Search Results (226)

Grid-based multi-agent path finding (MAPF) solvers scale to large teams, but their discrete schedules may not provide high-quality continuous finite-radius motions near the square-grid corner-passing threshold. We study endpoint-time-preserving geometric smoothing for disk agents at radius $0.35$. We establish an embedded-graph corner-passing threshold for synchronized finite-radius local passes and derive the square-grid radius $r_c=\sqrt{2}/4$. Finite-radius realizations are formulated as Lipschitz trajectories, and we prove that standard four-neighbor schedules without vertex conflicts or head-on edge swaps are pairwise continuously feasible up to this threshold. The smoother replaces windows by shortcuts only when speed, obstacle-clearance, pairwise continuous-collision detection, and length checks pass. Accepted shortcuts preserve endpoint times, schedule-level makespan, discrete arrival records, and discrete sum-of-costs while enforcing geometric length non-increase; the strict-decrease subset yields the reported geometric sum-of-costs reductions. Across six MovingAI map settings, LaCAM solves 575 benchmark instances; 570 smoothed trajectories pass finite-radius validation, with median geometric sum-of-costs reductions of 9.9% on the main slice and 11.2% on the five-map extension. A targeted 100-agent radius sweep further supports the threshold interpretation by showing a clean feasibility transition around the predicted corner-passing radius. The results support time-preserving smoothing as a validated geometric-quality layer for scalable grid planners. Full article

6082 aluminum alloy is widely applied in marine engineering, rail transportation and other industries owing to its excellent comprehensive performance. Welding heat source characteristics exert a decisive influence on the microstructure and mechanical properties of welded joints and become a major constraint for the application of medium-thick aluminum alloy welded structures. In this work, comparative tests of TIG and MIG welding were carried out on 16 mm thick 6082 aluminum alloy plates. Combining thermal simulation, metallographic observation and mechanical property tests, the temperature field distribution, microstructure, microhardness, tensile properties and bending properties of the two kinds of joints were systematically studied. The results show that TIG welding possesses high heat input, forming a broad temperature field with steep thermal gradients. Its weld microstructure is coarse and accompanied by severe coarsening of Mg₂Si precipitates, and the joint presents a highly fluctuating M-shaped microhardness distribution. The average tensile strength of TIG welded joints is 194 MPa, and all specimens fracture in the heat-affected zone. By contrast, MIG welding with low heat input produces a uniform temperature field, as well as a fine and homogeneous weld microstructure with dispersed precipitates. Its microhardness distribution is stable, and the average tensile strength reaches 256 MPa, 32% higher than that of TIG joints. Both welding methods deliver favorable bending performance. The difference in heat input and cooling behavior changes the grain evolution and precipitate characteristics and further dominates the final mechanical performance of joints. MIG welding is more suitable for multi-layer, multi-pass welding of 16 mm thick 6082 aluminum alloy. This work clarifies the correlation between heat input, microstructure and mechanical properties, and the optimized process can effectively improve the microstructural uniformity of the weld joint and enhance its mechanical properties. Full article

Small and medium enterprises routinely face a paradox in financial monitoring: their accounting documents exist, but the cost of converting heterogeneous PDFs into timely financial signals is prohibitive without dedicated analytical staff or specialized software. This paper presents a two-layer artifact, designed under Design Science Research, that bridges this gap using only public-web large language models (LLMs) and a parsimonious multi-criteria decision routine. Layer 1 implements a structured LLM-driven workflow that extracts account–value pairs from annual tax balance sheets without code, APIs, or fine-tuning. Layer 2 reconstructs auditable accounting aggregates and ranks yearly financial condition through TOPSIS with min–max normalization—a deliberate replacement for classical vector normalization, which fails when profitability indicators are negative, as routinely occurs in distress years. To avoid size effects and algebraic redundancy, the decision matrix uses only three criteria spanning liquidity, profitability, and solvency. The artifact is demonstrated in a four-year case study of an anonymized construction SME (2021–2024), with accountant-verified document-level match rates of 0.810, 0.998, 0.950, and 0.909. Equal weighting is the only weighting configuration used; a supplementary entropy-based dispersion diagnostic yields the same ordinal ranking—2024 > 2023 > 2021 > 2022—and 10,000 Monte Carlo replications, with uncertainty injected at the reconstructed-aggregate level, confirm that the extreme ranks are invariant across all runs. The contribution is methodological and practical: a transparent, low-infrastructure pipeline that brings first-pass financial screening within reach of SMEs operating under severe data and budget constraints. Full article

In this study, manual welding electrodes with varying niobium (Nb) contents (0, 0.05, and 0.1 wt%) were developed for 960 MPa grade low-alloy high-strength steel, and deposited metals were produced through multilayer multipass welding. Microstructural characterization and mechanical testing were performed using scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction (XRD), electron backscatter diffraction (EBSD), and a universal testing machine to investigate the influence of Nb content and elucidate the strengthening mechanisms. The results demonstrate that under identical welding conditions, multipass thermal cycles induced a primary microstructural transformation from martensite to tempered martensite in all deposited metals, which predominantly comprised tempered martensite with minor fractions of bainite and second-phase particles. Increasing Nb content led to significant grain refinement. The second-phase particles exhibited sizes of 0.158 μm, 0.176 μm, and 0.168 μm, respectively, with volume fractions of 5.69%, 5.82%, and 5.90%. Nb addition substantially enhanced hardness and strength while causing a noticeable reduction in low-temperature impact toughness, though the values remained within acceptable limits. The deposited metal containing 0.05 wt% Nb exhibited optimal comprehensive mechanical properties, with a hardness of 386.7 HV, tensile strength of 1060 MPa, yield strength of 962 MPa, and Charpy impact energies of 41.95 J and 33.17 J at −40 °C and −60 °C, respectively. Theoretical calculations revealed that the dislocation strengthening contribution in martensite increased from 526 MPa to 600 MPa with increasing Nb content, representing the dominant strengthening mechanism, while grain refinement strengthening increased from 135.5 MPa to 157.6 MPa. Full article

MD-Hill-SPN is the first Hill-based construction to combine a multi-tier diffusion mix layer, a memory-hard KDF, and a simultaneous multi-metric empirical evaluation. Two independent runs of the full metric suite yield: (a) full plaintext avalanche from round 1 (mean 63.97–64.67 of 128 bits, ideal 64); (b) the differential-probability sampling floor of 2 × 10⁻⁵ reached at round 4 (50,000 of 50,000 output differences distinct, both sessions); (c) algebraic-degree lower-bound saturation at the maximum observable value from round 1; (d) linear-bias indistinguishable from random (combined exceedance 4.40%, below the 4.55% noise floor); and (e) branch numbers at the Singleton (MDS) bound for every tier (B = 5 for 4 × 4, B = 9 for 8 × 8, B = 17 for 16 × 16), computed exhaustively over weight-1 inputs. MD-Hill-SPN therefore moves beyond theoretical construction to a construction that passes a defined empirical evaluation suite: avalanche, differential sampling, linear-bias probing, algebraic-degree lower bounds, and MDS branch numbers under single-key, known-plaintext conditions with fixed parameters, an evaluation no prior Hill cipher variant has reported in full. Full article

Human activity recognition (HAR) using Wi-Fi channel state information (CSI) offers a privacy-preserving and contactless sensing modality suitable for smart homes, healthcare monitoring, and pervasive mobile Internet of Things (IoT) environments. However, existing CSI-based HAR approaches often suffer from computational inefficiency, high latency, and limited feasibility on resource-constrained embedded platforms. This work presents STAR (Sensing Technology for Activity Recognition), an edge AI-optimized framework that integrates lightweight temporal modeling, adaptive signal processing, and hardware-aware co-optimization to enable real-time, energy-efficient HAR on low-power embedded devices. STAR employs a streamlined three-layer Gated Recurrent Unit (GRU) architecture that reduces model parameters by 33% compared to conventional Long Short-Term Memory (LSTM) designs while maintaining strong temporal modeling capability. To enhance signal quality, STAR incorporates a multi-stage pre-processing pipeline consisting of median filtering, an eighth-order Butterworth low-pass filtering, and empirical mode decomposition (EMD) to denoise CSI amplitude measurements and extract stable spatial-temporal features. For on-device deployment, the system is implemented on a Rockchip RV1126 processor equipped with an embedded Neural Processing Unit (NPU) and interfaced with an ESP32-S3 CSI acquisition module. Experimental results demonstrate a mean recognition accuracy of 93.52% across seven activity classes and 99.11% for human-presence detection using a compact 97.6k-parameter model. INT8-quantized inference achieves a processing throughput of 33 MHz with only 8% CPU utilization, achieving a six-fold improvement in inference speed over CPU-based execution. With sub-second response latency and low power consumption, the system ensures real-time, privacy-preserving HAR, offering a practical, scalable solution for mobile and pervasive computing environments. Full article

Drought is a slowly evolving, multi-driver hydro-meteorological hazard whose accurate early prediction is a cornerstone of climate-smart agriculture and water-resource planning. Existing data-driven drought forecasting frameworks suffer from three persistent limitations: (i) most models concatenate heterogeneous climate variables into a single flat feature vector, implicitly assuming a single dominant driver such as precipitation, even though atmospheric moisture demand, radiation and wind-mediated evapotranspiration co-determine drought onset; (ii) wavelet preprocessing is typically applied to the full series, introducing future-information leakage that violates the operational causality requirement of forecasting; and (iii) most architectures predict a single horizon and provide no causal attribution explaining when, where and which climatic variables initiated the event. This study proposes WaveDroughtNet, a multi-modal, multi-horizon deep-learning framework that addresses these limitations through five integrated components: (a) a strictly causal Daubechies-4 wavelet decomposition computed in a rolling fashion; (b) six modality-specific encoders with stochastic modality dropout (p = 0.15); (c) cross-modal multi-head attention with four heads; (d) a four-layer temporal convolutional network (TCN) backbone with dilation factors yielding a 240-step receptive field; and (e) a post hoc DroughtOriginTracer that combines temporal attention, modal-attribution and inter-district propagation scans. The Standardised Precipitation Evapotranspiration Index (SPEI), used as the supervisory target, is computed following the canonical Vicente-Serrano formulation. water balance $D=P-\mathrm{P}\mathrm{E}\mathrm{T}$ (Hargreaves PET) at a 4-week (≈1-month) timescale, fitted with a three-parameter log-logistic distribution via L-moments, validated by Kolmogorov–Smirnov goodness-of-fit testing ($\alpha =0.05$) per district, and standardised through the inverse-normal cumulative distribution function. Trained on 18,304 weekly district records from NASA POWER reanalysis (2014–2025) covering all 32 districts of Tamil Nadu, India, WaveDroughtNet uses only 256,869 parameters and produces, in a single forward pass, four forecasts (1 week, 1 month, 3 months, 1 year). On the held-out 2024 test partition ($N=1728$), the model attains weighted $F1=0.9221$ and $R^2=0.8512$ at the 1-week horizon, and weighted $F1=0.8498$ and $R^2=0.6812$ at the 1-year horizon. Diebold–Mariano tests confirm that WaveDroughtNet significantly outperforms naive persistence, seasonal naive, LSTM, ConvLSTM and a vanilla Transformer at the 3-month and 1-year horizons (p < 0.001). The DroughtOriginTracer successfully back-projects 15 Coimbatore events to causal origins 29–41 weeks prior to onset. We explicitly acknowledge three limitations that constrain operational deployment in its current form—zero severe events in the 2024 test partition ($F{1}_{severe}$ = 0.000), static inter-district modelling, and absence of vegetation-index supervision—and propose concrete mitigation pathways in the Discussion. Full article

Controllable symbolic music generation must preserve a reference melody while remaining responsive to style prompts. Existing hierarchical diffusion systems typically reuse a shared condition vector across harmony, rhythm, and timbre stages, which can entangle stylistic factors and weaken melody preservation. We present HCDMG++, a hierarchical diffusion framework that addresses these two limitations through stage-aware style routing and differentiable melody regularization. The routing module uses a residual multi-layer perceptron (MLP) with zero-initialized scalar gates to project text-derived style embeddings into harmony-, rhythm-, and timbre-specific subspaces, whereas the regularization branch aligns soft pitch histograms and contour trajectories with the conditioning melody during training without breaking the differentiable computation graph. We evaluate the integrated system on a 384-sample benchmark covering four melodies, eight styles, four random seeds, and three denoising budgets, supplemented by a matched legacy-compatible reference and inference-time component ablation that contrasts legacy behavior, silenced gates, an automated uniform gamma routing sweep, and the full forward pass. HCDMG++ produces valid four-track outputs in all 384 runs, reaches a peak pitch histogram similarity score of 0.508 under a 64-step budget, and improves pitch histogram alignment over Legacy-HCDMG by roughly two orders of magnitude on the matched slice, while attaining a positive Fisher-style style separability score where the legacy benchmark is too sparse to support one. These results indicate that stage-specific conditioning and differentiable structural guidance jointly improve controllability in symbolic music diffusion, while also exposing the remaining limitations in long-form generalization and perceptual validation, which motivate the future work outlined at the end of this paper. Full article

Real-time energy management for large electric vehicle (EV) clusters requires both fast aggregate flexibility estimation and executable per-vehicle dispatch. Classical LP/MILP/MPC formulations provide strong feasibility and optimality guarantees when the model is fully specified, but their online solve time increases rapidly with cluster size; learning-based methods are fast but often rely on soft constraint penalties or external feasibility repair. We propose the Physics-Constrained Mamba-3 MIMO Aggregator (PC-M3), an amortized, constraint-aware sequence model that integrates a MIMO Mamba backbone, a history-dependent differentiable projection, a sparse routing layer, and an aggregation–disaggregation consistency loop, scaling AI-EMS from a single battery to ten-thousand-vehicle clusters in one forward pass. PC-M3 assigns every EV to one channel of a multi-input multi-output (MIMO) state-space recurrence and embeds the per-vehicle state-of-charge, power and energy constraints as a differentiable in-loop projection, jointly producing the cluster-level flexibility envelope and the per-vehicle charging trajectory. A sparse Routing-Mamba mixture-of-experts layer adaptively allocates capacity to behaviourally distinct sub-populations without supervised labels, and a consistency-trained aggregation–disaggregation loop binds the predicted envelope to the executed dispatch, forming a digital-twin-style predictive EMS pipeline that couples cluster dispatch with per-vehicle SoC evolution. On a single NVIDIA A100, PC-M3 sustains 0.34 s inference for 10,000 EVs over a 24-h horizon, about 18× faster than an Informer baseline and 2.4× faster than PowerMamba. Evaluated on the open ACN-Data and ElaadNL workplace and public charging corpora and on a 10,000-vehicle NREL dsgrid-TEMPO 2030 stress test, PC-M3 reduces the normalised envelope Hausdorff distance from 9.7% (PowerMamba) to 3.4%, cuts closed-loop cluster tracking RMSE from 1.45 MW (model predictive control) to 0.82 MW, and maintains zero observed feasibility violations with respect to the specified or imputed per-vehicle polytopes on every evaluated session. The framework provides a scalable, predictive, constraint-aware AI-EMS for V2G/G2V virtual-power-plant operation of large EV fleets. Full article

Diabetic retinopathy (DR) and diabetic macular oedema (DME) are two of the most significant preventable contributors to blindness in the adult population worldwide, yet current automated screening systems typically address each condition in isolation and rely on a single imaging modality. In this study, we propose a deep learning model that simultaneously grades DR severity and detects DME by fusing paired colour fundus and optical coherence tomography (OCT) images acquired from the same eye during the same clinical visit. Our architecture employs two parallel EfficientNet-B0 backbones pre-trained on ImageNet, one for each modality, whose 1280-dimensional feature vectors are concatenated into a 2560-dimensional joint representation. This fused representation passes through a shared fully connected block before branching into a three-class DR classification head and a binary DME detection head. We train and evaluate the model on a private dataset of 425 paired fundus and OCT eye images (850 images). The proposed architecture adopts feature-level fusion, in which modality-specific deep features are independently extracted from fundus and OCT images using separate convolutional backbones and subsequently concatenated to form a joint representation for multi-task learning. On the held-out test set (n= 85), the fusion model achieves 82.4% DR accuracy (area under the receiver operating characteristic curve [AUC] = 0.929, macro sensitivity = 0.81, macro specificity = 0.905) and 97.6% DME accuracy (AUC = 0.999, sensitivity = 0.833, specificity = 1.000). The fusion model detects 10 of 12 DME-positive eyes compared with only 7 of 12 for either the fundus-only or OCT-only baselines, representing a 43% relative improvement in DME sensitivity. Stratified five-fold cross-validation (n = 425 aggregated predictions) corroborates these findings, with the fusion model reaching 87.1% DR accuracy (AUC = 0.978) and 99.1% DME accuracy (AUC = 1.000). Gradient-weighted class activation mapping visualisations confirm that the fundus branch attends to clinically relevant macular lesions, whereas the OCT branch highlights retinal layer disruptions and subretinal fluid, providing interpretability. To the best of our knowledge, the proposed MultiRetNet is the first lightweight, task-specific multimodal architecture to jointly grade DR severity and detect DME from paired same-eye, same-visit fundus and OCT images through explicit feature-level fusion within a single end-to-end multi-task framework, distinct from recent generalist ophthalmic foundation models, supporting the value of multimodal fusion for comprehensive diabetic eye screening pending external validation. Full article

316LN stainless steel is widely used in critical nuclear fusion structural components due to its excellent mechanical properties and machinability. However, its high thermal expansion coefficient and low thermal conductivity promote welding distortion, while work hardening causes residual stress accumulation. Thermo-elastic–plastic finite element modeling (FEM) is the primary numerical method for predicting these effects. Yet, despite hardware advances, full-scale simulations—especially for thick plates with multi-pass welds—remain computationally expensive, hindering the balance between efficiency and accuracy. To address the inherent trade-off between welding efficiency and dimensional accuracy in multi-pass, multi-layer welding of thick-section components, this study employs MSC. Marc to develop a finite element model of a 15 mm thick butt-welded joint fabricated from 316LN stainless steel. Three distinct heat source models—instantaneous, enhanced moving, and moving element-set—are systematically implemented to simulate transient temperature fields, residual stress distributions, and welding deformation. All numerical predictions are rigorously validated against experimental measurements to comprehensively assess both accuracy and computational efficiency. Results indicate that: (i) the predicted molten pool geometries and characteristic thermal cycle profiles from all three models exhibit strong agreement with experimental observations; (ii) longitudinal residual stress distributions predicted by all models align closely with measured values; (iii) transverse residual stresses predicted by the moving element-set and enhanced moving heat sources agree well with experiments, whereas those from the instantaneous heat source show marked deviation; (iv) angular distortion predictions from the moving element-set heat source achieve over 90% conformity with experimental data, while the instantaneous heat source substantially underestimates angular distortion, and the enhanced moving heat source yields approximately 65% agreement; and (v) in terms of computational efficiency, the instantaneous heat source requires only ~40% of the computation time needed by the moving heat source. Full article

Aspect-Based Sentiment Analysis explores sentiment trends related to specific opinion aspects and holds significant commercial potential for monitoring brand reputation, understanding customer satisfaction, and personalizing recommendations. However, traditional methods rely exclusively on textual input and often struggle when the target aspect is not mentioned in the sentence. Multimodal Aspect-Based Sentiment Analysis addresses this limitation by incorporating both textual and visual modalities to enable more comprehensive sentiment understanding. Despite advancements in deep learning and transformer-based architectures, existing models often suffer from suboptimal modality fusion and weak aspect grounding, limiting their classification accuracy. To overcome these challenges, we propose an Attention-Driven Feature Fusion (ADFF) approach based on a three-stage hierarchical attention mechanism. First, it only fuses text and image embeddings. Second, it incorporates aspect-level features. Third, a multi-head attention layer further enhances cross-modal dependencies. The resulting representation is passed to a Long Short-Term Memory (LSTM) classifier for sentiment polarity prediction. We evaluate our model on three benchmark datasets, namely Twitter-2015, Twitter-2017, and MASAD. The experimental results demonstrate that the proposed model substantially outperforms state-of-the-art multimodal and unimodal baselines, improves both accuracy and F1-score, achieving 82.55% accuracy and 81.05% F1-score on Twitter-2015, 77.07% accuracy and 77.15% F1-score on Twitter-2017, and up to 99.67% accuracy and F1-score in the Plant domain of MASAD, where we observe consistent improvements across all seven domains. These results highlight the effectiveness and scalability of the hierarchical attention-based fusion strategy for real-world aspect-based sentiment analysis tasks. Full article

The controllable fabrication of tapered three-dimensional (3D) microstructures by ion-beam processing remains challenging, especially when both profile fidelity and geometric controllability are required. Tapered conical microstructures are of interest because they are relevant to a variety of applications, including micro-optical elements, functional textured surfaces, biomimetic interfaces, and field-enhancing emitter-related structures, where taper angle, aspect ratio, and structural uniformity strongly influence the resulting performance. In this work, a longitudinal layer-wise strategy is proposed for tapered micro-cone fabrication by ion-beam etching. The core idea is to discretize a continuous cone profile along the vertical direction into a sequence of annular layers whose dimensions are determined by the local geometry of the target three-dimensional structure. After this geometric discretization step, each individual layer is executed using a conventional multi-pass strategy, thereby combining longitudinal profile construction with stabilized local material removal. A dedicated pattern-design software, EBWriter, was developed to automatically generate annular patterns and process files from user-defined geometric parameters. Experimental validation was carried out on single-crystal silicon substrates using a dual-beam microscope platform operated at 30 kV. The results show that increasing the longitudinal layer number effectively weakens the staircase effect and improves the continuity of the reconstructed cone profile. For positive micro-cones fabricated using annular patterns with a nominal outer processing diameter of 3 μm, the increasing-inner-radius strategy enables preservation of the cone apex and reconstruction of tapered morphologies with improved fidelity. Under the present processing conditions, an empirical correspondence between the target geometric ratio and the recommended layer number was further summarized: layer numbers of approximately 50, 100, and 300 support cone structures with base-diameter-to-height ratios close to 1:2, 1:3, and 1:4, respectively. In addition, a 3 × 3 positive micro-cone array was successfully fabricated, with a total processing time of about 80 s. The measured cone base diameter and height were 0.886 ± 0.005 μm and 2.354 ± 0.023 μm, respectively, with dimensional variations controlled within ±2%. These results demonstrate that the proposed method provides a feasible layer-wise ion-beam fabrication route for tapered microstructures and offers a useful process basis for future studies on micro-optical surfaces, functional textured interfaces, and emitter-related microstructures. Full article

Unsupervised anomaly detection in industrial fabric inspection remains a formidable challenge due to the complexity of background textures and the subtle, irregular nature of real-world defects. Although the teacher-student distillation paradigm has demonstrated promising performance without reliance on anomalous data, existing methods still struggle in the presence of complex textures, largely due to limited semantic guidance, insufficient frequency modeling, and inadequate multi-scale representation. To address these limitations, we propose a novel reverse distillation framework tailored for fabric defect detection. The core of our method is the frequency decoupling Feature fusion module (FDFM), which achieves frequency domain alignment between teacher and student features through spatially adaptive and learnable filter banks, namely the adaptive high-pass filter (AHPF) and the adaptive low-pass filter (ALPF). Specifically: (1) the high-frequency pathway employs deconvolutional residual enhancement to emphasize boundary details; (2) the low-frequency pathway leverages the CARAFE operator to Handle these normal fluctuations to prevent the model from mistakenly identifying background changes as abnormal areas. This design not only maintains a lightweight architecture but also significantly improves sensitivity to fine-grained anomalies. Furthermore, we introduce a cross-layer residual alignment mechanism that guides the student network in reconstructing deep semantic representations from the teacher-student feature pairs. To balance detection accuracy and deployment efficiency, we develop two model variants: a high-capacity version optimized for precision, and a lightweight version tailored for real-time industrial applications. Compared with other methods from recent years, the experimental results of FAD-RNet validate its superiority in relevant metrics. It should be noted that this study is conducted based on the data organization and processing protocol of the ZJU-Leaper dataset, which may introduce certain dataset-specific characteristics. Full article

The effect of treatment time on arc-enhanced glow discharge plasma-assisted nitriding (AEGD-PAN) of AISI M2 high-speed steel was investigated for non-heat-treated and heat-treated substrates. Nitriding treatments were carried out at 350 °C for 1.5 and 3.5 h, producing diffusion layers with thicknesses ranging from approximately 38 to 75 µm without formation of a continuous brittle compound layer. X-ray diffraction combined with Rietveld refinement revealed the progressive formation of γ′-Fe₄N and ε-Fe₂₃N nitrides together with lattice expansion of the α-Fe matrix, indicating nitrogen supersaturation and precipitation strengthening within the diffusion zone. Heat-treated specimens exhibited higher surface hardness, reaching ~1350 HV_0.1, while non-heat-treated substrates developed pronounced hardness gradients associated with diffusion-controlled layer growth. Scratch testing showed improved resistance to contact-induced damage with increasing nitriding time, particularly for the 3.5 h treatment, where lateral cracking was significantly reduced and load-bearing capacity increased. Multi-pass scratch wear tests revealed a reduction in the Archard wear coefficient by up to four orders of magnitude compared with untreated M2 steel. These results demonstrate that AEGD-PAN at moderate temperature enables efficient diffusion layer formation and significant improvement in the tribological performance of high-alloy tool steels. Full article