Neural Network Implementation of Hierarchical Fuzzy Model of Dynamic Objects Speed Control

The aim is to create an intelligent control system based on soft computing for controlling dynamic objects moving along one of the defined routes in real-time systems. The parameters of objects in the real environment are characterized by high nonlinearity, dependence on the state of the environment, and time-varying dynamics when some parameters and states of objects are not available for measurement. Taking this into account, the hierarchical structure of the system is developed based on the classical fuzzy algorithms of Mamdani and Takagi-Sugeno-Kang, and an adaptive fuzzy neural network that implements the model. The application of a neuro-fuzzy model to controlling the movement of dynamic objects with many parameters and incomplete certainty through the use of expert knowledge is considered. A mathematical description of the fuzzy hierarchical model, a learning algorithm, and computer modeling are presented on the example of controlling the speed of rolling cars from a sorting hill. An example of the application of fuzzy rules built on numerical data is considered. The results of modeling with visualization of the results for the synthesized data are presented. The scientific innovation of the obtained results lies in the development of a hierarchical neuro-fuzzy model designed for forecasting and controlling dynamic objects. The modeling results confirm the ability of the proposed model to predict the unknown mapping of the input data vector, which consists of measured and unmeasured parameters, into the desired numerical value at certain points of the path at the model output. The obtained results demonstrate effective prediction of motion dynamics, the ability to achieve high forecasting accuracy and the possibility of intellectualizing the control of the technological process. When approximating the nonlinear dependence, the use of a multilayer neural network ensures the adaptability of the model to a specific area of application, and synergy with a fuzzy algorithm allows automating the process of controlling the technological process at the level of a human operator