Journal of the Korean Society for Precision Engineering

Development of Transformer-based Model for Prediction of PEMFC Remaining Useful Life: Da Hye Geum, Hyeon Do Han, Hyunjun Yang, Heejun Shin, Suk Won Cha, Gu Young Cho; J. Korean Soc. Precis. Eng. 2025;42(12):981-986.
Published online December 1, 2025; DOI: https://doi.org/10.7736/JKSPE.025.00015

Abstract
PDF

A Transformer model to predict the remaining useful life of a fuel cell, which has demonstrated superior performance in analyzing time series data. The dataset was created from long-term performance evaluation experiments conducted in rated power mode, with measurements taken every 10 hours. We preprocessed the raw data using a moving average, allocating 70% for training and 30% for evaluation. The model's performance, evaluated through MAE, MSE, and MAPE, was excellent. The fuel cell's critical voltage, defined as 94.5% of its initial voltage, was measured at 0.719 V. During the experimental run, the actual critical time was 106.6 hours, while the model predicted 106.8 hours, resulting in a 0.19% error. Since the predictions were based on data collected up to 93 hours, the estimated remaining life was 13.8 hours.

349 View
21 Download

Techniques for Tool Life Prediction and Autonomous Tool Change Using Real-time Process Monitoring Data: Seong Hun Ha, Min-Suk Park, Hoon-Hee Lee; J. Korean Soc. Precis. Eng. 2025;42(11):949-958.
Published online November 1, 2025; DOI: https://doi.org/10.7736/JKSPE.025.077

Abstract
PDF

Materials such as titanium alloys, nickel alloys, and stainless steels are difficult to machine due to low thermal conductivity, work hardening, and built-up edge formation, which accelerate tool wear. Frequent tool changes are required, often relying on operator experience, leading to inefficient tool use. While modern machine tools include intelligent tool replacement systems, many legacy machines remain in service, creating a need for practical alternatives. This study proposes a method to autonomously determine tool replacement timing by monitoring machining process signals in real time, enabling automatic tool changes even on conventional machines. Tool wear is evaluated using current and vibration sensors, with the replacement threshold estimated from the maximum current observed in an initial user-defined interval. When real-time signals exceed this threshold, the system updates controller variables to trigger tool changes. Results show vibration data are more sensitive to wear, whereas current data provide greater stability. These findings indicate that a hybrid strategy combining both sensors can enhance accuracy and reliability of tool change decisions, improving machining efficiency for difficult-to-cut materials.

81 View
4 Download

Study on the Life Prediction Analysis Methodology of Worm Gear for the TV Driving Mechanism: Dong Uk Kim, Tae Bae Kim, Il Joo Chang; J. Korean Soc. Precis. Eng. 2025;42(8):595-602.
Published online August 1, 2025; DOI: https://doi.org/10.7736/JKSPE.025.020

Abstract
PDF

In the case of TV products, space constraints and design requirements make it advantageous to use a worm gear that has a small volume and a self-locking function. Single enveloping worm gear teeth are classified as ZA, ZN, ZK, ZI, and ZC according to international standards. However, combining worm shafts and worm wheels with different tooth profiles can significantly worsen meshing transmission errors and reduce the lifespan of the worm gear. Despite these challenges, due to processing limitations, ease of manufacturing, and cost reduction, combinations of worm shafts and worm wheels with different tooth profiles are still considered. In this study, we confirmed the meshing transmission error for a worm gear that combined a ZA tooth shape worm shaft with a ZI tooth shape worm wheel. Additionally, we examined the contact stress and fatigue life characteristics of the material combinations using finite element analysis (FEM).

71 View
6 Download

Prediction of Steel Plate Deformation in Line Heating Process Using Conditional Generative Adversarial Network (cGAN): Young Soo Yang, Kang Yul Bae; J. Korean Soc. Precis. Eng. 2025;42(6):411-420.
Published online June 1, 2025; DOI: https://doi.org/10.7736/JKSPE.025.010

Abstract
PDF

This study proposed a conditional generative adversarial network (cGAN) model for predicting steel plate deformation based on heating line positions in a line heating process. A database was constructed by performing finite element analysis (FEA) to establish relationships between heating line positions and deformation shapes. Deformation shapes were converted into color map images. Heating line positions were used as conditional labels for training and validating the proposed model.
During the training process, generator and discriminator loss values, along with MSE and R² metrics, converged stably, demonstrating that generated images closely resembled the actual data. Validation results showed that predicted deformation magnitudes had an average relative error of approximately 3% and a maximum error of less than 7%. These findings confirm that the proposed model can effectively predict steel plate deformation shapes based on heating line positions in the line heating process, making it a reliable predictive tool for this application.

61 View
12 Download

Degradation Pattern Classification for Predicting Remaining Useful Life of Rolling-element Bearings: Yoonjae Lee, Dongju Seo, Sangyoon Lee, Changwoo Lee; J. Korean Soc. Precis. Eng. 2024;41(12):973-990.
Published online December 1, 2024; DOI: https://doi.org/10.7736/JKSPE.024.101

Abstract
PDF

In continuous-process systems, failures of rolling-element bearings typically cause accidents, reduced productivity, and production-related financial losses. Therefore, predicting both the lifespan of rolling-element bearings and their replacement time is crucial for preventing machine system failures. Accordingly, numerous studies have reported various machine and deep learning classifiers for predicting the lifespan of bearings. However, these studies did not consider degradation trends of bearings. Thus, this study aimed to develop an algorithm to predict the lifespan of a bearing by considering its degradation trend. A vibration dataset of bearings was obtained at low and high speeds. Using a second-order curve-fitting model, various degradation patterns in the dataset were classified. Appropriate time-domain or frequency-domain feature variables applicable to the design of a classifier were determined according to classified patterns. In addition, the classifier was trained using multiple bidirectional long short-term memories. Finally, the performance of the developed classifier was verified experimentally.

40 View
0 Download

Risk Prediction in Daily Activities and Falls based on Deep Learning: Seunghee Lee, Bummo Koo, Sumin Yang, Dongkwon Kim, Youngho Kim; J. Korean Soc. Precis. Eng. 2023;40(12):1003-1009.
Published online December 1, 2023; DOI: https://doi.org/10.7736/JKSPE.023.102

Abstract
PDF

Predicting fall risk is necessary for rescue and accident prevention in the elderly. In this study, deep learning regression models were used to predict the acceleration sum vector magnitude (SVM) peak value, which represents the risk of a fall. Twenty healthy adults (aged 22.0±1.9 years, height 164.9±5.9 cm, weight 61.4±17.1 kg) provided data for 14 common daily life activities (ADL) and 11 falls using IMU (Inertial Measurement Unit) sensors (Movella Dot, Netherlands) at the S2. The input data includes information from 0.7 to 0.2 seconds before the acceleration SVM peak, encompassing 6-axis IMU data, as well as acceleration SVM and angular velocity SVM, resulting in a total of 8 feature vectors used to model training. Data augmentations were applied to solve data imbalances. The data was split into a 4 : 1 ratio for training and testing. The models were trained using Mean Squared Error (MSE) and Mean Absolute Error (MAE). The deep learning model utilized 1D-CNN and LSTM. The model with data augmentation exhibited lower error values in both MAE (1.19 g) and MSE (2.93g²). Low-height falls showed lower predicted acceleration peak values, while ADLs like jumping and sitting showed higher predicted values, indicating higher risks.

32 View
0 Download

Analysis of the Possibility of Classifying Field Hockey Positions Using Random-forest: Ji Eung Kim, Seung Hun Lee, Hoi Deok Jeong; J. Korean Soc. Precis. Eng. 2023;40(7):527-532.
Published online July 1, 2023; DOI: https://doi.org/10.7736/JKSPE.023.055

Abstract
PDF

The purpose of this study was to check the position classification prediction rate based on the movement data of field hockey players using the random forest algorithm. In order to achieve the purpose of this study, movement data were collected using wearable devices in 15 practice matches. The collected information was then analyzed using the Random Forest algorithm, one of the ensemble techniques, with Python, a high-level, general-purpose programming language. As a result of this study, first, the position classification prediction rate was 52.4±3.3% when data measured by GPS sensors were used. Second, when using the data measured by an inertial measurement unit (IMU) sensor, the position classification prediction rate was 50.8±2.4%. Third, when both Global Positioning System (GPS) and IMU data were used, the position classification prediction rate was 55.6±2.0%. As a result of the study, it showed that the prediction rate was the highest when both GPS and IMU data were used.

49 View
0 Download

A Study on the Prediction Model of the Radius of Curvature of the Subtle Feature of the Automotive Parts for Different Forming Conditions: Jae-Hyeong Yu, Kyu-Seok Jung, Yunchan Chung, Chang-Whan Lee; J. Korean Soc. Precis. Eng. 2023;40(1):49-55.
Published online January 1, 2023; DOI: https://doi.org/10.7736/JKSPE.022.101

Abstract
PDF

The subtle feature is one of the characteristic lines and represents the most noticeable line in the automotive panel. In this study, we proposed a method to predict the radius of curvature of products according to the material, its thickness, its punch angle, and its punch radius. The radius of curvature was divided into three regions, namely, the non-linear, transition, and linear regions. In the non-linear region, the prediction model for the radius of curvature with different forming conditions was derived using the finite element analysis. In the linear region, the radius of curvature was assumed to be the sum of the punch radius and the thickness of the material. In the transition region, a model connecting two regions (Non-linear and linear region) was developed based on the continuity condition. The prediction model presented a very small RMSE with the value of 0.314 mm. Using the prediction model, the radius of curvature with various forming variables could be predicted and the required radius of punch, to obtain a certain value of the radius of curvature, could be precisely predicted.

52 View
0 Download

A Study on 3D Printing Conditions Prediction Model of Bone Plates Using Machine Learning: Song Yeon Lee, Yong Jeong Huh; J. Korean Soc. Precis. Eng. 2022;39(4):291-298.
Published online April 1, 2022; DOI: https://doi.org/10.7736/JKSPE.021.096

Abstract
PDF

Bone plates made of biodegradable polymers have been used to fix broken bones. 3D printers are used to produce the bone plates for fracture fixing in the industry. The dimensional accuracy of the product printed by a 3D printer is less than 80%. Fracture fixing plates with less than 80% dimensional accuracy cause problems during surgery. There is an urgent need to improve the dimensional accuracy of the product in the industry. In this paper, a methodology using machine learning was proposed to improve the dimensional accuracy. The proposed methodology was evaluated through case studies. The results predicted by the machine learning methodology proposed in this paper and the experimental results were compared through the experiment. After verification, results of the proposed prediction model and the experimental results were in good agreement with each other.

37 View
0 Download

Fatigue Life Analysis and Experimental Study of the Input Shaft of 6-Speed Automatic Transmission: Jianhua Lv, Xing Zhong, Rui Zhou, Zhen Qin, Qi Zhang, Sungki Lyu; J. Korean Soc. Precis. Eng. 2020;37(8):607-613.
Published online August 1, 2020; DOI: https://doi.org/10.7736/JKSPE.020.011

Abstract
PDF

The input shaft of gearbox usually bears a cyclic variation of torque, which may lead to the risk of experiencing a fatigue fracture. To evaluate the fatigue life accurately and identify the weak parts, the ANSYS is used to simulate the torsional fatigue of the input shaft for the gearbox, and the fatigue life of the weak part is obtained, which is then tested and verified by the torsional fatigue testing in the MTS torsional fatigue test rig. The test results show that the maximum difference is 14% between the calculated life and the testing results, indicating that the simulation value can reflect the actual fatigue life accurately. Notably, the cracks appear in the large oil holes, and its life is mainly concentrated in the crack initiation stage, accounting for 99.2% of the total life. The analysis results show that the fatigue life of the software simulation has the guiding significance for the life evaluation. The fatigue life of the shaft can be quickly calculated by the simulation to reduce the number of fatigue tests and achieve cost-effectiveness.

61 View
1 Download

Integrated Analytical Approach for the Definition and the Control of the Radial-Axial Ring Rolling Process: Do yeon Kim, Luca Quagliato; J. Korean Soc. Precis. Eng. 2019;36(9):821-835.
Published online September 1, 2019; DOI: https://doi.org/10.7736/KSPE.2019.36.9.821

Abstract
PDF

This research aims to provide a useful algorithm for the prediction of the geometrical expansion of flat rings in the radialaxial ring rolling process in case of multiple variations of the mandrel feeding speed during the process. The proposed algorithm was subjected to a 2-phases validation process, where results were compared with those of laboratory experiments, conducted at 150℃ on rings made of AA-1070 and AA-6061 aluminum alloys, and with numerical simulations, considering 7 different rings with outer diameter ranging from 800 to 2000 ㎜ and made of 42CrMo4 steel alloy, Ti6Al4V titanium alloy and AA-6061 aluminum alloys. In the first and second validation phases, the maximum deviation in the estimation of the outer diameter of the ring has been calculated in 1.7% and 6.82%, respectively. According to the results of the validation, the proposed algorithm is able to properly predict the geometrical expansion of the ring for multiple variations of the mandrel feeding speed during the process and has good accordance with both relatively small and large rings.

43 View
0 Download

Prediction of Low-Cycle Fatigue Life of In738LC Using Plastic Strain Energy Density: Sung Uk Wee, Chang Sung Seok, Jae Mean Koo, Jeong Min Lee; J. Korean Soc. Precis. Eng. 2019;36(4):401-406.
Published online April 1, 2019; DOI: https://doi.org/10.7736/KSPE.2019.36.4.401

Abstract
PDF

Gas turbine blades are important parts of a power plant, and thus, it is necessary to be able to predict the low-cycle fatigue life of the blades. In this study, a low-cycle fatigue test of In738LC, which is used primarily in gas turbine blade manufacture, was performed at various high temperatures (750oC, 800oC, and 850oC). From the test results, the stressstrain curve and the stress-strain hysteresis loop were obtained. It was established that In738LC has no strain hardening or softening. The life prediction equations for low-cycle fatigue were derived using the Coffin-Manson equation and the energy model. In conclusion, one equation for predicting the life low-cycle fatigue was obtained using the energy level with temperature as the varying factor.

44 View
0 Download

Low Cycle Fatigue Characteristics of a Ni-Based Single Crystal Superalloy CMSX-4 at Elevated Temperature: Jae Gu Choi, Chang-Sung Seok, Sung Uk Wee, Eui-Suck Chung, Byoung-Gwan Yun, Suk-Hwan Kwon; J. Korean Soc. Precis. Eng. 2019;36(3):271-279.
Published online March 1, 2019; DOI: https://doi.org/10.7736/KSPE.2019.36.3.271

Abstract
PDF

Isothermal low cycle fatigue (LCF) behavior of a crystal nickel-based superalloy CMSX-4, a material for high-pressure turbine first stage rotor blade, was investigated at elevated temperatures. Strain-controlled LCF tests were performed under various test conditions, such as mechanical strain amplitude. Stress response and cyclic deformation were investigated, and equations of LCF life prediction were derived through the Coffin-Manson method. In addition, fatigue-induced fracture mechanism and microstructural evolution were investigated, using scanning electron microscopy (SEM). Results revealed that cyclic behavior of the CMSX-4 superalloy, was characterized by cyclic softening with increasing number of cycles at 800oC and 900oC. LCF of the CMSX-4 superalloy at 800oC and 900oC could be affected mainly by elastic damage in fatigue processing. Fatigue cracks were initiated in the surface oxide layer of the specimen. The plane of fracture surface was tilted toward <001> direction. The fatigue fracture mechanism was quasi-cleavage fracture at 800oC and 900oC. In all broken specimens, the γˊ phase morphology maintained cuboidal shape.

Citations

Citations to this article as recorded by

Mechanical Loading Effect on Stress States and Failure Behavior in Thermal Barrier Coatings
Da Qiao, Wengao Yan, Wu Zeng, Jixin Man, Beirao Xue, Xiangde Bian
Crystals.2023; 14(1): 2. CrossRef
A method for predicting the delamination life of thermal barrier coatings under thermal gradient mechanical fatigue condition considering degradation characteristics
Damhyun Kim, Kibum Park, Keekeun Kim, Chang-Sung Seok, Jongmin Lee, Kyomin Kim
International Journal of Fatigue.2021; 151: 106402. CrossRef
Low-cycle fatigue behavior of K416B Ni-based superalloy at 650 °C
Jun Xie, De-long Shu, Gui-chen Hou, Jin-jiang Yu, Yi-zhou Zhou, Xiao-feng Sun
Journal of Central South University.2021; 28(9): 2628. CrossRef

57 View
1 Download
Crossref

Prediction of Smart Greenhouse Temperature-Humidity Based on Multi-Dimensional LSTMs: Young Eun Song, Aekyung Moon, Su-Yong An, Hoeryong Jung; J. Korean Soc. Precis. Eng. 2019;36(3):239-246.
Published online March 1, 2019; DOI: https://doi.org/10.7736/KSPE.2019.36.3.239

Abstract
PDF

The objective of this study is to investigate a novel temperature and humidity prediction algorithm for smart greenhouse based on the machine learning method. The smart greenhouse is known to increase farm production by automatically controlling temperature and humidity and other factors. However, maintaining constant inside temperature and humidity in the conventional smart greenhouse system is still a problem because of the multiple time delay elements. To solve the problems, prediction control scheme is required. But, since the system is highly nonlinear with the lack of sensory data, predicting accurate temperature and humidity is very challenging. In this paper, the multi-dimensional Long Short-Term Memory networks (LSTMs) is being applied to deal with the unstructured greenhouse environmental data. The designed LSTMs learning model is trained with the 27 dimensional data which comprises of all the greenhouse control parameter and environmental sensory data. The prediction performance was evaluated using the short, mid and long term experiments. Also, the comparison with the conventional recurrent neural networks (RNNs) based prediction algorithm was done using the experimental results and later on discussions.

Citations

Citations to this article as recorded by

Data-Driven Optimization Method for Recurrent Neural Network Algorithm: Greenhouse Internal Temperature Prediction Model
Kwang Cheol Oh, Sunyong Park, Seok Jun Kim, La Hoon Cho, Chung Geon Lee, Dae Hyun Kim
Agronomy.2024; 14(11): 2545. CrossRef
Development and Verification of Smart Greenhouse Internal Temperature Prediction Model Using Machine Learning Algorithm
Kwang Cheol Oh, Seok Jun Kim, Sun Yong Park, Chung Geon Lee, La Hoon Cho, Young Kwang Jeon, Dae Hyun Kim
Journal of Bio-Environment Control.2022; 31(3): 152. CrossRef
Deep-Learning Temporal Predictor via Bidirectional Self-Attentive Encoder–Decoder Framework for IOT-Based Environmental Sensing in Intelligent Greenhouse
Xue-Bo Jin, Wei-Zhen Zheng, Jian-Lei Kong, Xiao-Yi Wang, Min Zuo, Qing-Chuan Zhang, Seng Lin
Agriculture.2021; 11(8): 802. CrossRef

54 View
0 Download
Crossref

Prediction of System Behavior by Sinusoidal Extrapolation Prediction Filter: Son Mook Oh, Jung Han Kim; J. Korean Soc. Precis. Eng. 2018;35(11):1063-1070.
Published online November 1, 2018; DOI: https://doi.org/10.7736/KSPE.2018.35.11.1063

Abstract
PDF

Predicting the response of a system, even several steps ahead, offers tremendous advantage to improve the system performance, to acquire an ideal model of a system and disturbances. The best way of predicting a response signal from a system is to use the sinusoidal extrapolation based on its frequency characteristics. Sinusoidal extrapolation is a statistical method for predicting future data through frequency analysis of past data. Practically speaking, the prediction from a frequency analysis in a control system is appropriate, because the output of a system can be modeled by several dominant frequencies from input and system models. In this study, we developed a novel and reliable prediction filter, using multi frequency sinusoidal extrapolation and a prediction error compensation algorithm. In this paper, we also suggest the design guidelines, regularity, and overall process of obtaining optimal predictions from an efficient and practical view, for the widely used industrial equipment. Results show that the performance of the proposed prediction filter is considered reliable and effective for improving the performance of a system, such as a motion controller.