American Journal of Applied Science and Technology

Hybrid Machine Learning Architecture for SAP Production Planning Optimization and Lean Manufacturing Performance

2026-09-01T10:32:17+00:00

Modern production planning increasingly requires intelligent decision-support mechanisms capable of integrating demand variability, production constraints, inventory conditions, and operational performance objectives. SAP-based production planning provides an enterprise-level environment for coordinating these activities, but conventional planning logic may be insufficient when manufacturing systems exhibit nonlinear demand patterns, fluctuating capacities, bottlenecks, and complex interactions among planning variables. This research proposes a Hybrid Machine Learning Architecture for SAP Production Planning Optimization and Lean Manufacturing Performance that combines predictive learning, feature-aware representation, residual learning, attention mechanisms, and optimization-oriented decision support. Because the supplied literature primarily concerns machine learning architectures for medical-image segmentation rather than SAP or manufacturing, these studies are treated as methodological foundations rather than direct evidence of SAP performance. In particular, residual learning, attention mechanisms, recurrent contextual learning, and encoder–decoder architectures provide transferable principles for constructing a hybrid analytical architecture (He et al., 2016; Hu et al., 2018; Cai et al., 2019; Ronneberger et al., 2015). The proposed framework uses SAP planning data to generate forecasts, estimate production risks, identify operational constraints, and recommend planning adjustments. A conceptual evaluation demonstrates how the architecture can be assessed through forecasting accuracy, inventory efficiency, throughput, schedule stability, and lean-performance indicators. The study contributes a structured research framework for integrating machine learning with enterprise production planning while emphasizing interpretability, data quality, scalability, and operational feasibility.

Conditions for The Safe Installation of a Hydrogen Generator Unit on Board A Vehicle and Its Operational Analysis

2026-09-16T11:42:43+00:00

This article scientifically and technically investigates the conditions for the safe installation of a hydrogen generator (electrolyzer) unit on board vehicles, particularly commercial transport vehicles and hybrid buses. Although the use of a hydrogen-air mixture in internal combustion engines (ICE) increases thermal efficiency and reduces toxic gas emissions, integrating the unit on board requires serious mechanical and gas-dynamic safety measures. During the research, issues concerning the strength of the unit's brackets (stresses and deformations), vibration protection, as well as the viscosity properties of the electrolyte and the diagnostics of hydrogen leaks were examined. The results indicate that proper placement of the generator and optimal application of the sensor system make it possible to fully achieve operational safety and environmental standards.

Influence of Linear Dimensions and Technological Preparation Parameters of Cotton Seeds on Hulling Efficiency and Reduction of Oil Losses

2026-09-16T06:30:53+00:00

The article presents the results of a study of the influence of linear dimensions, lint content and moisture content of technical cotton seeds on the efficiency of the processes of hulling and subsequent separation. It has been established that the optimal kernel moisture content for seeds of grades I-III is 8.5-9.5 %, and for grade IV - 9.5-10.5 %, which ensures minimal oil losses with the husk. It is shown that when the lint content of seeds is less than 8 %, hulling units operate at 100 % of their rated capacity, whereas when this indicator is exceeded, it is advisable to reduce the load to 80-85 %.

A Transparent Predictive Model for SSD Failure Detection Using Local and Global Explainability

2026-09-07T05:28:54+00:00

Solid-state drive (SSD) reliability is increasingly important in computing environments where storage failures can cause data loss, service interruption, and substantial operational costs. Conventional predictive models can identify failure patterns from telemetry and health indicators, but their practical adoption is constrained when predictions cannot be interpreted by engineers and system administrators. This paper proposes a transparent predictive architecture for SSD failure detection that integrates predictive modeling with local and global explainability. The framework distinguishes between instance-level explanations, which identify why a particular SSD is classified as approaching failure, and global explanations, which reveal the general relationships between storage-health attributes and predicted failure risk. The theoretical foundation is motivated by the broader development of interpretable computational systems and transparent model reasoning. The methodology establishes a pipeline encompassing telemetry acquisition, preprocessing, feature construction, predictive classification, local explanation, global feature analysis, consistency assessment, and decision-oriented interpretation. LIME and SHAP are positioned as complementary mechanisms for explaining individual predictions and aggregate model behavior, following the transparency-oriented perspective established by Kumar (2026). The resulting framework is designed to transform an otherwise opaque failure score into an actionable diagnostic representation. The analysis indicates that combining local and global explanations can improve diagnostic usefulness, facilitate model validation, and support human oversight. However, explainability does not inherently guarantee causal validity, and the proposed architecture therefore emphasizes explanation consistency, operational context, and careful interpretation.

Modeling Security Threats in Enterprise Federated Identity Systems: Attack Vectors, Vulnerabilities, and Mitigation Strategies

2026-09-02T12:04:53+00:00

Enterprise information environments increasingly depend on federated identity architectures to provide centralized authentication and authorization across heterogeneous applications, organizational domains, and cloud services. Although federation and multi-factor authentication (MFA) reduce several weaknesses associated with isolated credential management, they also create concentrated trust relationships and technically complex attack surfaces. This research and review paper develops a structured threat-modeling perspective for enterprise federated identity systems by examining authentication flows, federation trust relationships, token-processing components, identity providers, service providers, and MFA mechanisms as interconnected security assets. A STRIDE-oriented analytical model is employed to classify major threat categories, including spoofing, tampering, repudiation, information disclosure, denial of service, and elevation of privilege. The methodology combines architectural decomposition, attack-surface identification, threat classification, vulnerability assessment, and mitigation mapping. The theoretical foundation is strengthened through the supplied literature on robust estimation, nonsmooth optimization, piecewise-affine modeling, machine-learning behavior, and statistical learning. These works provide useful methodological perspectives for analyzing uncertain, nonlinear, and heterogeneous security environments, although they do not directly investigate federated identity. The analysis indicates that the most consequential risks arise at trust boundaries, token issuance and validation points, MFA recovery processes, administrative interfaces, and identity-provider dependencies. The study further demonstrates that security controls should be evaluated as an interconnected system rather than as independent authentication mechanisms. The resulting framework provides a systematic basis for identifying attack vectors, prioritizing vulnerabilities, and designing layered mitigation strategies for enterprise federated identity infrastructures.

Improving the Effectiveness of Geological and Technical Measures and Optimizing Development of The UMID Oil Gas Condensate Field

2026-09-16T11:45:42+00:00

This study evaluates geological and technical measures at Uzbekistan’s Umid oil–gas–condensate field using retrospective dissertation data. The effective oil-saturated thickness is 6.7 m. Reported concurrent oil and gas-condensate development was associated with annual oil production increasing from 14.8 to 59.0 thousand tonnes. After treatment, reported gas rates rose by 20 and 45 thousand m³/d in wells 78 and 12 and oil production by 0.5 t/d in well 77. High water cut and limited oil-well availability constrain performance. Coordinated gas withdrawal, well-specific interventions, and gas-lift assessment are priorities; sustained recovery and economic benefits remain unverified.

Effects of Harrow Tooth Length and Mass Per Tooth on The Performance of a Harrow-Roller Implement

2026-09-16T11:39:47+00:00

This study presents experimental results on the effects of harrow tooth length and mass per tooth on the agrotechnical and energy performance of a harrow-roller. The evaluated indicators included tillage depth and its standard deviation, degree of soil crumbling, standard deviation of field-surface irregularity heights, and specific draft resistance. The results showed that, under the tested field conditions and travel speeds, a harrow tooth length of 140–160 mm and a mass per tooth of 2.10–2.25 kg provided the required performance.

Leadership of Or Participation in Research Projects and Contractual Research Activities (International and Domestic Research Projects and Contracts)

2026-09-14T03:27:06+00:00

This article analyzes the role of academic and pedagogical personnel in leading and participating in international and local scientific projects, as well as in economic (business) contracts within the modern system of higher education and science. The theoretical foundations of research project management are examined, including university-industry-government cooperation mechanisms based on the Triple Helix model. The national grant system of Uzbekistan (fundamental, applied, and innovative projects), international grant programs (Erasmus+, Horizon Europe), and the practice of economic contracts between higher education institutions and manufacturing enterprises are analyzed. The professional competencies and responsibilities of a research project leader are also discussed.

Contextual Risk-Based Verification for Strengthening JWT-Based Authentication

2026-09-03T15:01:43+00:00

JSON Web Token (JWT)-based authentication provides a scalable mechanism for representing authenticated identity and authorization claims across distributed applications, APIs, and microservices. However, conventional JWT validation generally emphasizes cryptographic validity, issuer, audience, expiration, and token structure while providing limited consideration of the contextual circumstances in which a valid token is presented. This creates an important security limitation: possession of a correctly signed and unexpired token may be treated as sufficient evidence of legitimacy even when the surrounding request context is anomalous. This paper proposes a contextual risk-based verification framework that supplements conventional JWT validation with adaptive assessment of contextual signals, including device characteristics, request origin, temporal behavior, access patterns, session consistency, and resource sensitivity. The methodology conceptualizes authentication as a continuous risk evaluation process rather than a one-time token-validation event. A multi-stage architecture is developed consisting of cryptographic validation, contextual feature extraction, risk scoring, policy evaluation, and adaptive response. The analysis indicates that contextual verification can strengthen JWT-based authentication by distinguishing technically valid tokens from potentially suspicious token usage. The approach also introduces trade-offs involving latency, privacy, false positives, implementation complexity, and policy calibration. The proposed framework therefore positions JWT as one component of a broader adaptive authentication architecture rather than as an independently sufficient security control.