DYNAMIC SYSTEMS MODELING USING ARTIFICIAL NEURAL NETWORKS FOR AGRICULTURAL MACHINES
МОДЕЛИРОВАНИЕ ДИНАМИЧЕСКИХ СИСТЕМ С ИСПОЛЬЗОВАНИЕМ ИСКУССТВЕННОЙ НЕЙРОННОЙ СЕТИ ДЛЯ СЕЛЬСКОХОЗЯЙСТВЕННЫХ МАШИН
DOI : https://doi.org/10.35633/inmateh-58-07
Authors
(*) Corresponding authors:
Abstract
The tasks of designing complex dynamic systems (an agricultural machine, a car, a metalworking machine) are always multi-criteria, since choosing a reliable option needs taking into account many various requirements for the technical systems. The methodologies and methods of dynamic systems research are currently improving due to the need in developing a functional component that takes into account the unlimited possibilities of computers, in some cases changing over to “virtual reality”. The purpose of the study is to reveal the essence of a promising heuristic approach to the assessment of functional relationships between the functioning elements of dynamic systems and variables describing the state of a given system. Technological production processes can be considered as a dynamic system containing resistance forces. In dynamic systems (machines), a transitory phenomenon occurs when starting and stopping, when switching from one mode to another, as well as when resetting or increasing the working load. In many cases, when studying transitory phenomena in dynamic systems, it is convenient to use not the classical method of integrating differential motion equations, but an operational calculus based on a promising area of applied mathematics – artificial neural networks, and one of the promising methods for the development and design of various dynamic systems is simulation by artificial neural networks. The technological process model for onion harvesting machines presented by artificial neural networks is able to assess the qualitative indicators, separate functioning elements of the cleaning machine performance out of input factors with different physical nature, while further research is based on previous model constructions. The methodology for modeling working processes of dynamic systems by using artificial neural networks in the form of reality objects significantly expands the opportunities for arrangement and reuse of the results obtained, makes it possible to use the analytical apparatus of the information theory (message transmission in the presence of interference) for searching and optimizing the design and operating parameters of the machines under development.
Abstract in Ukrainian