1. Surface water is one of the important sources of water for human life and a major component of water resources. With the continuous advancement of the global industrialization process, under the...

The article presents research on environmental forecasting using machine learning methods for water quality analysis in the Southern Bug River. It focuses on the application of binary logistic regression and random forest regression to estimate the effects of various environmental parameters on key water quality indicators. The study involved data preparation by reading and processing CSV files with economic and environmental indicators and generating feature matrices and target vectors for model training. Both models were trained using nine years of data. The results can be used to develop water resource management strategies, predict the ecological state of water basins, and assess the influence of anthropogenic factors on the hydrochemical regime of rivers. In addition, the revealed relationships between the hydrochemical characteristics of water contribute to increasing the automation of water quality monitoring processes.

This work deals with the issue of operational environmental monitoring of the urban water environment. The specified general provisions regarding the functioning of monitoring methods and the monitoring programme of water bodies. Maximum permissible concentrations of various water quality indicators are set. The main sources of pollution and the main requirements for water quality control systems are defined. On the basis of the above, an industrial parameter control system based on the principles of the Internet of Things technology for remote monitoring and operator control was developed. In the process of performing the work, structural and functional schemes of the operational monitoring system were created and local and remote level work algorithms were compiled. The analysis of measurements of a large number of parameters of the water environment was carried out. The proposed approach proved the possibility of ensuring high accuracy and reliability of measurement results obtained using the developed computer monitoring system.

This article introduces and validates an advanced approach to adaptive automatic control of pyrolysis reactors, aimed at optimizing the utilization of agricultural plastics. By integrating fuzzy logic with enhanced metaheuristic optimization technique, the proposed method achieves superior performance in regulating reactor temperature modes. A key feature of this approach is the incorporation of the reactor's current loading level as an additional variable in the antecedents of the adaptive fuzzy controller's rule base. This dynamic adaptation is further enhanced through the application of the hybrid improved gray wolf optimization algorithm, which optimizes the rule base parameters to increase the robustness and accuracy of control. To assess the effectiveness of this methodology, a fuzzy control system was designed for a specific pyrolysis reactor. Simulation results confirm the approach's superiority over traditional controllers in response speed, accuracy and robustness in various operating conditions. In particular, at a reactor loading level of 100% at startup of the installation and subsequent transition to steady-state operation, the application of the proposed approach made it possible to reduce the transient process duration by 2.26 times and decrease in overshoot by 48.4% compared to existing systems. This, in turn, will lead to a reduction in molecular weight fluctuations and an improvement in the quality of the resulting fuel fractions, thereby significantly enhancing the general energy efficiency and economic performance of agricultural enterprises utilizing pyrolysis technologies.

This paper presents the development and evaluation of a lightweight onboard deep learning module for real-time oil spill detection within a distributed UAV-based maritime environmental monitoring system. The study investigates four object detection architectures under controlled training conditions using both baseline and augmented dataset configurations: three convolutional one-stage detectors (YOLOv8s, YOLO11s, YOLO26s) and one transformer-based model (RF-DETR Nano). Performance was assessed using precision, recall, mAP@50, mAP@50–95, and inference latency on an NVIDIA Tesla T4 GPU. Experimental results show that RF-DETR Nano achieves the highest localization accuracy (mAP@50–95 = 0.863) while maintaining real-time throughput (~ 50 FPS), whereas YOLO26s provides the most favourable efficiency–accuracy trade-off among CNN-based models (mAP@50–95 = 0.806 at ~ 67 FPS). The findings demonstrate that accurate and computationally efficient oil spill detection can be performed directly onboard UAV platforms, enabling reliable autonomous maritime monitoring without dependence on continuous ground-based processing.