Novikov Kirill Sergeevich (Postgraduate student
Penza State University
)
Yurkov Nikolay Kondratievich (Doctor of Technical Sciences
Penza State University
)
Koshelev Nikita Dmitrievich (Postgraduate student
Penza State University
)
|
Background and Objectives. Environmental systems often exhibit complex spatial structures, high variability, and incomplete observations. These features pose significant challenges for predictive and control models. Traditional neural networks show strong data approximation capabilities but lack robustness and interpretability. In this work, we propose a neuro-fuzzy architecture that combines fuzzy logic with neural representations to analyze and manage distributed ecological processes. Materials and methods. The model integrates a fuzzy inference system with learnable neural components and is trained on multichannel monitoring data, including atmospheric, soil, and satellite observations. Control signals are formed using both data-driven inference and rule-based reasoning. Results. The architecture was tested on tasks of pollution forecasting and ecological load regulation. The model demonstrated improved robustness to missing data and enhanced interpretability of predictions. Conclusions. The proposed hybrid system successfully combines expert knowledge with adaptive learning, making it suitable for real-world ecological decision support under uncertainty.
Keywords:neuro-fuzzy systems, spatially distributed processes, environmental modeling, sustainable control, intelligent systems
|
|
|
Read the full article …
|
Citation link: Novikov K. S., Yurkov N. K., Koshelev N. D. NEURO-FUZZY NETWORK ARCHITECTURES FOR ANALYSIS AND CONTROL OF SPATIALLY DISTRIBUTED ENVIRONMENTAL PROCESSES // Современная наука: актуальные проблемы теории и практики. Серия: Естественные и Технические Науки. -2025. -№06/2. -С. 159-164 DOI 10.37882/2223-2966.2025.06-2.26 |
|
|