课程文献中心 更多>>

关键词： computer-aided design   design automation   vision-language models   artificial intelligence  

摘要： Efficient creation of accurate and editable 3D CAD models is critical in engineering design, significantly impacting cost and time to market in product innovation. Current manual workflows remain highly time consuming and demand extensive user expertise. While recent developments in AI-driven CAD generation show promise, existing approaches are often constrained by limited CAD representations, inability to generalize to real-world images, or low output accuracy due to a lack of domain-specific knowledge. To address these challenges, this article introduces CAD-Coder, an open-source vision-language model (VLM) explicitly fine-tuned to generate editable CAD code (CadQuery Python) directly from visual input. Leveraging a large-scale dataset that we constructed-GenCAD-Code, consisting of over 163k CAD model image and code pairs-CAD-Coder outperforms VLM baselines such as GPT-4.5 and Qwen2.5-VL-72B on image-conditioned CAD code generation, achieving a 100% valid syntax rate and the highest accuracy in 3D solid similarity. Notably, our VLM exhibits initial signs of generalizability, generating CAD code from real-world images and utilizing a CAD operation not explicitly seen during fine-tuning. The performance and adaptability of CAD-Coder highlight the potential of VLMs fine-tuned on code to streamline CAD workflows for engineers and designers. CAD-Coder is publicly available.

关键词： transfer learning   medical image classification   explainable AI   Grad-CAM   dermatology   Dermnet  

摘要： Accurate medical diagnostics for skin affections such as skin cancer, psoriasis, vascular tumors, or exanthems have become increasingly difficult due to the growing volume and visual variability of dermatological cases, as well as limited specialist availability. To address this, the present work introduces a complete and deployable deep-learning-based system capable of detecting ten distinct skin disease categories, trained using transfer learning with EfficientNet-B5 and enhanced with explainable AI through Grad-CAM. The proposed system achieves a top-3 accuracy of 95.96%, a weighted F1-score of 0.87, and class-specific F1-scores reaching 0.96 for acne and 0.95 for nail fungus. These results demonstrate strong predictive performance for the deep learning model trained, validated, and evaluated on a ten-class subset of the Dermnet dataset. The research conducted covers the visual explainability of the AI model classification process, including integration into a fully functional web application, usable as an expert system for image uploading, data processing and visualization of results. The AI visualizing technology based on Grad-CAM provides clear, class-specific heatmaps that highlight the most influential regions in each prediction, improving transparency and supporting clinical interpretability.

关键词： computer-aided design   computer-aided manufacturing   hydrofluoric acid   air abrasion   dental   composite resins   dental bonding  

摘要： In computer-aided design/computer-aided manufacturing (CAD/CAM) restorations for severely damaged teeth, the cavity floor or proximal margins may be elevated with composite resin to improve adhesion. This in vitro study investigated how different surface pretreatment methods affect the shear bond strength (SBS) of hybrid CAD/CAM materials to dentin or composite surfaces, simulating clinical situations of composite elevation. Hybrid CAD/CAM samples were bonded to dentin or composite substrates following different surface pretreatment protocols and cemented using a dual-cure adhesive resin cement. The samples were thermocycled and subjected to shear bond strength testing, and failure modes were analyzed. The SBS in the sandblasting (SB)+Dentin group and hydrofluoric acid (HF)+Dentin was significantly higher than that in the SB+Composite and HF+Composite groups (p < 0.05). Untreated+composite and untreated+dentin groups showed significantly lower SBS (p < 0.05). Failure mode analysis revealed a predominance of cohesive failures in the SB+Dentin group, while adhesive failures were more frequently observed in most of the other groups. SB-treated and HF-etched hybrid CAD/CAM materials showed more favorable bonding behavior to dentin than to composite, highlighting that bonding to the elevated composite layer may be less effective than bonding directly to prepared dentin.

关键词： Bond strength   PEEK   Post-core   polyetheretherketone   Push-out  

摘要： This study aimed to evaluate the bond strength to dentin of dental posts fabricated with different PEEK using the push-out test, while also examining the failure modes. Sixty mandibular premolars extracted for periodontal reasons were randomly allocated into three groups (n = 20 each). The custom post-cores were fabricated using CAD/CAM from 3 PEEK disks: unfilled PEEK (Group UP); 20% TiO2-filled PEEK (Group TP); 20% ceramic-filled PEEK (Group CP). Monobond Plus and *** adhesives were used for bonding. Following cementation, 180 dentin disks were obtained by sectioning each tooth into 1 mm segments from the coronal, middle, and apical thirds. Subsequently, the push-out bond strength (PBS) test was performed. Data were analyzed using MANOVA and Tukey’s HSD tests (α = 0.05). There was no significant difference between the bonding agents in Group CP. In Group UP, *** exhibited significantly higher push-out bond strength (PBS) than Monobond Plus in the middle and coronal thirds, while in Group TP, this superiority was observed in the apical and middle thirds. Among the tested post materials, Group CP demonstrated the highest overall PBS values, which were significantly greater than those of Group TP, though not significantly different from Group UP. The highest PBS values were consistently found in the coronal third, followed by the middle and apical thirds. Ceramic-filled PEEK posts achieved the greatest bond strength, and *** outperformed Monobond Plus as a bonding agent. A gradual decrease in PBS from the coronal to the apical thirds was observed across all groups. Within the limitations of this in vitro study, ceramic-reinforced PEEK posts provided the highest push-out bond strength to root dentin. The *** bonding agent demonstrated superior performance compared to Monobond Plus, and the PBS of PEEK posts decreased progressively from the coronal to the apical third. Careful selection of both the PEEK post type and the bonding agent i

关键词： Antioxidants enzymes   Chlorophyll a fluorescence   Mesembryanthemum crystallinum   PEPC   Plant pigments   RuBisCO  

摘要： Facultative CAM plants possess a unique ability to reversibly initiate and suppress beta-carboxylation, a metabolic pathway characterised by nocturnal stomatal opening and high water-use efficiency. In the present study, we focused on determining how continuous light modifies the interaction of the Mesembryanthemum crystallinum, a model CAM facultative species, with salinity stress - an ever-growing global-scale plant diversity and productivity threat. Expression of stress-dependent PEPC form and RbcL, chlorophyll a fluorescence, pigment composition and antioxidant enzyme activity were analysed. Expression of both mentioned genes was predictably influenced by salinity stress, with stress-dependent PEPC amount increase accompanied by a concurrent RbcL decline. However, exposing plants already affected by salinity stress to continuous light resulted in a limited RbcL decline - a side-effect exclusively induced by salinity stress. Moreover, a limited expression of salinity-dependent PEPC was concomitantly observed;in M. crystallinum leaves, the content of cryptoxanthin, beta-carotene, alpha-carotene, zeaxanthin, violaxanthin, antheraxanthin and neoxanthin has increased, especially in the case of salinity with continuous light treatments. The total SODs activity was increased in plants exposed to salinity combined with continuous light. This study unravels novel evidence that continuous light treatment may alleviate the detrimental effects of high salinity in facultative CAM plants;these conclusions, however, may potentially be extrapolated to other plant groups, including glycophytes.

关键词： Dual-axis roller-cam turntable   Finite element analysis (FEA)   Experimental modal analysis (EMA)   Positioning accuracy   Principal component analysis (PCA)   Intelligent diagnosis  

摘要： The growing demand for high-precision multiaxis machining in aerospace, energy, and advanced manufacturing sectors has heightened the need for rotary indexing systems that offer both structural rigidity and intelligent health management capabilities. However, existing studies typically address structural design, accuracy validation, or fault diagnosis independently, resulting in limited integration between mechanical performance evaluation and data-driven intelligence. This study proposes an integrated evaluation and diagnostic framework for a dual-axis roller-cam rotary indexing table, combining finite element modeling, experimental modal analysis, precision accuracy verification, and principal component analysis (PCA)-based intelligent fault diagnosis within a unified platform. The theoretical foundation of the proposed approach is based on structural dynamics theory and multivariate statistical feature extraction, enabling the coupling of physical vibration behavior with probabilistic classification models. Experimental results demonstrate indexing accuracy within 11.6 '' to 19.8 '' and repeatability below 2.2 '' across all axes after compensation. A PCA-Gaussian mixture model (GMM)-based diagnostic system achieved a classification accuracy of 93.5%, with an area under the ROC curve exceeding 0.92, outperforming conventional FFT- and wavelet-based methods. This study presents one of the first integrated structural-diagnostic frameworks specifically designed for dual-axis roller-cam rotary systems, effectively bridging the long-standing gap between mechanical accuracy validation and intelligent condition monitoring. The proposed methodology establishes a new paradigm by combining physics-based structural verification with data-driven fault intelligence, thereby enabling scalable predictive maintenance and cyber-physical integration in next-generation intelligent machining systems.

关键词： Ceramic   Colour   Glaze   Polish   Roughness   Translucency   Zirconia  

摘要： This in vitro study aims to evaluate the effects of two different surface finishing procedures-glazing and polishing-on the stability of colour, the translucency, and the roughness of surface with three CAD/CAM ceramic materials: lithium disilicate ceramic (LDC), zirconia-reinforced lithium silicate ceramic (ZLC), and translucent zirconia (TrZ). The process is evaluated before and after coffee thermocycling. A total of 120 samples (1.0 ± 0.1 mm) were prepared from LDC (IPS *** CAD HT A2), ZLC (Celtra Duo HT A2), and TrZ (Cercon HT A2). All samples were subjected to standardized grinding for 20 s to simulate chairside adjustments and were then assigned to glaze (G) or polish (P) subgroups (n = 20). Surface roughness (Ra, Rz) was measured by using a contact profilometer, while colour coordinates (L*, a*, b*), colour difference (ΔE₀₀) (∆E00), and translucency parameter (TP) were recorded with a colorimeter. All samples underwent 10,000 coffee thermocycles (5–55 °C). Data were analysed by using one-way ANOVA, repeated-measures ANOVA, Kruskal–Wallis, and appropriate post-hoc tests (α = 0.05). It is seen that grinding produced the highest Ra and Rz values in ZLC, followed by LDC and TrZ. When surface treatments were completed, polished TrZ showed a significantly lower Ra compared with glazed TrZ, while ZLC and LDC exhibited lower roughness after glazing, although not significantly. Only the ZLC-P group demonstrated a perceptible colour change, yet all groups remained within the clinically acceptable threshold (< 2.23). Before and after thermocycling, LDC showed the highest TP values, followed by ZLC and TrZ. Thermocycling significantly increased Ra (except in TrZ-P and ZLC-P) and decreased TP values in all materials except LDC-P. Based on the results, it is concluded that surface finishing techniques and coffee thermocycling affect the optical and surface properties of CAD/CAM ceramics. Glazing generally provides smoother surfaces for glass-ceramics, w

关键词： nonlinear vibration isolation   dual negative stiffness   quasi-zero stiffness   wire-rope isolator   roller-cam mechanism   geometric nonlinearity   tetrahedral structure  

摘要： Low-frequency vibration isolation is essential for precision systems;however, conventional linear isolators are limited by the inherent coupling between static stiffness and natural frequency. To overcome this limitation, a dual negative-stiffness quasi-zero-stiffness (DNS-QZS) isolator is developed by integrating a roller-cam mechanism, steel wire-rope elements with hysteretic damping, and a symmetric tetrahedral support structure. An improved Bouc-Wen hysteresis model is formulated to characterize the nonlinear restoring force behavior. Experimental and numerical results show that the proposed isolator begins to attenuate transmitted vibration energy in the low-frequency range and achieves effective vibration isolation when the excitation frequency exceeds approximately 7.7 Hz, where the transmissibility falls below 0 dB and remains suppressed over a broad frequency band. The DNS-QZS isolator exhibits low equivalent stiffness near equilibrium and enhanced damping, enabling stable and broadband low-frequency isolation while retaining sufficient load-bearing capability. These findings demonstrate the practical applicability of the proposed design in advanced vibration control scenarios.

关键词： 3D printing   CAD/CAM milling   MOD inlay   Fracture resistance   Failure mode  

摘要： Objective This study investigated the fracture resistance and mode of failure of CAD/CAM milled versus 3D printed resin composite inlays under the influence of thermocycling. Materials and methods Brilliant Crios CAD/CAM milled and Crowntec 3D printed resin composite materials were used in this study. Twenty sound freshly extracted human mandibular third molars were selected for this study and randomly divided based on manufacturing technique into two groups: CAD/CAM milled, and 3D printed resin composite inlays. A universal testing machine was used to estimate the fracture resistance, and the mode of failure was investigated for fractured specimens. Inferential statistics for evaluating and comparing milling and 3D-printed manufacturing techniques were carried out via independent samples t tests at the 0.05 significance level. Results While the 3D printing group recorded numerically higher fracture resistance, no statistically significant difference was found (p > 0.05). Evaluating the mode of failure of the CAD/CAM milling group specimens revealed that there was no significant difference between all the recorded failure modes (P > 0.05). On the other hand, a significant difference was revealed between the different recorded modes of failure in the 3D printing group (P < 0.05). Conclusions Both 3D-printed, and CAD/CAM milled resin inlays had comparable fracture resistance that exceeded the physiological masticatory force in molar region;however, most of the failure modes recorded for both materials were irreparable failures involving more than half of the tooth and extending below the CEJ. Clinical significance 3D printed resin composite material is a promising option in fabrication of indirect restorations with lower cost and more time saving. Long term clinical studies are recommended to evaluate clinical performance of 3D printed resin composite materials.

关键词： 高光谱成像   机器学习   废旧纺织品   聚酯纤维   氨纶   神经网络  

摘要： 针对复杂废旧纺织品中聚酯纤维和氨纶纤维含量检测精度低、模型易过拟合及泛化能力不足等问题,本文提出了一种基于近红外高光谱成像与优化机器学习算法的快速无损检测方法,用于实现废旧纺织品的自动化识别与分选。采用工作波段为1 000~1 700 nm的推扫式高光谱成像系统采集不同混纺比例样本的光谱数据,并通过黑白板校正、Savitzky-Golay滤波及标准正态变换对数据进行预处理,以降低噪声和散射干扰。针对聚酯纤维含量预测,利用主成分分析提取光谱特征,并构建基于蛇优化算法优化的多层感知机模型(MLP-SO);针对氨纶纤维检测,通过相关性分析筛选敏感波段,结合核主成分分析特征建立随机森林(RF)模型。结果表明,MLP-SO模型在聚酯含量预测中取得决定系数(R2)为0.963、均方根误差(RMSE)为3.546的性能,显著优于传统机器学习模型;RF模型在氨纶含量预测中获得R2为0.901 78、RMSE为1.48,明显优于偏最小二乘回归模型。相关性分析显示1 000~1 100 nm和1 400~1 500 nm为氨纶含量敏感波段,特征可视化结果验证了不同氨纶浓度样本的良好可分性。研究结果表明,高光谱成像结合优化算法可实现复杂废旧纺织品中聚酯与氨纶纤维的快速、无损和定量检测,为纺织品资源回收与绿色循环利用中的高效自动化分选提供了可靠技术支撑。