The performances of defect inspection have been severely hindered by insufficient images in industries, which can be alleviated generating more samples as data augmentation. We propose the first image generation method challenging few-shot cases. Given just a handful and relatively defect-free ones, our goal is to augment dataset with new images. Our consists two training stages. First, we train data-efficient StyleGAN2 on backbone. Second, attach defect-aware residual blocks backbone, learn...

5G wireless networks are expected to support new applications demanding ultra high reliability and low latency, such as self-driving cars, industrial control, real-time gaming. Such services known ultra-reliable low-latency communications (URLLC) targeting at 99.999% transmission correctness within 1 ms delay bound. Current Long Term Evolution (LTE) system is far from supporting services, especially for uplink. Even with around 10 times shortened interval, due the waste of time in requesting...

This manuscript investigates the equity-oriented train regulation optimization problem by formulating it as a Markov Decision Process and addressing through Lagrange relaxation methodology. To ensure equitable service meet safety requirements, we propose constrained Deep Reinforcement Learning approach. approach utilizes reshaped reward functions to enforce hard constraints enables dynamic adjustment of policies throughout learning phase. Leveraging an actor-critic framework, this handles...

Abstract To solve the problem of low magnetic field utilization traditional magnetorheological (MR) brakes, a cylindrical MR brake with inner and outer fluid flow channels was proposed in this paper. By adding non-magnetic device to rotating sleeve cylinder, flux lines meandered through six effective damping gaps axial channels, allowing achieve better braking performance without changing its external dimensions. The structure working principle were depicted, circuit analysis also carried...

Train operation control in urban railways is challenging due to its high dynamics, complex environment, and level of comfort safety. To address these challenges, this article, the authors propose a new deep reinforcement-based train (DRTO) method which includes: 1) A deterministic reinforcement learning algorithm, 2) dynamic incentive system, used ensure safe multitrain 3) an event-driven method, improve DRTO performance based on strategy. evaluate performance, we thoroughly compare proposed...

Traction transformer is vital equipment in high-speed railway. Insulation status decides its safety and reliability, the frequency-domain dielectric spectrum (FDS) test one of most effective methods reflecting changing insulation status. For field applications, following problems should be addressed: 1) how to obtain result paper from combined oil-paper system 2) discriminate defects between aging damp. In this article, first problem was transferred a nonlinear equation set, cuckoo search...

Vision transformers (ViTs) have recently received explosive popularity, but the huge computational cost is still a severe issue. Since computation complexity of ViT quadratic with respect to input sequence length, mainstream paradigm for reduction reduce number tokens. Existing designs include structured spatial compression that uses progressive shrinking pyramid computations large feature maps, and unstructured token pruning dynamically drops redundant However, limitation existing lies in...

This manuscript focuses on investigating the metro network expansion problem, which is formulated as a Markov Decision Process and addressed using sequential station selection methodology. To identify an effective strategy, we introduce multi-objective reinforcement learning framework, encompasses objectives such traffic demands, social equity, accessibility. The proposed method effectively eliminates need for domain-specific rules expert guidance by leveraging reward calculations to...

Transportation-related consumer electronics technology has advanced rapidly, particularly for automated train operation on high-speed railways. To maximize transport capacity and meet growing demands, this manuscript proposes a new elastic tracking control method, that compresses the interval while maintaining safety. The process is formulated as Monte Carlo Twin Delayed Deep Deterministic policy gradient (TD3) algorithm used to generate basic strategy. A three-stage principle requirements...

The next generation new radio (NR) network, or the fifth (5G) system, will be able to enhance significantly current Long Term Evolution (LTE), fourth (4G) network from many perspectives and initiations. One of such initiations is support Ultra Reliable Low Latency Communication (URLLC) services, especially transmission small but critical control related packets with periodic aperiodic traffic patterns very stringent latency (e.g, less than 1 ms)and reliability (up 99.999% 99.9999%)....

Traction transformers are one of the most important equipment in traction power supply system. Thermal aging transformer insulation system determines its reliable life. To achieve an estimation life operated under real load condition, effects loads on thermal oil-paper were investigated. First, design principles for accelerated experiment studied and proposed by considering magnitude (i.e., rate) frequency (i.e. spare time lasting time) load. Then, experimental was established laboratory to...

N-type GaN films bombarded with different highly charged 126Xeq+-ions（9≤q≤30） at room temperature was studied by atomic force microscopy. The experimental results show that when q exceeds the threshold value 18, remarkable swelling turns into obvious erosion in irradiated area. On other hand, surface disorder of strongly depends on charge state ions, incident angle and ion influence, damage behavior is unrelated to kinetic energy within scope parameters（180 keV≤Ek≤600 keV）.For q=18,...

Abstract For passengers traveling by the railway, a reasonable train timetable can improve passenger satisfaction efficiently, and service quality of railway transport. Existing related research focuses on reprogramming from perspective operating time income, but how to optimize rationality travel has not been greatly resolved. Therefore, this paper proposes multi-objective compilation method based sleep period checking. In system, quantitative calculation preference is defined, arrival...

This paper proposes a multi-objective reinforcement learning approach for the metro network expansion problem. The problem is modeled as Markov Decision Process, treating it sequential station selection procedure. Through reward calculations to fine-tune policy during process, proposed method eliminates need expert guidance and problem-specific heuristics. To fully utilize traffic demands address curse of dimensionality, , an actor-critic framework employed generate next evaluate current...

Virtual clothes try-on has emerged as a vital feature in online shopping, offering consumers critical tool to visualize how clothing fits. In our research, we introduce an innovative approach for virtual try-on, utilizing self-supervised Vision Transformer (ViT) coupled with diffusion model. Our method emphasizes detail enhancement by contrasting local image embeddings, generated ViT, their global counterparts. Techniques such conditional guidance and focus on key regions have been...

ML Framework for PPCPs fate in WWTPs.

Inadequate information capture and transmission of multi-energy coupled integrated energy systems can lead to low utilization poor economic social benefits. To address this problem, paper proposes a real-time optimal scheduling strategy for in parks based on digital twin. Firstly, an system model with coupling the park is constructed twin technology; secondly, particle swarm algorithm used solve problem objective minimizing total operation cost. Finally, test (PIES) built CloudPSS platform...