https://jst.vn/index.php/ssad en-US Sun, 15 Jun 2025 00:00:00 +0000 OJS 3.2.1.1 http://blogs.law.harvard.edu/tech/rss 60 https://jst.vn/index.php/ssad/article/view/417 The study presents a simple way to optimally design a Proportional-Differential (PD) controller and apply it to the vibration control of a quarter car model's active suspension system. First, the optimization objectives are determined, including minimizing the vehicle body acceleration and the suspension deflection. The tyre deflection and road holding constraints are also considered. Next, the variables, including the components in the gain vector of the PD controller, are optimized using the Balancing Composite Motion Optimization (BCMO) algorithm. Different controller configurations, according to the two above optimization objectives, are simulated to verify the performance of the controllers for the nominal system and for the system when its mass and stiffness are varied. An H∞ controller in a reputable published study is also included for comparison. The simulation results show the proposed PD controllers' high control efficiency and robustness, especially the PD controller, which is based on minimizing the vehicle body's acceleration. Hai-Le Bui, Thi-Thoa Mac, Van-Quyen Nguyen, Ngoc-An Tran, Tung-Anh Le, Quy-Cao Tran, Sy-Tai Nguyen Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/417 https://jst.vn/index.php/ssad/article/view/418 Automating friction clutch engagement is essential to improving vehicle dynamic and driving comfort. The study proposes a Model Predictive Control (MPC) strategy for automated clutch engagement in truck launching, evaluated through simulation. The powertrain models are developed for designing and evaluating the controller. The MPC algorithm calculates the required friction torque to be transmitted through the clutch by minimizing the deviation between actual and desired parameters. Key performance metrics, including longidutianal jerk, specific friction work, and the dynamic load factor, are used to assess the effectiveness of the proposed control strategy. To further evaluate the controller’s impact on ride comfort, longitudinal jerk is analyzed through a co-simulation approach using specialized software. Simulation results for a first-gear launch under moderate intensity conditions show that the specific friction work is 18.4 J/cm², the dynamic load factor is 1.8, and the longitudinal jerk is 16.84 m/s³. These results confirm that the proposed MPC-based clutch control strategy ensures smooth engagement, enhances driving comfort, and meets performance requirements. Van Nghia Le, Quoc Trieu Nguyen, Hoang Phuc Dam, Trong Dat Tran, Ngoc Khanh Duong Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/418 https://jst.vn/index.php/ssad/article/view/419 The study tries to classify the axisymmetric boattail models with minimum drag using numerical simulation and neural networks. Numerical simulation was conducted for the boattail model in a range of angles from 0 to 22°and length from 0.5 to 1.5 diameter of the model. The Mach number was changed from 0.1 to 3.0. The results revealed that, the angle with minimum drag is around 14° at subsonic but it dramatically shifts to 7-9° at supersonic conditions. The maximum error of the neural network in predicting aerodynamic drag is less than 2%. At subsonic flow, the angle with minimum drag is around 14° and boattail length was 1.5 times the model diameter. At supersonic conditions, the angle and length are around 7° and 1.5 diameter of the model, respectively. Increasing boattail length results in reducing drag. This study provides a good reference for further design of flying objects and proposes control method for drag reduction. The Hung Tran, Cong Truong Dao, Dinh Anh Le, Trang Minh Nguyen, Cong Truong Dinh Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/419 https://jst.vn/index.php/ssad/article/view/420 In the study, the authors developed a portable, non-invasive smart device for real-time monitoring of electric motors' working conditions based on IoT technology and artificial intelligence. The device collects vibration data of an electric motor, predicting anomalies using deep learning algorithms. Additionally, an application was built to track the real-time working conditions of the electric motors. Whenever an anomaly is detected, an alert message is immediately sent to the user via their smartphone. For anomaly prediction, two types of vibration data were utilized in the deep learning algorithms: one in the time domain and the other in the frequency domain, obtained through a discrete Fourier transform. Various feature extraction models in deep learning algorithms were employed to assess the accuracy of each model in predicting electric motor anomalies. Experiments were conducted on a grinding machine operating under various grinding conditions to evaluate the accuracy of the developed device in predicting anomalies. The results indicate that predicting the working condition of an electric motor using time-domain vibration data is more accurate than using frequency-domain data. It was found that the Serenest26d_32x4d and Reset 34 feature extraction models achieved better training results with time-domain vibration data compared to other models. The Reset 34 feature extraction model achieves the highest accuracy, with an F1-score of 1, when predicting the working condition of the grinding machine. The running time for all prediction models is under 0.02 seconds, demonstrating the capability for real-time monitoring of the electric motor's working condition using the developed device. Duc Phuc Truong, Hoang Vu Nguyen, Tran Bach Dang Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/420 https://jst.vn/index.php/ssad/article/view/421 Many engineering applications require smooth orientation planning, i.e., interpolating the orientation of a rigid body so that its motion is smooth through intermediate poses. This smooth motion ensures for instance the continuity of the control torques. There are several ways to represent the orientation of a rigid body, so there are also different ways to plan motion for orientation. Each way has its advantages and disadvantages. In general, the problem of motion planning for the orientation has been less studied due to its complexity compared to motion planning for the endpoint. This paper presents the motion planning for the orientation using Euler parameters when the initial and final directions, and a set of intermediate directions are known. First, the Euler parameters are interpolated using cubic splines, and then they are normalized. Numerical simulations are carried out to validate the effectiveness of the proposed method. The proposed algorithms presented here preserve the fundamental properties of the interpolated rotation. The algorithms presented in this paper provide interpolation tools for rotation that are accurate, easy to implement. Quang Hoang Nguyen, Minh Hai Duong, Van Phong Dinh Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/421 https://jst.vn/index.php/ssad/article/view/422 The paper investigates the dynamic behavior of a 220 kV mixed overhead-underground transmission line compensated with a shunt reactor, with a focus on the current zero-missing phenomenon and switching overvoltages. The zero-missing phenomenon, a condition in which the current through the circuit breaker fails to reach zero, can affect breaker operation and system stability, particularly in shunt-compensated systems. Key factors influencing this issue include cable length, cable configuration, reactive power compensation, and switching strategies. Using EMTP/ATP simulations based on a real grid model, the study evaluates various mitigation strategies for the zero-missing phenomenon, such as pre-insertion resistors, compensation ratios, and connections to high-power loads. Transient overvoltages on cable cores and sheaths during switching operations are also analyzed. Thanh Chung Pham, Van Top Tran Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/422 https://jst.vn/index.php/ssad/article/view/423 Accurate solar power forecasting is crucial for optimizing grid operations and balancing energy supply and demand. Due to the high variability of solar radiation, advanced machine learning methods are needed to enhance forecasting accuracy. This study compares three models: Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and a single-hidden-layer Bidirectional Gated Recurrent Unit (BiGRU). XGBoost and LightGBM are decision tree-based boosting models known for fast training and high accuracy, while BiGRU is a recurrent neural network designed for time-series data but prone to overfitting. Experimental results show that XGBoost and LightGBM train significantly faster and achieve lower errors (Normalized Mean Absolute Percentage Error-NMAPE is lower than 5%), demonstrating superior generalization. In contrast, BiGRU exhibits overfitting with NMAPE equal to 23.986% and Root Mean Squared Error (RMSE) equal to 18,763.12 kW on June 30, 2021. Notably, on December 31, 2021, XGBoost and LightGBM closely followed actual power generation trends, whereas BiGRU struggled to capture variations, further indicating its generalization issues. The findings highlight XGBoost and LightGBM as more suitable models for solar power forecasting, providing valuable insights for researchers and engineers in power grid management. Manh-Hai Pham, Tuan Anh Nguyen, Minh Phap Vu, Van Duy Pham, Thanh Doanh Le, Thi Anh Tho Vu, Dang Toan Nguyen Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/423 https://jst.vn/index.php/ssad/article/view/424 The article presents a practical approach for implementing traditional Model Predictive Control (MPC) on a rotary inverted pendulum, a highly nonlinear and inherently unstable system. The study begins with the development of a mathematical model of the pendulum, followed by the application of a predictive controller to this model. The proposed algorithm is subsequently validated on an experimental platform, the Quanser QUBE-Servo2. The paper emphasizes the advantages of MPC, particularly its ability to incorporate system constraints and effectively manage nonlinear dynamics, thus making it a widely adopted strategy in industrial applications. Nevertheless, it also addresses the inherent challenges of MPC implementation, notably the construction of accurate system models and the computational burden associated with solving complex optimization problems. The control objective is to maintain the pendulum in its upright equilibrium position. The study evaluates the effectiveness of MPC with and without uncertainty compensation by analyzing key performance metrics, including response time, settling time, overshoot, and steady-state error, through both simulations and experiments. The results illustrate the comparative benefits and limitations of the uncertainty-compensated MPC algorithm relative to the traditional MPC controller. Hoang Dieu Dang, Thu Ha Nguyen, Thi Lan Anh Dinh, Duc Quang Nguyen Copyright (c) 2025 JST: Smart Systems and Devices https://creativecommons.org/licenses/by-nc/4.0 https://jst.vn/index.php/ssad/article/view/424