Javanshir Zeynalov

Javanshir Zeynalov

Auto-generated profile, based on publicly available information

Javanshir Zeynalov

Nakhchivan State University

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 July 2025 in Artificial Intelligence in Applied Sciences
Verified
Journal Article
Open access
PDF file
YÇYiğitcan Çakmak
Yiğitcan Çakmak
JZJavanshir Zeynalov
Javanshir Zeynalov
Breast cancer represents a significant global health challenge, ranking as one of the most prevalent malignancies among women. Early and accurate diagnosis through medical imaging is paramount for improving patient outcomes, with mammography serving as the gold standard for screening. However, the interpretation of mammograms can be challenging and subject to inter-observer variability. This study aims to comparatively evaluate the performance and computational efficiency of four prominent Convolutional Neural Network (CNN) architectures for the automated classification of breast cancer from mammogram images. Utilizing a publicly available dataset comprising 3,383 mammogram images classified as either Benign or Malignant, we trained and evaluated four distinct models: InceptionV3, DenseNet169, InceptionV4, and ResNet50. The results demonstrate that the DenseNet169 architecture achieved superior performance across all evaluated metrics, attaining the highest accuracy (73.33%), precision (70.45%), recall (67.83%), and F1-score (68.60%). Notably, DenseNet169 also exhibited the highest computational efficiency, featuring the lowest parameter count (12.49M) among the tested models. These findings suggest that DenseNet169 offers an optimal balance between diagnostic accuracy and model efficiency, positioning it as a highly promising candidate for integration into clinical decision support systems to aid radiologists in the early detection of breast cancer.

26 June 2025 in Electrica
Verified
Journal Article
Open access
PDF file
FPF. H. Pashayev
F. H. Pashayev
JZJ. I. Zeynalov
J. I. Zeynalov
HNH. T. Najafov
H. T. Najafov
The article proposes a four-level generalized architecture of technological corporate computer networks (TCCNs)and assesses the elements of this architecture and information exchange via communication channels betweentheir levels. The volume of information that is generated at the levels of sensors and technologicalprocesses, whichcarry out most of their activities and functions without human participationand are transmitted to higher levels,is estimated, the functions of traffic monitoring in channels are investigated, and the functions of the distributionof the information volume over time are considered. It is noted that in channels associated with sensors andtechnological processes, the distribution function mainly behaves as a constant function. However, traffic incommunication channels associated with human activity at the upper levels of the network has a certain dailyperiodicity. Both cases can be used when monitoring specific channels. Formulas are given in the form of integraland discrete sum functions for determining the volume of information passing through a channel in a given timeinterval. Here, the average volume of information passing through a selected channel per unit of time and thepeak factor of the information volume are determined. Various methods and tools are used to assess the technicalcondition of the TCCNs channels and their resistance to Internet threats. To address this, the article proposes athree-level backpropagation neural network. Indicators with informative features are created to determine theinputs and outputs of the neural network. Among these indicators, there is an important indicator that is formedbased on the experience of the operating personnel. Five different values are formed from the impressions ofthe operating personnel as indicator estimates. The inputs of the neural network are mainly formed from thesegenerated indicators. The outputs of the neural network are also determined on the basis of the impressions ofthe operating personnel. A network training data set is formed from the inputs and outputs of neural networks.The final result of the article is the learning algorithms based on backpropagation algorithms.Cite this article as: F. H. Pashayev, J. I. Zeynalov and H. T. Najafov, "Monitoring and security assessment of traffic in communication channels of technologicalcorporate computer networks," Electrica, 25, 0216, 2025. doi: 10.5152/electrica.2025.24216.

12 October 2025 in IETE Journal of Research
Verified
Journal Article
Closed access
KKKübra Karadede
Kübra Karadede
ERTUĞRUL ADIGÜZELERTUĞRUL ADIGÜZEL
ERTUĞRUL ADIGÜZEL
JZJavanshir Zeynalov
Javanshir Zeynalov
AEAysel Ersoy
Aysel Ersoy

17 September 2026 in Electrica
Verified
Journal Article
Open access
PDF file
JZJavanshir Zeynalov
Javanshir Zeynalov
VKVitaly Kargin
Vitaly Kargin
BABeyali Ahmedov
Beyali Ahmedov
VAViktor Artemyev
Viktor Artemyev
AHAnar Hajiyev
Anar Hajiyev
GZGultaj Zeynalzade
Gultaj Zeynalzade
With the rapid growth of technological and biological data, the development of intelligent predictive control systems has become a pressing task to enhance the reliability and safety of industrial processes. This article explores streaming data processing models in the context of building a predictive control system for biological risks in Clean-in-Place (CIP) processes. Architectural approaches are examined for integrating streaming analytics with machine learning and predictive modeling tools that enable real-time identification of potential microbiological contamination risks, assessment of their dynamics, and forecasting of critical situations. Special attention is given to the selection and adaptation of streaming technologies such as Apache Kafka, Apache Flink, and Spark Streaming, considering the specific characteristics of biotechnological processes, high data variability, and the necessity of compliance with regulatory requirements in pharmaceutical and food production. It is demonstrated that hybrid models combining statistical methods, machine learning algorithms, and digital twins of CIP processes not only improve prediction accuracy but also allow proactive risk management within automated systems. The article also addresses the integration of predictive control with existing Quality Management Systems and Manufacturing Execution Systems platforms, thereby creating a unified digital monitoring ecosystem. The results of experimental modeling are presented, confirming the effectiveness of streaming models in minimizing the probability of biological risks and optimizing operational costs.

1 July 2026 in Science Progress
Verified
Journal Article
Open access
ERTUĞRUL ADIGÜZELERTUĞRUL ADIGÜZEL
ERTUĞRUL ADIGÜZEL
NSNADIR SUBAŞI
NADIR SUBAŞI
JZJavanshir Zeynalov
Javanshir Zeynalov
MAMaftun Aliyev
Maftun Aliyev
SRSevinj Rzayeva
Sevinj Rzayeva
AEAysel Ersoy
Aysel Ersoy
This study presents a systematic comparative evaluation of ten regression-based machine-learning models for day-ahead photovoltaic (PV) energy forecasting under semi-arid climatic conditions. The analysis is conducted using a limited six month dataset (May–October 2024) of real operational production data obtained from a 22 MW grid-connected PV plant in Nakhchivan, Azerbaijan, integrated with key meteorological variables including solar irradiance, air temperature, relative humidity, and wind speed. Linear Regression, Ridge, Lasso, ElasticNet, Support Vector Regression (SVR), Decision Tree, Random Forest, Gradient Boosting, XGBoost, and a Multi-Layer Perceptron (MLP) were benchmarked using mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), coefficient of determination (R 2 ), and five-fold cross-validated R 2 . The results indicate that regularised linear models—particularly Lasso Regression—provide the most consistent balance between predictive accuracy and generalisation stability under moderate data availability, while Random Forest demonstrates strong cross-validated robustness and the MLP exhibits overfitting behaviour, highlighting sensitivity to limited training data. These findings demonstrate that increased model complexity does not necessarily translate into improved forecasting reliability in data-constrained semi-arid environments. The novelty of this work lies in its unified benchmarking framework that simultaneously evaluates predictive accuracy, interpretability, and generalisation performance using real-world utility-scale operational data. By explicitly linking forecasting reliability with sustainability-oriented planning, the study contributes to more reliable life-cycle cost and emission assessment of large-scale PV systems in emerging renewable-energy markets. The findings offer practical guidance for energy planners and policy-makers seeking transparent, computationally efficient forecasting strategies in semi-arid climates.

30 October 2026 in Electrica
Verified
Journal Article
Closed access
PDF file
JZJavanshir Zeynalov
Javanshir Zeynalov
MAMaftun Aliyev
Maftun Aliyev
LILeyla Ibrahimova
Leyla Ibrahimova
GZGultaj Zeynalzada
Gultaj Zeynalzada
SMSabina Mahmudova
Sabina Mahmudova
TNTofiqa Nadirova
Tofiqa Nadirova
Accurate estimation of photovoltaic (PV) power output is crucial for improving the reliability and efficiency of solar energy systems, especially in regions exposed to high dust loads and temperature variability. This study is scoped as a concurrent power-estimation (soft-sensor) study, inferring instantaneous direct current (DC) power from synchronous environmental measurements rather than multi-step-ahead forecasting; extension to horizon-based forecasting is left for future work. A comparative modeling framework is presented to estimate the power output of a 280 W polycrystalline-silicon PV module in the semi-arid climate of Nakhchivan, using a sixmonth, high-resolution time series of synchronized environmental parameters, irradiance, ambient temperature, humidity, wind, and dust, collected via the Nakhchivan State Meteorological Service and a Benning PV-2 tester. Five supervised learning models were evaluated: multiple linear regression, Ridge, Lasso, Random Forest (RF), and Gradient Boosting, assessed using mean absolute error (MAE), root mean square error (RMSE), symmetric mean absolute percentage error (sMAPE), and Theil’s U1 (the inequality coefficient, bounded from zero to one; no random-walk baseline is claimed). Random Forest showed the lowest error among the models compared, with RMSE of 27.93 W and Theil’s U1 of 0.0896, though this evaluation used a random, non-blocked, train/test split; the magnitude of the RF advantage should be interpreted as an upper bound pending re-evaluation under a leakagefree, blocked time-series protocol. Plane-of-array irradiance, ambient temperature, and dust concentration emerged as the most significant predictors. Ensemble-based models show promising generalization within the sampled December to May window; performance under a leakage-free protocol and generalization to genuine forecasting horizons and to summer remain to be established. These findings can support solar power estimation, fault detection, and performance optimization in distributed energy systems, particularly in dust-prone or datasparse regions. Physics and Astronomy Classification Scheme (PACS) numbers: numbers: 88.40.−j [Solar energy], 88.40.fc [Modelling and analysis].