Search Results (1,777)

Early SOC balancing techniques primarily centered on simple hardware circuit designs. Passive balancing circuits utilize resistors to consume energy, aiming to balance the SOC among batteries; however, this approach leads to considerable energy wastage. As research progresses, active balancing circuits have garnered widespread attention. Successively, active balancing circuits utilizing capacitors, inductors, and transformers have been proposed, enhancing balancing efficiency to some extent. Nevertheless, challenges persist, including energy wastage during transfers between non-adjacent batteries and the complexity of circuit designs. In recent years, SOC balancing methods based on software algorithms have gained popularity. For instance, intelligent control algorithms are being integrated into battery management systems to optimize control strategies for SOC balancing. However, these methods may encounter issues such as high algorithmic complexity and stringent hardware requirements in practical applications. This paper proposes a fast state-of-charge (SOC) balance control strategy that incorporates a weighting factor within a modular battery energy storage system architecture. The modular distributed battery system consists of battery power modules (BPMs) connected in series, with each BPM comprising a battery cell and a bidirectional buck–boost DC-DC converter. By controlling the output voltage of each BPM, SOC balance can be achieved while ensuring stable regulation of the DC bus voltage without the need for external equalization circuits. Building on these BPMs, a sliding mode control strategy with adaptive acceleration coefficient weighting factors is designed to increase the output voltage difference of each BPM, thereby reducing the balancing time. Simulation and experimental results demonstrate that the proposed control strategy effectively increases the output voltage difference among the BPMs, facilitating SOC balance in a short time. Full article

This study leverages IoT technology to develop a real-time monitoring system for large motorcycles. We collaborated with professional mechanics to define the required data types and system architecture, ensuring practicality and efficiency. The system integrates the NB-IoT for efficient remote data transmission and uses MQTT for optimized messaging. It also includes advanced database management and intuitive data visualization for enhancing the user experience. For hardware installation, the system follows strict guidelines to avoid damaging the motorcycle’s original structure, comply with Taiwan’s legal standards, and prevent unauthorized modifications. The implementation of this real-time monitoring system is anticipated to significantly reduce safety risks associated with mechanical failures as it continuously monitors inappropriate driving behaviors and detects mechanical abnormalities in real time. The study indicates that the integration of advanced technologies, such as the NB-IoT and multi-sensor systems, can lead to improved driving safety and operational efficiency. Furthermore, the research suggests that the system’s ability to provide instant notifications and alerts through the platforms’ instant messaging can enhance user responsiveness to potential hazards, thereby contributing to a safer riding experience. Full article

Anomalous Sound Detection (ASD) systems are pivotal in the Industrial Internet of Things (IIoT). Through the early detection of machines’ anomalies, these systems facilitate proactive maintenance, thereby mitigating potential losses. Although prior studies have improved system accuracy using various advanced machine learning technologies, they frequently neglect the associated substantial computing and storage demands, which are crucial in resource-constrained IIoT environments. In this paper, we propose an ASD system that is efficiently optimized for both software and hardware considerations regarding edge intelligence. For the software aspect, we identify signal variation as a critical issue for ASD. Hence, we introduce a suite of lightweight yet robust processing techniques that enhance accuracy while minimizing resource consumption. As for the hardware aspect, we find that memory constraints may be a significant challenge for deploying ASD systems on microcontrollers (MCUs). Therefore, we propose a memory-aware pruning algorithm specialized for ASD to fit into MCUs’ constraints. Finally, we evaluate our method on the DCASE dataset, and the results show that our system achieves favorable outcomes in both accuracy and resource efficiency, marking our contribution to ASD system practice. Full article

For electrically driven commercial vehicles equipped with three-speed automatic mechanical transmission (AMT), the transmission control unit (TCU) without vehicle mass and road gradient estimation function will lead to frequent shifting and insufficient power during vehicle full-load or grade climbing. Therefore, it is necessary to estimate the mass and road gradient for the electrically driven commercial vehicles equipped with the three-speed AMT, and to adjust the shift rule according to the estimation results. Given the above problems, this paper focuses on the control and development of the electrically driven three-speed AMT and takes the shift controller with the vehicle mass and road gradient estimation as the research goal. The mathematical model and simulation model of vehicle dynamics are established to verify the shift function of TCU. The least squares method and calibration techniques are applied to estimate the vehicle mass and road gradient. According to the estimation results, the existing shift strategy is optimized, and the software-in-the-loop simulation of the transmission controller is carried out to verify the function of the control algorithm software. The hardware-in-the-loop test model is established to verify the shift strategy’s optimization effect, which shortens the controller’s forward development cycle. According to the estimation results of mass and gradient, the error result of the proposed method is controlled within 4.5% for mass and 8.6% for gradient. The experiment verifies that the optimized shift strategy can effectively improve the dynamic performance of the vehicle. The HIL experimental results show that the vehicle can maintain low gear while climbing the hill, and the vehicle speed does not decrease significantly. Full article

Deep learning techniques have been widely investigated as an effective method for signal measurement in recent years. However, most existing deep learning-based methods still face difficulty in deploying on embedded platforms and perform poorly in real-time applications. To address this, this paper develops two accelerators, as the core of the signal measurement system, for intelligent signal processing. Firstly, by introducing the idea of automated framework, we propose a simplest deep neural network (DNN)-based hardware structure, which automatically maps algorithms to hardware modules, supports configurable parameters, and has the advantage of low latency, with an average inference time of only 3.5 μs. Subsequently, another accelerator is designed with the efficient hardware structure of the long short-term memory (LSTM) + DNN model, demonstrating outstanding performance with a classification accuracy of 98.82%, mean absolute error (MAE) of 0.27°, and root mean square errors (RMSE) of 0.392° after model compression. Moreover, parallel optimization strategies are exploited to further reduce latency and support simultaneous frequency and direction measurement tasks. Finally, we test the actual collected signal data on the XCVU13P field programmable gate array (FPGA). The results show that the time of inference saves 28–31% for the DNN model and 71–73% for the LSTM + DNN model compared to running on graphic processing unit (GPU). In addition, the parallel strategies further decrease the delay by 23.9% and 37.5% when processing continuous data. The FPGA-based and deep learning-assisted hardware accelerators significantly improve real-time performance and provide a promising solution for signal measurement. Full article

This paper introduces a model that predicts four common smartphone-holding postures, aiming to enhance user interface adaptability. It is unique in being completely independent of platform and hardware, utilizing the inertial measurement unit (IMU) for real-time posture detection based on sensor data collected around tap gestures. The model identifies whether the user is holding and operating the smartphone with one hand or using both hands in different configurations. For model training and validation, sensor time series data undergo extensive feature extraction, including statistical, frequency, magnitude, and wavelet analyses. These features are incorporated into 74 distinct sets, tested across various machine learning frameworks—k-nearest neighbors (KNN), support vector machine (SVM), and random forest (RF)—and evaluated for their effectiveness using metrics such as cross-validation scores, test accuracy, Kappa statistics, confusion matrices, and ROC curves. The optimized model demonstrates a high degree of accuracy, successfully predicting the holding hand with a 95.7% success rate. This approach highlights the potential of leveraging sensor data to improve mobile user experiences by adapting interfaces to natural user interactions. Full article

Floating photovoltaic systems are rapidly gaining popularity due to their advantages in conserving land resources and their high energy conversion efficiency, making them a promising option for photovoltaic power generation. However, these systems face challenges in offshore environments characterized by high salinity, humidity, and variable irradiation, which necessitate effective maximum power point tracking (MPPT) technologies to optimize performance. Currently, there is limited research in this area, and few reviews analyze it comprehensively. This paper provides a thorough review of MPPT techniques applicable to floating photovoltaic systems, evaluating the suitability of various methods under marine conditions. Traditional algorithms require modifications to address the drift phenomena under uniform irradiation, while different GMPPT techniques exhibit distinct strengths and limitations in partial shading conditions (PSCs). Hardware reconfiguration technologies are not suitable for offshore use, and while sampled data-based techniques are simple, they carry the risk of erroneous judgments. Intelligent technologies face implementation challenges. Hybrid algorithms, which can combine the advantages of multiple approaches, emerge as a more viable solution. This review aims to serve as a valuable reference for engineers researching MPPT technologies for floating photovoltaic systems. Full article

Human activity recognition (HAR) technology is related to human safety and convenience, making it crucial for it to infer human activity accurately. Furthermore, it must consume low power at all times when detecting human activity and be inexpensive to operate. For this purpose, a low-power and lightweight design of the HAR system is essential. In this paper, we propose a low-power and lightweight HAR system using point-cloud data collected by radar. The proposed HAR system uses a pillar feature encoder that converts 3D point-cloud data into a 2D image and a classification network based on depth-wise separable convolution for lightweighting. The proposed classification network achieved an accuracy of 95.54%, with 25.77 M multiply–accumulate operations and 22.28 K network parameters implemented in a 32 bit floating-point format. This network achieved 94.79% accuracy with 4 bit quantization, which reduced memory usage to 12.5% compared to existing 32 bit format networks. In addition, we implemented a lightweight HAR system optimized for low-power design on a heterogeneous computing platform, a Zynq UltraScale+ ZCU104 device, through hardware–software implementation. It took 2.43 ms of execution time to perform one frame of HAR on the device and the system consumed 3.479 W of power when running. Full article

With the rapid growth of connected vehicles and the vulnerability of embedded systems against cyber attacks in an era where quantum computers are becoming a reality, post-quantum cryptography (PQC) is a crucial solution. Yet, by nature, automotive sensors are limited in power, processing capability, memory in implementing secure measures. This study presents a pioneering approach to securing automotive systems against post-quantum threats by integrating the Ascon cipher suite—a lightweight cryptographic protocol—into embedded automotive environments. By combining Ascon with the Controller Area Network (CAN) protocol on an Artix-7 Field Programmable Gate Array (FPGA), we achieve low power consumption while ensuring high performance in post-quantum-resistant cryptographic tasks. The Ascon module is designed to optimize computational efficiency through bitwise Boolean operations and logic gates, avoiding resource-intensive look-up tables and achieving superior processing speed. Our hardware design delivers significant speed improvements of 100 times over software implementations and operates effectively within a 100 MHz clock while demonstrating low resource usage. Furthermore, a custom digital signal processing block supports CAN protocol integration, handling message alignment and synchronization to maintain signal integrity under automotive environmental noise. Our work provides a power-efficient, robust cryptographic solution that prepares automotive systems for quantum-era security challenges, emphasizing lightweight cryptography’s readiness for real-world deployment in automotive industries. Full article

Recurrent neural networks (RNNs) have produced significant results in many fields, such as natural language processing and speech recognition. Owing to their computational complexity and sequence dependencies, RNNs need to be deployed on customized hardware accelerators to satisfy performance and energy-efficiency constraints. However, designing hardware accelerators for RNNs is challenged by the vast design space and the reliance on ineffective optimization. An efficient automated design space exploration (DSE) strategy that can balance conflicting objectives is wanted. To address the low efficiency and insufficient universality of the resource allocation process employed for hardware accelerators, we propose an automated two-stage design space exploration (DSE) strategy for customized RNN accelerators. The strategy combines a genetic algorithm (GA) and a reinforcement learning (RL) algorithm, and it utilizes symmetrical exploration and exploitation to find the optimal solutions. In the first stage, the area of the hardware accelerator is taken as the optimization objective, and the GA is used for partial exploration purposes to narrow the design space while maintaining diversity. Then, the latency and power of the hardware accelerator are taken as the optimization objectives, and the RL algorithm is used in the second stage to find the corresponding Pareto solutions. To verify the effectiveness of the developed strategy, it is compared with other algorithms. We use three different network models as benchmarks: a vanilla RNN, LSTM, and a GRU. The results demonstrate that the strategy proposed in this paper can provide better solutions and can achieve latency, power, and area reductions of 9.35%, 5.34%, and 11.95%, respectively. The HV of GRMD is reduced by averages of 6.33%, 6.32%, and 0.67%, and the runtime is reduced by averages of 18.11%, 14.94%, and 10.28%, respectively. Additionally, given different weights, it can make reasonable trade-offs between multiple objectives. Full article

The Winograd algorithm reduces the computational complexity of convolutional neural networks (CNNs) by minimizing the number of multiplications required for convolutions, making it particularly suitable for resource-constrained edge devices. Concurrently, most edge hardware accelerators utilize 8-bit integer arithmetic to enhance energy efficiency and reduce inference latency, requiring the quantization of CNNs before deployment. Combining Winograd-based convolution with quantization offers the potential for both performance acceleration and reduced energy consumption. However, prior research has identified significant challenges in this combination, particularly due to numerical instability and substantial accuracy degradation caused by the transformations required in the Winograd domain, making the two techniques incompatible on edge hardware. In this work, we describe our latest training scheme, which addresses these challenges, enabling the successful integration of Winograd-accelerated convolution with low-precision quantization while maintaining high task-related accuracy. Our approach mitigates the numerical instability typically introduced during the transformation, ensuring compatibility between the two techniques. Additionally, we extend our work by presenting a custom-optimized CUDA implementation of quantized Winograd convolution for NVIDIA edge GPUs. This implementation takes full advantage of the proposed training scheme, achieving both high computational efficiency and accuracy, making it a compelling solution for edge-based AI applications. Our training approach enables significant MAC reduction with minimal impact on prediction quality. Furthermore, our hardware results demonstrate up to a 3.4× latency reduction for specific layers, and a 1.44× overall reduction in latency for the entire DeepLabV3 model, compared to the standard implementation. Full article

Leak detection in oil and gas pipeline networks is a climacteric and frequent issue in the oil and gas field. Many establishments have long depended on stationary hardware or traditional assessments to monitor and detect abnormalities. Rapid technological progress; innovation in engineering; and advanced technologies providing cost-effective, rapidly executed, and easy to implement solutions lead to building an efficient oil pipeline leak detection and real-time monitoring system. In this area, wireless sensor networks (WSNs) are increasingly required to enhance the reliability of checkups and improve the accuracy of real-time oil pipeline monitoring systems with limited hardware resources. The real-time transient model (RTTM) is a leak detection method integrated with LoRaWAN technology, which is proposed in this study to implement a wireless oil pipeline network for long distances. This study will focus on enhancing the LoRa network parameters, e.g., node power consumption, average packet loss, and delay, by applying several machine learning techniques in order to optimize the durability of individual nodes’ lifetimes and enhance total system performance. The proposed system is implemented in an OMNeT++ network simulator with several frameworks, such as Flora and Inet, to cover the LoRa network, which is used as the system’s network infrastructure. In order to implement artificial intelligence over the FLoRa network, the LoRa network was integrated with several programming tools and libraries, such as Python script and the TensorFlow libraries. Several machine learning algorithms have been applied, such as the random forest (RF) algorithm and the deep extreme learning machine (DELM) technique, to develop the proposed model and improve the LoRa network’s performance. They improved the LoRa network’s output performance, e.g., its power consumption, packet loss, and packet delay, with different enhancement ratios. Finally, a hybrid deep extreme learning machine model was built and selected as the proposed model due to its ability to improve the LoRa network’s performance, with perfect prediction accuracy, a mean square error of 0.75, and an exceptional enhancement ratio of 39% for LoRa node power consumption. Full article

Due to the growth of renewable energies, which requires cost reduction and efficiency in terms of structural health assessment, failure prevention, effective maintenance scheduling, and equipment lifespan optimization, in this paper, we propose an Ultra Wideband (UWB)-based accelerometer Sensor Node for low-power applications in offshore platforms. The proposed Sensor Node integrates a high-resolution accelerometer together with an Impulse Radio Ultra-Wideband (IR-UWB) transceiver. This approach enables effective remote monitoring of structural vibrations. This provides an easy-to-install, scalable, and flexible wireless solution without sacrificing robustness and low power consumption in marine environments. Additionally, due to the diverse and highly demanding applications of condition monitoring systems, we propose two modes of operation for the Sensor Node. It can be remotely configured to either transmit raw data for further analysis or process data at the edge. A hardware (HW) description of the proposed Sensor Node is provided. Moreover, we describe the power management strategies implemented in our system at the firmware (FW) level. We show detailed power consumption measurements, including power profiles and the battery-powered autonomy of the proposed Sensor Node. We compare data from a wired acquisition system and the proposed wireless Sensor Node in a laboratory environment.The wired sensor integrated into this acquisition system, fully characterized and tested, is our golden reference. Thus, we validate our proposal. Furthermore, this research work is within the scope of the SUREWAVE Project and is conducted in collaboration with the MARIN Institute, where wave basin tests are carried out to evaluate the behavior of a Floating Photovoltaic (FPV) system. These tests have provided a valuable opportunity to assess the effectiveness of the proposed Sensor Node for offshore platforms and to compare its performance with a wired system. Full article

Field positioning (FP) is a key technique in the digitalization of agriculture. By integrating sensors and mapping techniques, FP can convey critical information such as soil quality, plant distribution, and topography. Utilizing vehicles for field applications provides precise control and scientific management for agricultural production. Compared to conventional methods, which often struggle with the complexities of field conditions and suffer from insufficient accuracy, this study employs a novel approach using self-developed multi-sensor array hardware as a portable field topographic surveying device. This innovative setup effectively navigates challenging field conditions to collect raw data. Data fusion is carried out using the Unscented Kalman Filter (UKF) algorithm. Building on this, this study combines the good point set and Opposition-based Differential Evolution for a joint improvement of the Slime Mould Algorithm. This is linked with the UKF algorithm to establish loss value feedback, realizing the adaptive parameter adjustment of the UKF algorithm. This reduces the workload of parameter setting and enhances the precision of data fusion. The improved algorithm optimizes parameters with an efficiency increase of 40.43%. Combining professional, mapping-grade total stations for accuracy comparison, the final test results show an absolute error of less than 0.3857 m, achieving decimeter-level precision in field positioning. This provides a new application technology for better implementation of agricultural digitalization. Full article

As user requirements become increasingly complex, the demand for product personalization is growing, but traditional hardware-centric production relies on fixed procedures that lack the flexibility to support diverse requirements. Although bespoke manufacturing has been introduced, it provides users with only a few standardized options, limiting its ability to meet a wide range of needs. To address this issue, a new manufacturing concept called the software-defined factory has emerged. It is an autonomous manufacturing system that provides reconfigurable manufacturing services to produce tailored products. Reinforcement learning has been suggested for flexible scheduling to satisfy user requirements. However, fixed rule-based methods struggle to accommodate conflicting needs. This study proposes a novel federated digital twin scheduling that combines large language models and deep reinforcement learning algorithms to meet diverse user requirements in the software-defined factory. The large language model-based literacy module analyzes requirements in natural language and assigns weights to digital twin attributes to achieve highly relevant KPIs, which are used to guide scheduling decisions. The deep reinforcement learning-based scheduling module optimizes scheduling by selecting the job and machine with the maximum reward. Different types of user requirements, such as reducing manufacturing costs and improving productivity, are input and evaluated by comparing the flow-shop scheduling with job-shop scheduling based on reinforcement learning. Experimental results indicate that in requirement case 1 (the manufacturing cost), the proposed method outperforms flow-shop scheduling by up to 14.9% and job-shop scheduling by 5.6%. For requirement case 2 (productivity), it exceeds the flow-shop method by up to 13.4% and the job-shop baseline by 7.2%. The results confirm that the literacy DRL scheduling proposed in this paper can handle the individual characteristics of requirements. Full article