• Journal of Navigation and Port Research
• /
• v.46 no.4
• /
• pp.367-374
• /
• 2022

The purpose of this study was to propose a deep learning algorithm that applies to the fault diagnosis of fuel pumps and purifiers of autonomous ships. A deep learning algorithm reflecting the time dependence of the measured signal was configured, and the failure pattern was trained using the vibration signal, measured in the equipment's regular operation and failure state. Considering the sequential time-dependence of deterioration implied in the vibration signal, this study adopts Conv1D with sliding window computation for fault detection. The time dependence was also reflected, by transferring the measured signal from two-dimensional to three-dimensional. Additionally, the optimal values of the hyper-parameters of the Conv1D model were determined, using the grid search technique. Finally, the results show that the proposed data preprocessing method as well as the Conv1D model, can reflect the sequential dependency between the fault and its effect on the measured signal, and appropriately perform anomaly as well as failure detection, of the equipment chosen for application.

• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.57-73
• /
• 2021

Maintenance and prevention of failure through anomaly detection of ICT infrastructure is becoming important. System monitoring data is multidimensional time series data. When we deal with multidimensional time series data, we have difficulty in considering both characteristics of multidimensional data and characteristics of time series data. When dealing with multidimensional data, correlation between variables should be considered. Existing methods such as probability and linear base, distance base, etc. are degraded due to limitations called the curse of dimensions. In addition, time series data is preprocessed by applying sliding window technique and time series decomposition for self-correlation analysis. These techniques are the cause of increasing the dimension of data, so it is necessary to supplement them. The anomaly detection field is an old research field, and statistical methods and regression analysis were used in the early days. Currently, there are active studies to apply machine learning and artificial neural network technology to this field. Statistically based methods are difficult to apply when data is non-homogeneous, and do not detect local outliers well. The regression analysis method compares the predictive value and the actual value after learning the regression formula based on the parametric statistics and it detects abnormality. Anomaly detection using regression analysis has the disadvantage that the performance is lowered when the model is not solid and the noise or outliers of the data are included. There is a restriction that learning data with noise or outliers should be used. The autoencoder using artificial neural networks is learned to output as similar as possible to input data. It has many advantages compared to existing probability and linear model, cluster analysis, and map learning. It can be applied to data that does not satisfy probability distribution or linear assumption. In addition, it is possible to learn non-mapping without label data for teaching. However, there is a limitation of local outlier identification of multidimensional data in anomaly detection, and there is a problem that the dimension of data is greatly increased due to the characteristics of time series data. In this study, we propose a CMAE (Conditional Multimodal Autoencoder) that enhances the performance of anomaly detection by considering local outliers and time series characteristics. First, we applied Multimodal Autoencoder (MAE) to improve the limitations of local outlier identification of multidimensional data. Multimodals are commonly used to learn different types of inputs, such as voice and image. The different modal shares the bottleneck effect of Autoencoder and it learns correlation. In addition, CAE (Conditional Autoencoder) was used to learn the characteristics of time series data effectively without increasing the dimension of data. In general, conditional input mainly uses category variables, but in this study, time was used as a condition to learn periodicity. The CMAE model proposed in this paper was verified by comparing with the Unimodal Autoencoder (UAE) and Multi-modal Autoencoder (MAE). The restoration performance of Autoencoder for 41 variables was confirmed in the proposed model and the comparison model. The restoration performance is different by variables, and the restoration is normally well operated because the loss value is small for Memory, Disk, and Network modals in all three Autoencoder models. The process modal did not show a significant difference in all three models, and the CPU modal showed excellent performance in CMAE. ROC curve was prepared for the evaluation of anomaly detection performance in the proposed model and the comparison model, and AUC, accuracy, precision, recall, and F1-score were compared. In all indicators, the performance was shown in the order of CMAE, MAE, and AE. Especially, the reproduction rate was 0.9828 for CMAE, which can be confirmed to detect almost most of the abnormalities. The accuracy of the model was also improved and 87.12%, and the F1-score was 0.8883, which is considered to be suitable for anomaly detection. In practical aspect, the proposed model has an additional advantage in addition to performance improvement. The use of techniques such as time series decomposition and sliding windows has the disadvantage of managing unnecessary procedures; and their dimensional increase can cause a decrease in the computational speed in inference.The proposed model has characteristics that are easy to apply to practical tasks such as inference speed and model management.

• Journal of KIISE:Databases
• /
• v.31 no.5
• /
• pp.468-478
• /
• 2004

It is essential in various application areas of data mining and bioinformatics to effectively retrieve the occurrences of interesting patterns from sequence databases. For example, let's consider a network event management system that records the types and timestamp values of events occurred in a specific network component(ex. router). The typical query to find out the temporal casual relationships among the network events is as fellows: 'Find all occurrences of CiscoDCDLinkUp that are fellowed by MLMStatusUP that are subsequently followed by TCPConnectionClose, under the constraint that the interval between the first two events is not larger than 20 seconds, and the interval between the first and third events is not larger than 40 secondsTCPConnectionClose. This paper proposes an indexing method that enables to efficiently answer such a query. Unlike the previous methods that rely on inefficient sequential scan methods or data structures not easily supported by DBMSs, the proposed method uses a multi-dimensional spatial index, which is proven to be efficient both in storage and search, to find the answers quickly without false dismissals. Given a sliding window W, the input to a multi-dimensional spatial index is a n-dimensional vector whose i-th element is the interval between the first event of W and the first occurrence of the event type Ei in W. Here, n is the number of event types that can be occurred in the system of interest. The problem of‘dimensionality curse’may happen when n is large. Therefore, we use the dimension selection or event type grouping to avoid this problem. The experimental results reveal that our proposed technique can be a few orders of magnitude faster than the sequential scan and ISO-Depth index methods.hods.

• The Journal of Korean Academy of Prosthodontics
• /
• v.33 no.3
• /
• pp.467-491
• /
• 1995

In order to analyze the occlusin of intercuspation with maximun bite force, fifteen healthy adult subjects with the ages 23 to 27 were studied(Group1 ; 5-normal occlusion with Angle's Class1, Group2 ; 5-Angle's Class2 malocclusion, Group3 ; 5-Angle’s Class3 malocclusion). Head Position was fixed with occlusal plane paralleling to horizontal line and occlusal registration r cord was made with polyether rubber impression material(Ramitec, ESPECo. West Germany). After all subject were trained for maximum intercuspation at least 5 times, occlusal registration procedure was repeated for this study. Lower posterior rubber occlusal registration records were sliced with 1mm thickness using precision metal sliding channel(Hitachi Ind. Co., Japan). Gross sectional drawings were traced from occluding view of upper and lower posterior teeth on the rubber slices using digitizer, and superimposed for the determination of each drawing distance(Superimposed Rubber Pattern Method). Based on superimposed rubber pattern drawings, total area of occlusal view, sum of each area of the 5 divided occlusal contact provinces and its ratio, total area and number of occlusal contact area were determined to elucidate occlusal stability in the normal and abnormal occlusion groups. The data were analysed by t-test(p=0.05) to determine statistical significance. The obtained results were as follows : 1. Group1 showed the largest standard area with occlusal view of the lower posterior teeth and Group3 showed the smallest area. There was a significant difference between Group2 and Group3(p=0.025), and Gropu1 was not statistically different for both Group2 and Group3. 2. Means and ratio of the under 2.0mm area(D) and ratio showed $197.49mm^2$, 59.76% in Group1, $188,69mm^2$, 56.10% in Group2, and $174.23mm^2$, 55.76% in Group3. The results that Group1 has the most area/ratio and Group3 has the least area/ratio can be considered Group1 is the most advantageous for masticatory effective area, and Group3 is the least adnantageous. 3. Means and ratio of the under 1.0mm area(C) were $198.96mm^2$, 42.65% in Group1, 123.06$mm^2$, 46.58% in Group2, and $92.24mm^2$, 29.52% in Group3. These data means that Group1 is the most advantageous in terms of masticatory effective area and Group3 is the least. 4. Means and ratio of the under 0.5mm area(B) were $86.68mm^2$, 26.68% in Group1, $62.98mm^2$, 18.71% in Group2, and $36.44mm^2$, 11.66% in Group3. These can also be considered Group1 is the most advantageous for masticatory effective area and occlusal stability. 5. Means and ratio of the under 0.05mm area(A) were $30.92mm^2$, 9.21% in Group1, $14.31mm^2$, 4.25% in Group2, and $7.59mm^2$, 2.43% in Group3. The area ratio of the each subject group was(4.1) : (1.9) : (1)and the data of the under 0.05mm area has the intimate relationship with inter-group and intra-group data/ratio. 6. First molar showed the most occlusal contact points in all subject group and Group1 showed somewhat uniformly distributed occlusal contact point except first premolar. In Group2, all contact point in posterior teeth showed significantly reduced distribution except first molar. Group3 showed evenly distributed contace points in first and second molars.

• Journal of Biomedical Engineering Research
• /
• v.24 no.5
• /
• pp.401-410
• /
• 2003

This study investigates the biomechanical efficacies of various cement augmentation techniques with or without pressurization for varying degrees of osteoporotic femur. For this study, a biomechanical analysis using a finite element method (FEM) was undertaken to evaluate surgical procedures, Simulated models include the non-cemented(i.e., hip screw only, Type I), the cement-augmented(Type II), and the cemented augmented with pressurization(Type III) models. To simulate the fracture plane and other interfacial regions, 3-D contact elements were used with appropriate friction coefficients. Material properties of the cancellous bone were varied to accommodate varying degrees of osteoporosis(Singh indices, II∼V). For each model. the following items were analyzed to investigate the effect surgical procedures in relation to osteoporosis of varying degrees : (a) von Mises stress distribution within the femoral head in terms of volumetric percentages. (b) Peak von Mises stress(PVMS) within the femoral head and the surgical constructs. (c) Maximum von Mises strain(MVMS) within the femoral head, (d) micromotions at the fracture plane and at the interfacial region between surgical construct and surrounding bone. Type III showed the lowest PVMS and MVMS at the cancellous bone near the bone-construct interface regardless of bone densities. an indication of its least likelihood of construct loosening due to failure of the host bone. Particularly, its efficacy was more prominent when the bone density level was low. Micromotions at the interfacial surgical construct was lowest in Type III. followed by Type I and Type II. They were about 15-20% of other types. which suggested that pressurization was most effective in limiting the interfacial motion. Our results demonstrated the cement augmentation with hip screw could be more effective when used with pressurization technique for the treatment of intertrochanteric fractures. For patients with low bone density. its effectiveness can be more pronounced in limiting construct loosening and promoting bone union.

• Proceedings of the Korean Geotechical Society Conference
• /
• 1998.05a
• /
• pp.35-81
• /
• 1998

Distinct Element Method(DEM) has a great advantage to model the discontinuous behaviour of jointed rock masses such as rotation, sliding, and separation of rock blocks. Geometrical data of joints by a field monitoring is not enough to model the jointed rock mass though the results of DE analysis for the jointed rock mass is most sensitive to the distributional properties of joints. Also, it is important to use a properly joint law in evaluating the stability of a jointed rock mass because the joint is considered as the contact between blocks in DEM. In this study, a stochastic modelling technique is developed and the dilatant rock joint is numerically modelled in order to consider th geometrical and mechanical properties of joints in DE analysis. The stochastic modelling technique provides a assemblage of rock blocks by reproducing the joint distribution from insufficient joint data. Numerical Modelling of joint dilatancy in a edge-edge contact of DEM enable to consider not only mechanical properties but also various boundary conditions of joint. Preprocess Procedure for a stochastic DE model is composed of a statistical process of raw data of joints, a joint generation, and a block boundary generation. This stochastic DE model is used to analyze the effect of deviations of geometrical joint parameters on .the behaviour of jointed rock masses. This modelling method may be one tool for the consistency of DE analysis because it keeps the objectivity of the numerical model. In the joint constitutive law with a dilatancy, the normal and shear behaviour of a joint are fully coupled due to dilatation. It is easy to quantify the input Parameters used in the joint law from laboratory tests. The boundary effect on the behaviour of a joint is verified from shear tests under CNL and CNS using the numerical model of a single joint. The numerical model developed is applied to jointed rock masses to evaluate the effect of joint dilation on tunnel stability.

• Journal of Chest Surgery
• /
• v.39 no.4 s.261
• /
• pp.275-280
• /
• 2006

Background: Aortopulmonary window (APW) is a very rare congenital heart anomaly, often associated with other cardiac anomalies. It causes a significant systemic to pulmonary artery shunt, which requires early surgical correction. Accurate diagnosis and surgical correction will bring good outcomes. The purpose of this study was to describe our 20-year experience of aortopulmonary window. Material and Method: Between March 1985 and January 2005, 16 patients with APW underwent surgical repair. Mean age at operation was $157.8{\pm}245.3$ ($15.0{\sim}994.0$) days and mean weight was $4.8{\pm}2.5$ ($1.7{\sim}10.7$) kg. Patent ductus arteriosus (8), atrial septal defect (7), interruptedaortic arch (5), ventricular septal defect (4), patent foramen ovate (3), tricuspid valve regurgitation (3), mitral valve regurgitation (2), aortic valve regurgitation (1), coarctation of aorta (1), left superior vena cavae (1), and dextrocardia (1) were associated. Repair methods included 1) division of the APW with primary closure or patch closure of aorta and pulmonary artery primary closure or patch closure (11) and 2) intra-arterial patch closure (3). 3) Division of the window and descending aorta to APW anastomosis (2) in the patients with interrupted aortic arch or coarctation. Result: There was one death. The patient had 2.5 cm long severe tracheal stenosis from carina with tracheal bronchus supplying right upper lobe. The patient died at 5th post operative day due to massive tracheal bleeding. Patients with complex aortopulmonary window had longer intensive care unit and hospital stay and showed more morbidities and higher reoperation rates. 5 patients had reoperations due to left pulmonary artery stenosis (4), right pulmonary artery stenosis (2), and main pulmonary artery stenosis (1). The mean follow-up period was $6.8{\pm}5.6$ (57.0 days$\sim$16.7 years)years and all patients belonged to NYHA class 1. Conclusion: With early and prompt correction of APW, excellent surgical outcome can be expected. However, optimal surgical method needs to be established to decrease the rate of stenosis of pulmonary arteries.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.30 no.6
• /
• pp.306-318
• /
• 2018

In order to evaluate the seismic capacity of massive vertical type breakwaters which have intensively been deployed along the coast of South Korea over the last two decades, we carry out the preliminary numerical simulation against the PoHang, GyeongJu, Hachinohe 1, Hachinohe 2, Ofunato, and artificial seismic waves based on the measured time series of ground acceleration. Numerical result shows that significant sliding can be resulted in once non-negligible portion of seismic energy is shifted toward the longer period during its propagation process toward the ground surface in a form of shear wave. It is well known that during these propagation process, shear waves due to the seismic activity would be amplified, and non-negligible portion of seismic energy be shifted toward the longer period. Among these, the shift of seismic energy toward the longer period is induced by the viscosity and internal friction intrinsic in the soil. On the other hand, the amplification of shear waves can be attributed to the fact that the shear modulus is getting smaller toward the ground surface following the descending effective stress toward the ground surface. And the weakened intensity of soil as the number of attacking shear waves are accumulated can also contribute these phenomenon (Das, 1993). In this rationale, we constitute the numerical model using the model by Hardin and Drnevich (1972) for the weakened shear modulus as shear waves go on, and shear wave equation, in the numerical integration of which $Newmark-{\beta}$ method and Modified Newton-Raphson method are evoked to take nonlinear stress-strain relationship into account. It is shown that the numerical model proposed in this study could duplicate the well known features of seismic shear waves such as that a great deal of probability mass is shifted toward the larger amplitude and longer period when shear waves propagate toward the ground surface.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• Journal of Korean Society of Forest Science
• /
• v.45 no.1
• /
• pp.11-25
• /
• 1979

There was much mass movement at many different mountain side of Peong Chang area in Kwangwon province by the influence of heavy rainfall through August/4 5, 1979. This study have done with the fact observed through the field survey and the information of the former researchers. The results are as follows; 1. Heavy rainfall area with more than 200mm per day and more than 60mm per hour as maximum rainfall during past 6 years, are distributed in the western side of the connecting line through Hoeng Seong, Weonju, Yeongdong, Muju, Namweon and Suncheon, and of the southern sea side of KeongsangNam-do. The heavy rain fan reason in the above area seems to be influenced by the mouktam range and moving direction of depression. 2. Peak point of heavy rainfall distribution always happen during the night time and seems to cause directly mass movement and serious damage. 3. Soil mass movement in Peongchang break out from the course sandy loam soil of granite group and the clay soil of lime stone and shale. Earth have moved along the surface of both bedrock or also the hardpan in case of the lime stone area. 4. Infiltration seems to be rapid on the both bedrock soil, the former is by the soil texture and the latter is by the crumb structure, high humus content and dense root system in surface soil. 5. Topographic pattern of mass movement spot is mostly the concave slope at the valley head or at the upper part of middle slope which run-off can easily come together from the surrounding slope. Soil profile of mass movement spot has wet soil in the lime stone area and loose or deep soil in the granite area. 6. Dominant slope degree of the soil mass movement site has steep slope, mostly, more than 25 degree and slope position that start mass movement is mostly in the range of the middle slope line to ridge line. 7. Vegetation status of soil mass movement area are mostly fire field agriculture area, it's abandoned grass land, young plantation made on the fire field poor forest of the erosion control site and non forest land composed mainly grass and shrubs. Very rare earth sliding can be found in the big tree stands but mostly from the thin soil site on the un-weatherd bed rock. 8. Dangerous condition of soil mass movement and land sliding seems to be estimated by the several environmental factors, namely, vegetation cover, slope degree, slope shape and position, bed rock and soil profile characteristics etc. 9. House break down are mostly happen on the following site, namely, colluvial cone and fan, talus, foot area of concave slope and small terrace or colluvial soil between valley and at the small river side Dangerous house from mass movement could be interpreted by the aerial photo with reference of the surrounding site condition of house and village in the mountain area 10. As a counter plan for the prevention of mass movement damage the technics of it's risk diagnosis and the field survey should be done, and the mass movement control of prevention should be started with the goverment support as soon as possible. The precautionary measures of house and village protection from mass movement damage should be made and executed and considered the protecting forest making around the house and village. 11. Dangerous or safety of house and village from mass movement and flood damage will be indentified and informed to the village people of mountain area through the forest extension work. 12. Clear cutting activity on the steep granite site, fire field making on the steep slope, house or village construction on the dangerous site and fuel collection in the eroded forest or the steep forest land should be surely prohibited When making the management plan the mass movement, soil erosion and flood problem will be concidered and also included the prevention method of disaster.