user is difficult to guarantee the QoS(Quality of Service) for the sensor data. In this paper, we propose APQTA(Adaptive Priority Queue-driven Task Allocation method for sensor data processing) to efficiently process the sensor data generated by the IoT environment. APQTA is to separate the data into job and by applying the priority allocation scheduling based on the deadline to ensure that guarantee the QoS at the same time increasing the efficiency of the data processing.

• 전기의세계
• /
• v.21 no.2
• /
• pp.9-19
• /
• 1972

The transient characteristics of separately excited d-c motor driven by thyristor d-c chopper is studied in this paper. The armature controlled system is applied. As a result of theoretrical analysis the following conculsions were drawn: (1) For the transient analysis, it is recognized that the state transition analysis is a more general method and powerful tool than the state equation method or signal flow graph method, although it includes iterative matrix calculations. And the system is dealt with a finite width sampled-data system in the state transition analysis. (2) The transient characteristics of the motor angular velocity and its torque to the sampling duration variation are compared with those due to the amplitude variation of d-c chopper voltage as follows. The attenuation rate of the transient characteristics is equal in both cases, but the initial value of the transient characteristics in former case is greater than in latter case. (3) The roots of characteristics equation of the system lie inside the unit circle of the Z-plane. Therefor the system is stable. Further it is found that as the sampling duration is decreased the relative stability is lessened.

• Wind and Structures
• /
• v.5 no.5
• /
• pp.441-462
• /
• 2002

This paper presents a practical numerical method to determine both the spatial and temporal distribution of driving rain on buildings. It is based on an existing numerical simulation technique and uses the building geometry and climatic data at the building site as input. The method is applied to determine the 3D spatial and temporal distribution of wind-driven rain on the facade a low-rise building of complex geometry. Distinct wetting patterns are found. The important causes giving rise to these particular patterns are identified : (1) sweeping of raindrops towards vertical building edges, (2) sweeping of raindrops towards top edges, (3) shelter effect by various roof overhang configurations. The comparison of the numerical results with full-scale measurements in both space and time for a number of on site recorded rain events shows the numerical method to yield accurate results.

• Proceedings of the KIEE Conference
• /
• 1991.07a
• /
• pp.143-146
• /
• 1991

This paper describes the analysis of a capacitor-driven induction coil-gun employing an equivalent circuit. The system differerntial equations are solved by using Runge-Kutta method. The velocity characteristics of projectile and current building in barrel ciols are studied. From the results, it is shown the optimal capacitance of capacitors, charging voltage and initial position of the projectile can be determined. These results will be used as the basis data for the design of capacitor driven coil-guns.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.5
• /
• pp.382-393
• /
• 1991

In this paper, we propose a method for recognizing printed Korean characters using the Preceding Layer Driven multi-layer perceptron. The new learning algorithm which assigns the weight values to an integer and makes use of the transfer function as the step function was presented to design the hardware. We obtained 522 Korean character-image as an experimental object through scanner with 600DPI resolution. The preprocessing for feature extraction of Korean character is the separation of individual character, noise elimination smoothing, thinnig, edge point extraction, branch point extraction, and stroke segmentation. The used feature data are the number of edge points and their shapes, the number of branch points, and the number of strokes with 8 directions.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.12
• /
• pp.27-37
• /
• 1991

In this paper, the novel PLD (Preceding Layer Driven) MLP (Multi Layer Perceptron) neural network model and its learning algorithm is described. This learning algorithm is different from the conventional. This integer weights and hard limit function are used for synaptic weight values and activation function, respectively. The entire learning process is performed by layer-by-layer method. the number of layers can be varied with difficulty of training data. Since the synaptic weight values are integers, the synapse circuit can be easily implemented with CMOS. PLD MLP neural network was applied to English Characters, arbitrary waveform generation and spiral problem.

• Journal of the Microelectronics and Packaging Society
• /
• v.31 no.2
• /
• pp.45-53
• /
• 2024

The maintenance of semiconductor equipment is crucial for the continuous growth of the semiconductor market. System management is imperative given the anticipated increase in the capacity and complexity of industrial equipment. Ensuring optimal operation of manufacturing processes is essential to maintaining a steady supply of numerous parts. Particularly, monitoring the status of substrate transfer robots, which play a central role in these processes, is crucial. Diagnosing failures of their major components is vital for preventive maintenance. Fault diagnosis methods can be broadly categorized into physics-based and data-driven approaches. This study focuses on data-driven fault diagnosis methods due to the limitations of physics-based approaches. We propose a methodology for data acquisition and preprocessing for robot fault diagnosis. Data is gathered from vibration sensors, and the data preprocessing method is applied to the vibration signals. Subsequently, the dataset is trained using Gradient Tree-based XGBoost machine learning classification algorithms. The effectiveness of the proposed model is validated through performance evaluation metrics, including accuracy, F1 score, and confusion matrix. The XGBoost classifiers achieve an accuracy of approximately 92.76% and an equivalent F1 score. ROC curves indicate exceptional performance in class discrimination, with 100% discrimination for the normal class and 98% discrimination for abnormal classes.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.18 no.5
• /
• pp.108-114
• /
• 2010

The injection nozzle of an electro-hydraulic injector is being opened and closed by movement of a injector's needle which is balanced by pressure at the nozzle seat and at the needle control chamber, at the opposite end of the needle. In this study, the effects of needle movement in a piezo-driven injector on unsteady cavitating flows behavior inside nozzle were investigated by cavitation numerical model based on the Eulerian-Lagrangian approach. Aimed at simulating the 3-D two-phase flow behavior, the three dimensional geometry model along the central cross-section regarding of one injection hole with real design data of a piezo-driven diesel injector has been used to simulate the cavitating flows for injection time by at fully transient simulation with cavitation model. The cavitation model incorporates many of the fundamental physical processes assumed to take place in cavitating flows. The simulations performed were both fully transient and 'pseudo' steady state, even if under steady state boundary conditions. As this research results, we found that it could analyze the effect the pressure drop to the sudden acceleration of fuel, which is due to the fastest response of needle, on the degree of cavitation existed in piezo-driven injector nozzle.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.6 no.10
• /
• pp.2473-2492
• /
• 2012

Energy conservation is a vital issue in wireless sensor networks. Recently, employing mobile sinks for data gathering become a pervasive trend to deal with this problem. The sink can follow stochastic or pre-defined paths; however the controlled mobility pattern nowadays is taken more into consideration. In this method, the sink moves across the network autonomously and changes its position based on the energy factors. Although the sink mobility would reduce nodes' energy consumption and enhance the network lifetime, the overhead caused by topological changes could waste unnecessary power through the sensor field. In this paper, we proposed EEDARS, an energy-efficient dual-sink algorithm with role switching mechanism which utilizes both static and mobile sinks. The static sink is engaged to avoid any periodic flooding for sink localization, while the mobile sink adaptively moves towards the event region for data collection. Furthermore, a role switching mechanism is applied to the protocol in order to send the nearest sink to the recent event area, hence shorten the path. This algorithm could be employed in event-driven and multi-hop scenarios. Analytical model and extensive simulation results for EEDARS demonstrate a significant improvement on the network metrics especially the lifetime, the load and the end-to-end delay.