• Proceedings of the KIEE Conference
• /
• 2007.11a
• /
• pp.206-207
• /
• 2007

We developed a measurement system for on-line test and evaluation of performances of KHST. The measurement system is composed of software part and hardware part. Perfect interface between multi-users is possible. A new method to measure temperature was applied to the measurement system. By using the system, fault diagnosis and performance evaluation of electric equipment in Korean High Speed Train was conducted during test running.

• Proceedings of the KIEE Conference
• /
• 2005.07b
• /
• pp.1619-1621
• /
• 2005

AC electric railway feeding system classifies into three groups such as normal, TIE and PP feeding method. If the feeding scheme of ac electric railway is changed, current distribution flowing through the line is also modified. And if the current distribution is altered according to the feeding scheme, returned tendency through rail load current or fault current of the train is changed. So the investigation about error correcting method of protective relay is needed considering feeding scheme. In this paper prior to error correcting of protective relay, through interpreting feeding circuit, changes in current distribution of the rail in accordance with feeding would be predicted and analyzed.

• International Journal of Internet, Broadcasting and Communication
• /
• v.11 no.1
• /
• pp.47-53
• /
• 2019

The idea of ensemble learning is to train multiple models, each with the objective to predict or classify a set of results. Most of the errors from a model's learning are from three main factors: variance, noise, and bias. By using ensemble methods, we're able to increase the stability of the final model and reduce the errors mentioned previously. By combining many models, we're able to reduce the variance, even when they are individually not great. In this paper we propose an ensemble model and applied it to classification problem. In iris, Pima indian diabeit and semiconductor fault detection problem, proposed model classifies well compared to traditional single classifier that is logistic regression, SVM and random forest.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.16 no.2
• /
• pp.128-139
• /
• 2017

Railway is one of the public transportation systems along with shipping and aviation. With the recent introduction of high speed train, its proportion is increasing rapidly, which results in the higher risk of catastrophic failures. The wheel bearing to support the train is one of the important components requiring higher reliability and safety in this aspect. Recently, many studies have been made under the name of prognostics and health management (PHM), for the purpose of fault diagnosis and failure prognosis of the bearing under operation. Among them, the most important step is to extract a feature that represents the fault status properly and is useful for accurate remaining life prediction. However, the conventional features have shown some limitations that make them less useful since they fluctuate over time even after the signal de-noising or do not show a distinct pattern of degradation which lack the monotonic trend over the cycles. In this study, a new method for feature extraction is proposed based on the observation of relative frequency energy shifting over the cycles, which is then converted into the feature using the information entropy. In order to demonstrate the method, traditional and new features are generated and compared using the bearing data named FEMTO which was provided by the FEMTO-ST institute for IEEE 2012 PHM Data Challenge competition.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.2
• /
• pp.1-7
• /
• 2016

Although the second driving device of KTX, which consists of the wheel and the axle reduction gears unit, is a mechanically integrated structure, its preventive maintenance (PM) requires two separate intervals due to the different technical requirements. In particular, these subsystems perform attaching and detaching work simultaneously according to the maintenance directive. Therefore, to reduce the unnecessary amount of PM and high logistic availability of the train, it is important to optimize PM with regard to reliability-centered maintenance toward a cost-effective solution. In this study, fault tree analysis and reliability of the subsystems, considering the criticality of the components, were performed using the data derived from field data in maintenance. The cost optimization of the PM was derived from a genetic algorithm considering the target reliability and improvement factor. The cost optimization was derived from a maximum of the fitness function of the individual in generation. The optimal TBO of them using the genetic algorithm was 2.85x106 km, which is reduced to approximately 21% compared to the conventional method.

• Proceedings of the KSR Conference
• /
• 2008.06a
• /
• pp.2173-2178
• /
• 2008

Railroad is the major factor of vibration source in railway vehicles, and it must carefully maintained the original condition to secure the safety and good ride comfort of passenger. Measuring the condition of rail irregularities such as surface, alignment, gauge, twist and cant etc is required to maintain the good performance of railroad. Currently, the various rail irregularity measurement systems(EM120, ROGER1000K and the Total Rail Irregularity Measurement system of Korea High Speed Train) are operated in Korea to estimate the rail irregularity. It is hard to verify the correlation of one rail irregularity data of a measurement system with the other, because they have been adopted different rail irregularity estimation methods. The best method securing the reliability of the irregularity data is the direct confirmation on the ground where the measurement system had detected as a fault section, but it is impossible to apply all sections simultaneously due to limitation of time, labor, cost and equipments. There is a method to secure the reliability of the data by using acceleration values. Rail irregularities, the major factor of vibration in railway vehicle, are transmitted to the vehicle acceleration through masses, springs, dampers and joints as the system dynamic formation. In this study, Transition Function has been adopted by using the rail irregularity and the acceleration value regarding as input & output parameters respectively. It has been verified by comparing the analyzed results with real measured irregularity data from the Total Rail Irregularity Measurement system of Korea High Speed Train. Also various methods has been accomplished to verify the correlation between rail irregularities and acceleration values.

• Journal of Advanced Navigation Technology
• /
• v.19 no.5
• /
• pp.370-378
• /
• 2015

The data acquisition unit, signal acquiring and processing equipment, processes and provides major data of an aircraft such as engine system, power train system, hydraulic system, etc. However, it had lots of failure related to the system during production test flight, and it is necessary to fix them perfectly as soon as possible because of the significance of the equipment. In this paper, it contains failure classification and analysis for each defect element to improve whole software as well as electrical circuit. Particularly, utilizing Fault Injection Method based interworking test, more efficient improvement activity was performed for not only equipment level but also aircraft level, and as a result, it is achieved that considerable betterment of Surion quality and flight safety.

• Journal of Korean Society on Water Environment
• /
• v.24 no.1
• /
• pp.91-98
• /
• 2008

This work was performed to develop a model possible to predict the influent flow and influent components, which are one of main disturbances causing process problems at the operation of municipal wastewater treatment plant. In this study, artificial neural network (ANN) was used in order to develop a model that was able to predict the influent flow, $COD_{Mn}$, SS, TN 1 day-ahead, 2day-ahead and 3 day ahead. Multi-layer feed-forward back-propagation network was chosen as neural network type, and tanh-sigmoid function was used as activation function to transport signal at the neural network. And Levenberg-Marquart (LM) algorithm was used as learning algorithm to train neural network. Among 420 data sets except missing data, which were collected between 2005 and 2006 at field plant, 210 data sets were used for training, and other 210 data sets were used for validation. As result of it, ANN model for predicting the influent flow and components 1-3day ahead could be developed successfully. It is expected that this developed model can be practically used as follows: Detecting the fault related to effluent concentration that can be happened in the future by combining with other models to predict process performance in advance, and minimization of the process fault through the establishment of various control strategies based on the detection result.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.5
• /
• pp.55-67
• /
• 2022

Sensor data can provide fault diagnosis for equipment. However, the cause analysis for fault results of equipment is not often provided. In this study, we propose an explainable convolutional neural network framework for the sensor-based time series classification model. We used sensor-based time series dataset, acquired from vehicles equipped with sensors, and the Wafer dataset, acquired from manufacturing process. Moreover, we used Cycle Signal dataset, acquired from real world mechanical equipment, and for Data augmentation methods, scaling and jittering were used to train our deep learning models. In addition, our proposed classification models are convolutional neural network based models, FCN, 1D-CNN, and ResNet, to compare evaluations for each model. Our experimental results show that the ResNet provides promising results in the context of time series classification with accuracy and F1 Score reaching 95%, improved by 3% compared to the previous study. Furthermore, we propose XAI methods, Class Activation Map and Layer Visualization, to interpret the experiment result. XAI methods can visualize the time series interval that shows important factors for sensor data classification.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.9
• /
• pp.351-358
• /
• 2016

An Electrical Point Machine (NS:New-type Switch), which is equipped and operated at railways in Korea, has been used since the 1960s after being imported from Japan. On the other hand, although the mechanical configuration has improved the position motor control circuit, the electrical connection has not been improved, so NS may have a problem, such as the interlocking system of automatic train operation. In addition, NS is the most vulnerable part in the railway system and a huge train accident may occur due to minor defects. The existing NS wiring set of the circuit controller should be checked only if fixed. Therefore, an excessive inspection time only by a Railroad Signal expert is required. In this paper, the improvement of electrical connection in a NS wiring set, such as the position motor control circuit, was developed and the prototype was installed at Seoul Metro in the distance to go section. The results can be used to help make appropriate adjustments. The improvement of the NS wiring set enhance the maintenance efficiency, passenger service and the stability of the signal system as well as reducing the maintenance cost.