• Journal of Power Electronics
• /
• v.4 no.1
• /
• pp.1-11
• /
• 2004

A stable adaptive sensorless speed controller for three-level inverter fed induction motor direct torque control (DTC) system using the radial-basis function network (RBFN) is presented in this paper. Torque ripple in the DTC system for high power induction motor could be drastically reduced with the foregoing researches of switching voltage selection and torque ripple reduction algorithms. However, speed control performance is still influenced by the inherent uncertainty of the system such as parametric uncertainty, external load disturbances and unmodeled dynamics, and its exact mathematical model is much difficult to be obtained due to their strong nonlinearity. In this paper, the inherent uncertainty is approximated on-line by the RBFN, and an additional robust control term is introduced to compensate for the reconstruction error of the RBFN instead of the rich number of rules and additional updated parameters. Control law for stabilizing the system and adaptive laws for updating both of weights in the RBFN and a bounding constant are established so that the whole closed-loop system is stable in the sense of Lyapunov, and the stability proof of the whole control system is presented. Computer simulations as well as experimental results are presented to show the validity and effectiveness of the proposed system.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.24 no.12
• /
• pp.3053-3060
• /
• 2000

Alternative formulation is presented for robust optimization problems and an efficient computational scheme for reliability estimation is proposed. Both design variables and design parameters considered as random variables about their nominal values. To ensure the robustness of objective performance a new cost function bounding the performance and a new constraint limiting the performance variation are introduced. The constraint variations are regulated by considering the probability of feasibility. Each probability constraint is transformed into a sub-optimization problem and then is resolved with the modified advanced first order second moment(AFOSM) method for computational efficiency. The proposed robust optimization method has advantages that the mean value and the variation of the performance function are controlled simultaneously and the second order sensitivity information is not required even in case of gradient based optimization. The suggested method is examined by solving three examples and the results are compared with those for deterministic case and those available in literature.

• The Journal of Korea Robotics Society
• /
• v.14 no.4
• /
• pp.357-362
• /
• 2019

This paper proposes a model and train method that can real-time detect objects and distances estimation based on a monocular camera by applying deep learning. It used YOLOv2 model which is applied to autonomous or robot due to the fast image processing speed. We have changed and learned the loss function so that the YOLOv2 model can detect objects and distances at the same time. The YOLOv2 loss function added a term for learning bounding box values x, y, w, h, and distance values z as 클래스ification losses. In addition, the learning was carried out by multiplying the distance term with parameters for the balance of learning. we trained the model location, recognition by camera and distance data measured by lidar so that we enable the model to estimate distance and objects from a monocular camera, even when the vehicle is going up or down hill. To evaluate the performance of object detection and distance estimation, MAP (Mean Average Precision) and Adjust R square were used and performance was compared with previous research papers. In addition, we compared the original YOLOv2 model FPS (Frame Per Second) for speed measurement with FPS of our model.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.1
• /
• pp.245-265
• /
• 2022

In response to problems such as insufficient extraction information, low detection accuracy, and frequent misdetection in the field of Thangka image defects, this paper proposes a YOLOv5 prediction algorithm fused with the attention mechanism. Firstly, the Backbone network is used for feature extraction, and the attention mechanism is fused to represent different features, so that the network can fully extract the texture and semantic features of the defect area. The extracted features are then weighted and fused, so as to reduce the loss of information. Next, the weighted fused features are transferred to the Neck network, the semantic features and texture features of different layers are fused by FPN, and the defect target is located more accurately by PAN. In the detection network, the CIOU loss function is used to replace the GIOU loss function to locate the image defect area quickly and accurately, generate the bounding box, and predict the defect category. The results show that compared with the original network, YOLOv5-SE and YOLOv5-CBAM achieve an improvement of 8.95% and 12.87% in detection accuracy respectively. The improved networks can identify the location and category of defects more accurately, and greatly improve the accuracy of defect detection of Thangka images.

• Journal of the Korean Society of Radiology
• /
• v.18 no.1
• /
• pp.21-26
• /
• 2024

The one of the famous deep learning models for object detection task is you only look once version 5 (YOLOv5) framework based on the one stage architecture. In addition, YOLOv5 model indicated high performance for accurate lesion detection using the bottleneck CSP layer and skip connection function. The purpose of this study was to evaluate the performance of YOLOv5 framework according to various hyperparameters in position emission tomogrpahy (PET) phantom images. The dataset was obtained from QIN PET segmentation challenge in 500 slices. We set the bounding box to generate ground truth dataset using labelImg software. The hyperparameters for network train were applied by changing optimization function (SDG, Adam, and AdamW), activation function (SiLU, LeakyRelu, Mish, and Hardwish), and YOLOv5 model size (nano, small, large, and xlarge). The intersection over union (IOU) method was used for performance evaluation. As a results, the condition of outstanding performance is to apply AdamW, Hardwish, and nano size for optimization function, activation function and model version, respectively. In conclusion, we confirmed the usefulness of YOLOv5 network for object detection performance in nuclear medicine images.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.4
• /
• pp.499-505
• /
• 2011

In this paper, we proposed an adaptive fuzzy sliding mode control for nonlinear systems without parameter projection method. By modifying the controller structure, the parameters of the estimated input gain function are guaranteed not being identically zero and it is shown that the control scheme will not cause any implementation problem even if the estimated value of input gain function is zero at any moment during on-line operations. Except for the input gain function which an approximate estimate for its lower bound is needed, the proposed control scheme does not assume a priori the exact values of the bounding parameters. Based on Lyapunov synthesis methods, the overall control system guarantees that the tracking error asymptotically converges to zero and that all signals involved in controller are uniformly bounded. This can be illustrated by the simulation results for an inverted pendulum system.

• Journal of IKEEE
• /
• v.24 no.4
• /
• pp.1141-1147
• /
• 2020

The hornet species are so similar in shape that they are difficult for non-experts to classify, and because the size of the objects is small and move fast, it is more difficult to detect and classify the species in real time. In this paper, we developed a system that classifies hornets species in real time based on a deep learning algorithm using a boundary box. In order to minimize the background area included in the bounding box when labeling the training image, we propose a method of selecting only the head and body of the hornet. It also experimentally compares existing boundary box-based object recognition algorithms to find the best algorithms that can detect wasps in real time and classify their species. As a result of the experiment, when the mish function was applied as the activation function of the convolution layer and the hornet images were tested using the YOLOv4 model with the Spatial Attention Module (SAM) applied before the object detection block, the average precision was 97.89% and the average recall was 98.69%.

• Journal of Communications and Networks
• /
• v.16 no.1
• /
• pp.67-80
• /
• 2014

Spectrum sensing is vital for secondary unlicensed nodes to coexist and avoid interference with the primary licensed users in cognitive wireless networks. In this paper, we develop models for bounding interference levels from secondary network to the primary nodes within a spectrum sensing framework. Instead of classical stochastic approaches where Poisson point processes are used to model transmitters, we consider a more practical model which takes into account the medium access control regulations and where the secondary Poisson process is judiciously thinned in two phases to avoid interference with the secondary as well as the primary nodes. The resulting process will be a modified version of the Mat$\acute{e}$rn point process. For this model, we obtain bounds for the complementary cumulative distribution function of interference and present simulation results which show the developed analytical bounds are quite tight. Moreover, we use these bounds to find the operation regions of the secondary network such that the interference constraint is satisfied on receiving primary nodes. We then obtain theoretical results on the primary and secondary throughputs and find the throughput limits under the interference constraint.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.16 no.6 s.97
• /
• pp.556-562
• /
• 2005

In this paper, we propose interleaved super-orthogonal convolutional coded ultra wide-band impulse radio(ISOC-UWB-IR) system, and analyze its performance. The proposed system uses interleaver and deinterleaver to decorrelate the temporal correlation of the fading process and to obtain the diversity gain. We also suggest the three types of interleavers, which are pulse-wise interleaver(PI), pulse sub-group-wise interleaver(PSGI), and pulse group-wise interleaver(PGI). Performance analysis result shows that the interleaving scheme, rather than the code rate of super orthogonal convolutional code(SOC) encoder, affects the performance for the Rayleigh fading channel.

• Proceedings of the Korean Society For Composite Materials Conference
• /
• 2005.04a
• /
• pp.1-4
• /
• 2005

A higher order zig-zag shell theory is developed to refine accurately predict deformation and stress of smart shell structures under the mechanical, thermal, and electric loading. The displacement fields through the thickness are constructed by superimposing linear zig-zag field to the smooth globally cubic varying field. Smooth parabolic distribution through the thickness is assumed in the transverse deflection in order to consider transverse normal deformation. The mechanical, thermal, and electric loading is applied in the sinusoidal distribution function in the in-surface direction. Thermal and electric loading is given in the linear variation through the thickness. Especially, in electric loading case, voltage is only applied in piezo-layer. The layer-dependent degrees of freedom of displacement fields are expressed in terms of reference primary degrees of freedom by applying interface continuity conditions as well as bounding surface conditions of transverse shear stresses. In order to obtain accurate transverse shear and normal stresses, integration of equilibrium equation approach is used. The numerical examples of present theory demonstrate the accuracy and efficiency of the proposed theory. The present theory is suitable for the predictions of behaviors of thick smart composite shell under mechanical, thermal, and electric loadings combined.