• Journal of KIISE
• /
• v.43 no.12
• /
• pp.1356-1364
• /
• 2016

This paper proposes a novel aesthetic photo recomposition method using a deep convolutional neural network (DCNN). Previous recomposition approaches define the aesthetic score of photo composition based on the distribution of salient objects, and enhance the photo composition by maximizing the score. These methods suffer from heavy computational overheads, and often fail to enhance the composition because their optimization depends on the performance of existing salient object detection algorithms. Unlike previous approaches, we address the photo recomposition problem by utilizing DCNN, which shows remarkable performance in object detection and recognition. DCNN is used to iteratively predict cropping directions for a given photo, thus generating an aesthetically enhanced photo in terms of composition. Experimental results and user study show that the proposed framework can automatically crop the photo to follow specific composition guidelines, such as the rule of thirds.

• The Transactions of the Korean Institute of Electrical Engineers B
• /
• v.51 no.5
• /
• pp.242-248
• /
• 2002

It is important to develop new effective technologies to increase the interruption capacity and to reduce the size of a UB(Gas Circuit Breakers). Major design parameters such as nozzle geometries and interrupting chamber dimensions affect the cooling of the arc and the breaking performance. But it is not easy to test real GCB model in practice as in theory. Therefore, a simulation tool based on a computational fluid dynamics(CFD) algorithm has been developed to facilitate an optimization of the interrupter. Special attention has been paid to the supersonic flow phenomena between contacts and the observation of hat-gas flow for estimating the breaking performance. However, there are many difficult problems in calculating the flow characteristics in a GCB such as shock wave and complex geometries, which may be either static or in relative motion. Although a number of mesh generation techniques are now available, the generation of meshes around complicated, multi-component geometries like a GCB is still a tedious and difficult task for the computational fluid dynamics. This paper presents the CFD program using CFB-CAD integration technique based on Cartesian cut-cell method, which could reduce researcher's efforts to generate the mesh and achieve the accurate representation of the geometry designed by a CAD tools.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.50 no.7
• /
• pp.339-349
• /
• 2001

This paper suggests an optimal identification method for complex and nonlinear system modeling that is based on Fuzzy-Neural Networks(FNN). The proposed Hybrid Identification Algorithm is based on Yamakawa's FNN and uses the simplified inference as fuzzy inference method and Error Back Propagation Algorithm as learning rule. In this paper, the FNN modeling implements parameter identification using HCM algorithm and hybrid structure combined with two types of optimization theories for nonlinear systems. We use a HCM(Hard C-Means) clustering algorithm to find initial apexes of membership function. The parameters such as apexes of membership functions, learning rates, and momentum coefficients are adjusted using hybrid algorithm. The proposed hybrid identification algorithm is carried out using both a genetic algorithm and the improved complex method. Also, an aggregated objective function(performance index) with weighting factor is introduced to achieve a sound balance between approximation and generalization abilities of the model. According to the selection and adjustment of a weighting factor of an aggregate objective function which depends on the number of data and a certain degree of nonlinearity(distribution of I/O data), we show that it is available and effective to design an optimal FNN model structure with mutual balance and dependency between approximation and generalization abilities. To evaluate the performance of the proposed model, we use the time series data for gas furnace, the data of sewage treatment process and traffic route choice process.

• Transactions on Electrical and Electronic Materials
• /
• v.17 no.5
• /
• pp.261-269
• /
• 2016

A high-performance transmit/receive (T/R) switch is essential for every radio-frequency (RF) device. This paper proposes a T/R switch that is designed in the CEDEC 0.13 μm complementary metal-oxide-semiconductor (CMOS) technology for 2.4 GHz ISM-band RF applications. The switch exhibits a 1 dB insertion loss, a 28.6 dB isolation, and a 35.8 dBm power-handling capacity in the transmit mode; meanwhile, for the 1.8 V/0 V control voltages, a 1.1 dB insertion loss and a 19.4 dB isolation were exhibited with an extremely-low power dissipation of 377.14 μW in the receive mode. Besides, the variations of the insertion loss and the isolation of the switch for a temperature change from - 25℃ to 125℃ are 0.019 dB and 0.095 dB, respectively. To obtain a lucrative performance, an active inductor-based resonant circuit, body floating, a transistor W/L optimization, and an isolated CMOS structure were adopted for the switch design. Further, due to the avoidance of bulky inductors and capacitors, a very small chip size of 0.0207 mm² that is the lowest-ever reported chip area for this frequency band was achieved.

• Journal of Power Electronics
• /
• v.18 no.5
• /
• pp.1458-1469
• /
• 2018

This paper proposes a novel common-mode voltage (CMV) elimination (CMV-EL) method based on model predictive control (MPC) to eliminate CMV for three-level T-type inverters (3LT2Is). In the proposed MPC method, only six medium and one zero voltage vectors (VVs) (6MV1Z) that generate zero CMV are considered as candidates to perform the MPC. Moreover, the influence of dead-time effects on the CMV of the MPC-based 6MV1Z method is investigated, and the candidate VVs are redesigned by pre-excluding the VVs that will cause CMV fluctuations during the dead time from 6MV1Z. Only three or five VVs are included to perform optimization in every control period, which can significantly reduce the computational complexity. Thus, a small control period can be implemented in the practical applications to achieve improved grid current performance. With the proposed CMV-EL method, the CMV of the $3LT^2Is$ can be effectively eliminated. In addition, the proposed CMV-EL method can balance the neutral point potentials (NPPs) and yield satisfactory performance for grid current tracking in steady and dynamic states. Simulation and experimental results are presented to verify the effectiveness of the proposed method.

• Korean Journal of Remote Sensing
• /
• v.33 no.4
• /
• pp.359-375
• /
• 2017

We developed fog detection algorithm (KNU_FDA) based on the optical and textural properties of fog using satellite (COMS) and ground observation data. The optical properties are dual channel difference (DCD: BT3.7 - BT11) and albedo, and the textural properties are normalized local standard deviation of IR1 and visible channels. Temperature difference between air temperature and BT11 is applied to discriminate the fog from other clouds. Fog detection is performed according to the solar zenith angle of pixel because of the different availability of satellite data: day, night and dawn/dusk. Post-processing is also performed to increase the probability of detection (POD), in particular, at the edge of main fog area. The fog probability is calculated by the weighted sum of threshold tests. The initial threshold and weighting values are optimized using sensitivity tests for the varying threshold values using receiver operating characteristic analysis. The validation results with ground visibility data for the validation cases showed that the performance of KNU_FDA show relatively consistent detection skills but it clearly depends on the fog types and time of day. The average POD and FAR (False Alarm Ratio) for the training and validation cases are ranged from 0.76 to 0.90 and from 0.41 to 0.63, respectively. In general, the performance is relatively good for the fog without high cloud and strong fog but that is significantly decreased for the weak fog. In order to improve the detection skills and stability, optimization of threshold and weighting values are needed through the various training cases.

• Ocean Systems Engineering
• /
• v.6 no.3
• /
• pp.265-287
• /
• 2016

The change of the global performance of a turret-moored FPSO (Floating Production Storage Offloading) with DP (Dynamic Positioning) control is simulated, analyzed, and compared for two different internal turret location cases; bow and midship. Both collinear and non-collinear 100-yr GOM (Gulf of Mexico) storm environments and three cases (mooring-only, with DP position control, with DP position+heading control) are considered. The horizontal trajectory, 6DOF (degree of freedom) motions, fairlead mooring and riser tension, and fuel consumptions are compared. The PID (Proportional-Integral-Derivative) controller based on LQR (linear quadratic regulator) theory and the thrust-allocation algorithm which is based on the penalty optimization theory are implemented in the fully-coupled time-domain hull-mooring-riser-DP simulation program. Both in collinear and non-collinear 100-yr WWC (wind-wave-current) environments, the advantage of mid-ship turret is demonstrated by the significant reduction in heave at the turret location due to the minimal coupling with pitch mode, which is beneficial to mooring and riser design. However, in the non-collinear WWC environment, the mid-turret case exhibits unfavorable weathervaning characteristics, which can be reduced by employing DP position and heading controls as demonstrated in the present case studies. The present study also reveals the plausible cause of the failure of mid-turret Gryphon Alpha FPSO in milder environment than its survival condition.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.1
• /
• pp.79-86
• /
• 2017

This research focused on deciding optimal manufacturing WIP (Work-In-Process) limit for a small production system. Reducing WIP leads to stable capacity, better manufacturing flow and decrease inventory. WIP is the one of the important issue, since it can affect manufacturing area, like productivity and line efficiency and bottlenecks in manufacturing process. Several approaches implemented in this research. First, two strategies applied to decide WIP limit. One is roulette wheel selection and the other one is elite strategy. Second, for each strategy, JIT (Just In Time), CONWIP (Constant WIP), Gated Max WIP System and CWIPL (Critical WIP Loops) system applied to find a best material flow mechanism. Therefore, pull control system is preferred to control production line efficiently. In the production line, the WIP limit has been decided based on mathematical models or expert's decision. However, due to the complexity of the process or increase of the variables, it is difficult to obtain optimal WIP limit. To obtain an optimal WIP limit, GA applied in each material control system. When evaluating the performance of the result, fitness function is used by reflecting WIP parameter. Elite strategy showed better performance than roulette wheel selection when evaluating fitness value. Elite strategy reach to the optimal WIP limit faster than roulette wheel selection and generation time is short. For this reason, this study proposes a fast and reliable method for determining the WIP level by applying genetic algorithm to pull system based production process. This research showed that this method could be applied to a more complex production system.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.45 no.8
• /
• pp.67-74
• /
• 2008

A compact and high-performance AES(Advanced Encryption Standard) encryption/decryption processor is designed by applying various hardware sharing and optimization techniques. In order to achieve minimized hardware complexity, sharing the S-Boxes for round transformation with the key scheduler, as well as merging and reusing datapaths for encryption and decryption are utilized, thus the area of S-Boxes is reduced by 25%. Also, the S-Boxes which require the largest hardware in AES processor is designed by applying composite field arithmetic on $GF(((2^2)^2)^2)$, thus it further reduces the area of S-Boxes when compared to the design based on $GF(2^8)$ or $GF((2^4)^2)$. By optimizing the operation of the 64-bit round transformation and round key scheduling, the round transformation is processed in 3 clock cycles and an encryption of 128-bit data block is performed in 31 clock cycles. The designed AES processor has about 15,870 gates, and the estimated throughput is 412.9 Mbps at 100 MHz clock frequency.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.12 no.1
• /
• pp.75-84
• /
• 2017

It becomes more and more more important for the storage that supports high speed recording and stable access from network environment. As one field of basic science which produces massive astronomical data, VLBI(: Very Long Baseline Interferometer) is now demanding more data writing performance and which is directly related to astronomical observation with high resolution and sensitivity. But most of existing storage are cloud model based for the high throughput of general IT, finance, and administrative service, and therefore it not the best choice for recording of big stream data. Therefore, in this study, we design storage system optimized for high performance of I/O and concurrency. To solve this problem, we implement packet read and writing module through the use of libpcap and pf_ring API on the multi core CPU environment, and build a scalable storage based on software RAID(: Redundant Array of Inexpensive Disks) for the efficient process of incoming data from external network.