Theory and Applications of Mathematics and Computer Science

Manufacturing Process of the SAV PALET Dacia Renault

Alexandru Puiu Barbos — Tue, 01 Sep 2026 00:00:00 +0300

The paper analyzes the technological workflow used to manufacture the welded SAV PALET G assembly intended for the
Renault/Dacia application, starting with the procurement of semi-finished materials and continuing with cutting, machining on
numerically controlled equipment, turning of the parts, welding, stress relieving, hot-dip galvanizing and final inspection. The
study uses technical data from the attached technological documentation and focuses on the engineering interpretation of the
manufacturing parameters. Calculations are presented for spindle speed, feed rate, theoretical machining time and an example
of dimensional verification. The results show that the sequence of operations must be established so that geometric accuracy,
welded-assembly stability and corrosion protection are maintained throughout the entire manufacturing cycle.

Mathematical Analysis and Structural Optimization of an Insulated Support for the 25 kv–50 hz Railway Contact Line

Alexandru Puiu Barbos — Tue, 01 Sep 2026 00:00:00 +0300

This paper presents a structural analysis of an insulated support assembly used in 25 kV–50 Hz railway overhead contact
lines. The study is based on the supplied engineering drawings of the stay rod, brace, direct steady arm, reverse fixing device,
bar, couplings, fixing plate and contact-wire clamp. The dimensions and materials explicitly readable in the drawings are incorporated
into a simplified analytical model for bending, shear and elastic displacement. The reference configuration mainly uses
S235/S235JR/FMN35 steels with corrosion protection by hot-dip galvanizing, while CuZn15 and other steels/stainless components
are present in the contact clamp and fasteners. AlCu4Mg is investigated as a lightweight alternative for selected structural members.
The analysis confirms the substantial mass-saving potential of aluminium alloys, but also the associated loss of sti ness;
therefore, a material substitution must be accompanied by geometric redesign. For the ?34 4 mm tubular direct steady arm, with
L = 400–1600 mm, the section properties are calculated and a conservative Fd = 90 N case is used to quantify the sensitivity of
bending moment and elastic displacement to length. The proposed methodology combines mass, strength and sti ness criteria and
can be extended by finite-element analysis.

Mathematical Modeling, Structural Resistance Calculation, and Technological Itinerary for Manufacturing the Y25 Bogie Frame

Paul Ilie Varsendan — Tue, 01 Sep 2026 00:00:00 +0300

This paper presents an analytical and mathematical approach to the dynamic stress phenomena acting on the supports and
protection devices of railway track inductors. The main objective of the research is the mathematical substantiation of the transition
from the use of conventional structural steels (S235J2) to superior alloys (S355J2) in railway infrastructure, through a rigorous
analysis of stress tensors and safety coecients. Based on equilibrium equations and bending models under transient shocks and
accelerations, it has been demonstrated that adopting the S355J2 material increases the structural safety coecient by over 50%,
ensuring the continuous operation of trac safety systems (PZB/Indusi).

Mathematical Analysis of Dynamic Stresses and Structural Optimization of Protection Devices for Railway Track Inductors

Paul Ilie Varsendan — Tue, 01 Sep 2026 00:00:00 +0300

TrueVision: A Human-Aligned Explainable Visual Completion Dataset for Incomplete Artworks

Roxana Buzuriu, Darian M. Onchis, Eduard Hogea — Fri, 18 Sep 2026 00:00:00 +0300

Large vision-language models are increasingly used in domains where the input image is incomplete, including damaged
paintings, cropped artworks, and restoration scenarios. In these settings, a plausible answer is not sucient: the model should
distinguish between visible evidence and inferred content, communicate uncertainty, and abstain when the available cues are too
weak. This paper presents the current repository-backed stage of TrueVision, a bilingual English-Romanian dataset and annotation
platform for explainable visual completion on artworks and other incomplete images. The current reportable experimental bundle
contains a shared 100-task overlap annotated by three production annotators, yielding 300 bilingual annotation instances and
deterministic 76/10/14 train-validation-test splits. The live platform already supports a richer structured workflow that separates
visible cues from inferred hidden content in both languages, but the committed gold overlap still follows the earlier single-prediction
annotation format and is therefore the basis of the quantitative results reported here. The paper includes four reproducible local
baselines: zero-shot and few-shot Qwen2-VL 2B prompting, a metadata-aware retrieval baseline, and a lightweight supervised
visual baseline trainable on Apple Silicon. On the held-out test split, retrieval remains the strongest non-learned reference in
English similarity (0.4820), while the trained baseline provides the strongest learned result, with 0.4746 Romanian similarity,
0.4507 English token F1, 0.3223 Romanian token F1, 0.5093 confidence macro-F1, and 0.4853 diculty macro-F1. These results
show that the TrueVision pipeline is operational end to end and already supports controlled comparison of explainable visual
completion methods in a low-resource bilingual setting.

Convolutional Neural Networks for Cardiac MRI Analysis: An Overview

Marcela Florea — Fri, 18 Sep 2026 00:00:00 +0300

Cardiac magnetic resonance imaging (cardiac MRI) provides detailed information about cardiac anatomy and function, but
analysing a large number of images can be time-consuming when performed manually. Deep learning methods, particularly convolutional
neural networks (CNNs), are widely used for automated medical image analysis. This paper provides an overview of
CNNs and their applications in cardiac MRI analysis, with a focus on cardiac image segmentation. The paper first presents the
basic principles of CNNs, including convolution, pooling and activation functions, followed by a discussion of U-Net and its use
in biomedical image segmentation. Applications of CNNs in cardiac MRI, including cardiac structure segmentation, disease classification
and quantitative assessment, are also discussed. Recent developments such as 2D and 3D CNNs, attention mechanisms
and hybrid CNN-Transformer architectures are presented. The paper also discusses current challenges, including limited annotated
datasets, di erences between imaging settings, generalization, interpretability and clinical validation. Overall, CNNs remain important
tools for cardiac MRI analysis, while recent research is exploring more robust models that can be applied across di erent
clinical settings.

Human-AI Collaboration in West-Romanian SMEs: A Qualitative Study of Managers’ Early-Stage Adoption, Risks and Value

Miruna Maura Trocan, Alexandru Nicolaita, Mircea Pele, Cosmin-Horatiu Danciu — Fri, 18 Sep 2026 00:00:00 +0300

This study focuses on Small and Medium Enterprises (SMEs) from the western side of Romania. The current research starts
from the hypothesis that SMEs started to experiment with generative artificial intelligence at a growing rate, in order to reduce costs
and accelerate customer communication. However, evidence from owner-managed companies in Central and Eastern Europe is
limited. This study addresses this research gap by bringing evidence gathered using the qualitative methodology. Evidence on this
topic was collected in Arad, Romania, and examines how managers in the service-sector SMEs adopt AI in their day-to-day work,
what benefits and risks they perceive, and which routines enable a responsible, human-in-the-loop use. This research is conducted
from a managerial point of view, showing the challenges that managers try to address and the backlash that some of them faced
from their own employees. It is important to mention that this study focuses on how these managers and their employees use AI as
a tool in their companies, not as a replacement for their human employees.
This study provides understanding of the benefits and challenges that are encountered by real-life managers and why it is
important to proceed with caution and adopt new technologies responsibly. Limitations of this study include a single region and
reliance on self-reported commercial outcomes. Future research should include evidence from a wider region.