• Korean Journal of Artificial Intelligence
• /
• v.11 no.1
• /
• pp.1-7
• /
• 2023

In this paper, a study was conducted to compare the prediction model of cardiovascular disease occurrence. It is the No.1 disease that accounts for 1/3 of the world's causes of death, and it is also the No. 2 cause of death in Korea. Primary prevention is the most important factor in preventing cardiovascular diseases before they occur. Early diagnosis and treatment are also more important, as they play a role in reducing mortality and morbidity. The Results of an experiment using Azure ML, Logistic Regression showed 88.6% accuracy, Decision Tree showed 86.4% accuracy, and Support Vector Machine (SVM) showed 83.7% accuracy. In addition to the accuracy of the ROC curve, AUC is 94.5%, 93%, and 92.4%, indicating that the performance of the machine learning algorithm model is suitable, and among them, the results of applying the logistic regression algorithm model are the most accurate. Through this paper, visualization by comparing the algorithms can serve as an objective assistant for diagnosis and guide the direction of diagnosis made by doctors in the actual medical field.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.36 no.5
• /
• pp.474-481
• /
• 2023

This paper is an experimental study on the optimal operating conditions of direct charging type electrospray for particulate matter collection. To perform the research, a direct charging type electrospray visualization system was configured to photograph the spray shape of microdroplets, and experiments were performed with varying electrode distance, flow rate, and applied voltage, which are the main factors affecting the particulate matter collection efficacy. Through image processing, the total number of microdroplets according to each condition was analyzed, and the number of microdroplets with a diameter of 1.5 mm or less was confirmed. In addition, by calculating the number of microdroplets per power consumption according to the applied voltage, the optimal operating conditions were derived in terms of energy consumption efficacy, and the microdroplet size distribution was analyzed under the optimal operating conditions. As a result of the experiment, it was confirmed that the optimal operating condition was at a flow rate of 10 mL/min and a voltage of -20 kV in case of 5 mm electrode distance, and at a flow rate of 15 mL/min and a voltage of -30 kV in case of 100 mm electrode distance.

• Smart Structures and Systems
• /
• v.33 no.4
• /
• pp.253-262
• /
• 2024

Current-based mechanical fault diagnosis is more convenient and low cost since additional sensors are not required. However, it is still challenging to achieve this goal due to the weak fault information in current signals. In this paper, a dual-loss convolutional neural network (DLCNN) is proposed to implement the intelligent bearing fault diagnosis via current signals. First, a novel similarity loss (SimL) function is developed, which is expected to maximize the intra-class similarity and minimize the inter-class similarity in the model optimization operation. In the loss function, a weight parameter is further introduced to achieve a balance and leverage the performance of SimL function. Second, the DLCNN model is constructed using the presented SimL and the cross-entropy loss. Finally, the two-phase current signals are fused and then fed into the DLCNN to provide more fault information. The proposed DLCNN is tested by experiment data, and the results confirm that the DLCNN achieves higher accuracy compared to the conventional CNN. Meanwhile, the feature visualization presents that the samples of different classes are separated well.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.14 no.6
• /
• pp.528-535
• /
• 2004

Ultrasonic Vibrator is designed to achieve the maximum vibration amplitude at 30 kHz by in-cluding a horn (diameter, 40 mm), mechanical vibration amplifier at the top of the ultrasonic vibrator in the system and making the complete system resonate. In addition, it is experimentally visualized by particle imaging velocimetry (PIV) that the acoustic streaming velocity in the gap is at maximum when the gap between the ultrasonic vibrator and stationary plate agrees with the multiples of half-wavelength of the ultrasonic wave. This fact results from the resonance of the sound wave and the theoretical analysis of that is also accomplished and verified by experiment. It is observed that the magnitude of the acoustic streaming dependent upon the gap between the ultrasonic vibrator and stationary plate possibly changes due to the measurement of the average velocity fields of the acoustic streaming induced by the ultrasonic vibration at resonance and non-resonance. There exists extremely small average velocity at non-resonant gaps while the relatively large average velocity exists at resonant gaps compared with non-resonant gaps. It also reveals that there should be larger axial turbulent intensity at the hub region of the vibrator and at the edge of it in the resonant gap where the air streaming velocity is maximized and the flow phenomena is conspicuous than that at the other region. Because the variation of the acoustic streaming velocity at resonant gap is more distinctive than that at non-resonant gap, shear stress increases more in the resonant gap and is also maximized at the center region of the vibrator except the local position of center (r〓0). At the non-resonant gap there should be low values of vorticity distribution, but in contrast to the non-resonant gap, high and negative values of it exist at the center region of the vibrator with respect to the radial direction and in the vicinity of the middle region with respect to the axial direction. Acoustic streaming is noise-free due to the ultrasonic vibration and maintenance-free because of the absence of moving parts. Moreover, the proposed method by acoustic streaming can be utilized to the nano and micro-electro mechanical systems as a driving mechanism in addition to the augmentation of the streaming velocity.

• Journal of KIISE:Computing Practices and Letters
• /
• v.9 no.6
• /
• pp.646-661
• /
• 2003

It is not easy that we visualize the large volume data stored in the every client computers of the web environment. One solution is as follows. First we compress volume data, second store that in the database server, third transfer that to client computer, fourth visualize that with direct-volume-rendering in the client computer. In this case, we usually use wavelet transform for compressing large data. This paper reports the experiments for acquiring the wavelet bases and the compression ratios fit for the above processing paradigm. In this experiments, we compress the volume data Engine, CThead, Bentum into 50%, 10%, 5%, 1%, 0.1%, 0.03% of the total data respectively using Harr, Daubechies4, Daubechies12 and Daubechies20 wavelets, then visualize that with direct-volume-rendering, afterwards evaluate the images with eyes and image comparison metrics. When compression ratio being low the performance of Harr wavelet is better than the performance of the other wavelets, when compression ratio being high the performance of Daubechies4 and Daubechies12 is better than the performance of the other wavelets. When measuring with eyes the good compression ratio is about 1% of all the data, when measuring with image comparison metrics, the good compression ratio is about 5-10% of all the data.

• Journal of Hydrogen and New Energy
• /
• v.23 no.4
• /
• pp.364-372
• /
• 2012

Transient mixing states of two different fuel oils, dimethylformamide (DMF) oil and JetA1 oil, were investigated by using a color image processing and a neural network. A tank ($D{\times}H$, $310{\times}370mm$) was filled with JetA1 oil. The DMF oil was filled at a top tank, and was mixed with the JetA1 oil in the tank mixing tank via a sudden opening which was performed by nitrogen gas with 1.9 bar. An impeller was rotated with 700 rpm for mixing enhancements of the two fuel oils. To visualize the mixing state of the DMF oil with the JetA1 oil, the DMF oil was coated with Rhodamine B whose color was red. A LCD monitor was used for uniform illumination. The color changes of the DMF oil were captured by a camcoder and the images were transferred to a host computer for quantifying the information of color changes. The color images of two mixed oils were captured with the camcoder. The R, G, B color information of the captured images was used to quantify the concentration of the DMF oil. To quantify the concentration of the DMF oil in the JetA1 oil, a calibration of color-to-concentration was carried out before the main experiment was done. Transient mixing states of DMF oil with the JetA1 oil since after the sudden infiltration were quantified and characterized with the constructed visualization technique.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.27 no.4
• /
• pp.278-285
• /
• 1991

The authors propose to use the slot system in order to improve of the efficiency for the cambered otter boards. The experiment is divided into 2 parts, one is the efficiency model test, and the other is the visualization model test. The hydrodynamic characteristics of the model otter boards were tested by efficiency model test to measure the shearing, drag force of the models and visualization test using hydrogen bubble method to observe the streak-line and time-line of flow around the models, and milk spout method to observe the separation zone in the wake behind the models. This study tested for 5 models such ad without slot, slot position 0.2C, 0.4C, 0.6C and 0.8C. The results obtained are as follows: \circled1 The maximum C sub(L) of model otter board with slot position 0.6C in attack angle 27$^{\circ}$ was the highest of all models, it's value was 1.59. \circled2 In general, the L/D ratio of the one slot otter boards were 16~28% higher than otter board without slot. \circled3 The slot position 0.6C was better than any other slot position, and it's conformed by visiualization. \circled4 As to the model otter board with slot position 0.6C, flow speed of the back side was faster 1.3 to 1.7 times than in the front side. \circled5 The size of the separated zone in case of the model otter board with 0.6C was smaller than that of any other models.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.1
• /
• pp.1-9
• /
• 2022

Non-small cell lung cancer (NSCLC) accounts for a high proportion of 85% among all lung cancer and has a significantly higher mortality rate (22.7%) compared to other cancers. Therefore, it is very important to predict the prognosis after surgery in patients with non-small cell lung cancer. In this study, the types of preoperative chest CT image patches for non-small cell lung cancer patients with tumor as a region of interest are diversified into five types according to tumor-related information, and performance of single classifier model, ensemble classifier model with soft-voting method, and ensemble classifier model using 3 input channels for combination of three different patches using pre-trained ResNet and EfficientNet CNN networks are analyzed through misclassification cases and Grad-CAM visualization. As a result of the experiment, the ResNet152 single model and the EfficientNet-b7 single model trained on the peritumoral patch showed accuracy of 87.93% and 81.03%, respectively. In addition, ResNet152 ensemble model using the image, peritumoral, and shape-focused intratumoral patches which were placed in each input channels showed stable performance with an accuracy of 87.93%. Also, EfficientNet-b7 ensemble classifier model with soft-voting method using the image and peritumoral patches showed accuracy of 84.48%.

• International Journal of Automotive Technology
• /
• v.6 no.1
• /
• pp.1-8
• /
• 2005

Recently, the HSDI (High Speed Direct Injection) diesel engine has been spotlighted as a next generation engine because it has a good potential for high thermal efficiency and fuel economy. This study was carried out to investigate the in-cylinder flow characteristics generated in a HSDI diesel engine with a 4-valve type cylinder head. The four kinds of cylinder head were manufactured to elucidate the effect of intake port geometry on the in-cylinder flow characteristics. The steady flow characteristics such as coefficient of flow rate $(C_{f})$, swirl ratio (Rs), and mass flow rate (m,) were measured by the steady flow test rig and the unsteady flow velocity within a cylinder was measured by PIV. In addition, the in-cylinder flow patterns were visualized by the visualization experiment and these results were compared with simulation results calculated by the commercial CFD code. The steady flow test results indicated that the mass flow rate of the cylinder head with a short distance between the two intake ports is $13\%$ more than that of the other head. However, the non-dimensional swirl ratio is decreased by approximately $15\%$. As a result of in-cylinder flow characteristics obtained by PIV and CFD calculation, we found that the swirl center was eccentric from the cylinder center and the position of swirl center was changed with crank angle. As the piston moves to near the TDC, the swirl center corresponded to the cylinder center and the velocity distribution became uniform. In addition, the results of the calculation are in good agreement with the experimental results.

• Asian-Australasian Journal of Animal Sciences
• /
• v.17 no.9
• /
• pp.1227-1231
• /
• 2004

Pig oocytes contain high levels of lipids in the ooplasm, which reduces the visibility of the germinal vesicle (GV) under microscopic examination. Therefore, the purposes of this study were to investigate the effects of relative centrifugation force (RCF) on the visibility and maturation rates of the GV stage oocytes after centrifugation. In Experiment 1, cumulus-oocyte-complexes (COCs) were collected from slaughterhouse ovaries and randomly allocated to different RCFs (3,000 rpm: 970 g; 6,000 rpm: 3,900 g; or 10,000 rpm: 10,840 g) for 10 or 20 min. Percentages of visible GV were 76-79% in the oocytes centrifuged with 10,000 rpm, which were significantly higher (p<0.01) than those with 3,000 and 6,000 rpm. No significant differences in GV visibility were observed among oocytes with different lengths of centrifugation (p<0.05) regardless of the RCFs. In esperiment 2, the maturation rate of the oocyte was found significantly lower in the 20 min than in the 10 min group received 10,840 g of RCF (30 vs. 75%, p<0.05). In conclusion, the GV of porcine oocytes can be clearly visible by centrifugation at 10,840 g for 10 min without compromising their subsequent maturation rates and a longer centrifugation time (20 min) had no beneficial influence on the visibility of GV stage pig oocytes.