Search

The usage of lithium-ion batteries has significantly increased by various applications in recent years due to the advantages of long lifespan, high energy density, high power density, and eco-friendly environment benefits for sustainable usage. Although it has attracted much interest on its manufacturing process from practitioner in Industry 4.0 now, academia has relatively less concern on addressing the manufacturing process in using hybrid techniques. Thus, an advanced hybrid four-stage detection model is proposed into the original process to help identify optimal real-time production performance to hold sustainable manufacturing. This proposed hybrid model creates four key detection stages corresponded to four core challenges, including autoregressive integrated moving average (ARIMA), If-Then-Else programming rule, convolutional neural networks (CNNs), and artificial neural networks (ANNs) for achieving the purpose of economizing manpower and material resources to save manufacturing cost. (1) In cyclic process stage, ARIMA makes a successful prediction of a 20-min discharging curve with a significant low 13.38% error rate. (2) If-Then-Else programming rule sets up min–max threshold range (MMTR) on both attributes of voltage and capacity to find out failure packs before final test 1 and benefit with saving manual testing time. (3) CNNs achieve 100% classification accuracy in a productive result and save average 3.5 h for each battery pack. (4) ANNs conclude an empirical result of 100% accuracy in predicting the battery pack pass or failure. Conclusively, the study makes a comparative research with various deep learning algorithms to evaluate its performance; the proposed hybrid detection model is never seen in the challengeable lines of predicting Li-ion battery pack, and thus, it has an innovative value and priority performance for benefiting the sustainable manufacturing on offering green and renewable energy. This study contributes the research rationality and practical significance. [ABSTRACT FROM AUTHOR]

Recent advancements in energy storage along with power electronic technology have made battery energy storage systems (ESS) a feasible alternative for current power applications. Battery packs with lithium-ion (Li-ion) pouch cells are the main source of ESS. However, it is a big trouble that improper voltage and impedance of laser welding significantly affect the whole battery module during battery pack manufacturing stages, causing the cell imbalance inside and eventually resulting in a thermal runaway of the battery pack and non-durable use. Importantly, the formation of nuggets welded can be classified as good (GD) or not-good (NG) based on the labels after peeling off the flyer of the Li-ion pouch cell. Interestingly, it is usually standard practice in this industry to include substantial numbers of redundant welds to gain confidence in the structural stability of the welded component. Thus, non-destructive and low-cost detection for identifying the nuggets is absolutely necessary. An effective methodology is motivated and proposed with three procedures for the identification of laser-welded nuggets. At first, the nuggets are detected and separated from a grayscale image. Image features are extracted to train the nugget images on the advanced detector model constructed to identify the GD and NG nuggets. Second, this research develops five models for achieving this purpose of detector; one is called the nugget model developed in the convolutional neural network (CNN) technique, and the others use the transfer learning of the most popular pre-trained models (i.e., InceptionV3, EfficientNet, ResNet, and MobileNet). From the comparative studies, it was found that the ResNet (residual network) model more effectively classifies the fault nuggets with a 100% accuracy rate than the other listed models. Finally, this research has significant application contributions for battery manufacturing industries to produce highly efficient welded nugget products by overcoming the cost-ineffective problems of manual inspection; thus, it further helps this industry simultaneously reduce productive inspection time and increase the manufacturing efficiency of ESS at a lower cost without human intervention than in the past. [ABSTRACT FROM AUTHOR]

Device-to-device (D2D) communications is designed to improve the overall network performance, including low latency, high data rates, and system capacity of the fifth-generation (5G) wireless networks. The system capacity can even be improved by reusing resources between D2D user equipments (DUEs) and cellular user equipments (CUEs) without causing harmful interference to the CUEs. A D2D resource allocation scheme is expected to have the characteristic that one CUE be allocated with a variable number of resource blocks (RBs), and the RBs be reused by more than one DUE. In this study, the Multi-Player Multi-Armed Bandit (MPMAB) reinforcement learning scheme is employed to model such a problem by establishing a preference matrix to facilitate greedy resource allocation. A fair resource allocation scheme is then proposed and shown to achieve fairness, prevent waste of resources, and alleviate starvation. Moreover, this scheme has better performance when there are not too many D2D pairs. [ABSTRACT FROM AUTHOR]

The International Telecommunication Union has required that the control plane (C-plane) latency in the fifth generation (5G) ultra-reliable low-latency communication (URLLC) application scenarios should not exceed 20 ms and encouraged technical innovation to further reduce it to less than 10 ms. However, the average C-plane latency in the fourth generation (4G) Long Term Evolution-Advanced (LTE-A) system is 80 ms. Such a high latency is because of the execution of the contention-based random access procedure (RAP). In this paper, we simplify the conventional contention-based RAP from 4 to 2 steps. Furthermore, utilization of demodulation reference signal for representing the UE ID and reservation of preambles for URLLC users significantly reduces the proposed 2-step RAP latency. From the perspectives of fixing the number of URLLC users and fixing the number of preambles reserved for URLLC users, simulation results show the percentage of successes for the 2-step RAP is 83.81% and 71.83% higher than that of the 4-step RAP, respectively. Consequently, the 10 ms latency requirement of the 5G URLLC is achieved. [ABSTRACT FROM AUTHOR]

The distinctive properties of partly sign-altered (PSA) coefficients in discrete cosine transform (DCT) domain are explored to achieve effective dual blind image watermarking. A host image is divided into non-overlapped blocks, each of which is then converted to a DCT representation separately. For each block, low-frequency DCT coefficients are selected for dual binary watermarking. The first binary bit is applied to the mean value of the PSA coefficients. Our formulation takes into account of human psychovisual characteristics. A stair-like quantization function is designed to not only regulate embedding strength but accommodate extra embedding for blocks with high contrast masking. In case second watermarking is requested, we tactically adjust the standard deviation of the PSA coefficients while keeping the PSA mean intact. The embedding strengths with respect to the involved parameters in the first and second watermarking are analytically examined. Experiment results indicate that the proposed scheme is capable of achieving excellent robustness at a peak signal-to-noise ratio (PSNR) around 38.7 dB. In particular, the exploitation of just noticeable distortion helps to enhance the robustness of the first watermark without causing noticeable quality degradation. The secondary watermark is embedded at the cost of approximately 1 dB PSNR. Yet the resultant performance is comparable with other compared methods in terms of bit error rates. [ABSTRACT FROM AUTHOR]

A windowed vector modulation (WVM) scheme incorporable with distortion compensation is proposed to enhance the imperceptibility and robustness of blind audio watermarking in the norm space of discrete wavelet transform (DWT) domain. The merits of WVM are demonstrated through performance comparison among three DWT-based watermarking methods that adaptively modulate the vectors collected from the approximation subband after 2-level DWT decomposition of audio signals. The incorporation of window weighting into modulation formulation can smooth the transitions across adjacent vectors, while the employment of distortion compensation allows the WVM to be executable using larger quantization steps without introducing additional distortion. Experiment results confirm that the perceptual quality is noticeably improved through the windowing process, and the bit error rates of retrieved watermarks are manifestly reduced after integrating WVM with distortion compensation. [ABSTRACT FROM AUTHOR]

Mining web facial images on the internet has become as a profitable and important paradigm towards auto face annotation technique. The unsupervised label refinement (ULR) is an effective method that can fix weakly labeled facial images data which are collected from the internet and included some images with wrong label. In order to improve the correction accuracy of ULR, particle swarm optimization (PSO) and binary particle swarm optimization (BPSO) are used for solving binary constraint optimization task in this study. A novel method named safe binary particle swam optimization (SBPSO) is also proposed to improve BPSO which has the probability over range problem for using the ULR. In addition, SBPSO is also employed for an enhanced ULR (EULR) objective function which is created by modifying the original formula of ULR to improve the accuracy of labeled facial image. An experimental database is queried from IMDb website which collected the actors who were bored in 1950 to 1990. Some error flags are randomly added in the database for the correction tests by different methods. The results showed that the SBPSO Algorithm for the EULR in automatic face annotation have the better label correction rate and convergence effect. [ABSTRACT FROM AUTHOR]

In this study, advanced computing in fuzzy rule-based preprocessing classifier is used to categorize five distribution types that are identified as noise by noise detection from the service of next era cloud-empowered computing and techniques. The proposed method extracts useful local information from the corrupted image that is supported by filter processing and resulted in more image details to preserve. The fuzzy rule-based preprocessing classifier goes through a four-phase detection procedure to determine the condition of central pixels for the local image window by using the similarity between the neighboring pixels. A filters' system is set up with our proposed fuzzy rule-based preprocessing classifier with several effective filters to verify its performance. Simulation results are compared with other individual filters by objective numerical measurements and subjectively visual inspection to indicate that our proposed method performs significantly better in terms of noise suppression and detail preservation. [ABSTRACT FROM AUTHOR]

This paper presents a self-synchronizing blind audio watermarking method developed based on the distinctive features of multilevel discrete wavelet transform (MDWT) and discrete cosine transform (DCT). In the proposed design, the repetitive pattern hidden in the 11th approximation subband is used to control frame synchronization, and the wideband signal covering the first to ninth detail subbands is used to hide binary information. Tracing the zero-crossings across the extracted synchronous signal makes it possible to recalibrate frame positions and achieve synchronous watermarking on a rational dither modulation (RDM) framework. We also developed a windowed vector modulation (WVM) scheme suited for the time as well as DCT domain to enhance watermark imperceptivity. The proposed watermarking method currently achieves a payload capacity of 86 bits per second. The results of PEAQ testing confirm that the watermarked audio signal retains quality nearly indistinguishable from the original. Compared with two existing synchronized methods designed specifically to cope with playback speed modification, the proposed MDWT-DCT-RDM-WVM method demonstrates superior robustness to commonly encountered attacks. When encountering desynchronization attacks, such as cropping and resampling time-scale modification, the proposed technique is capable of retrieving the watermark bits with a high degree of accuracy. [ABSTRACT FROM AUTHOR]