Neuro-fuzzy adaptive control of the power plant of an electric vehicle

Abstract

To date, a number of features of the IVCS for power plants of electric and hybrid vehicles remain insufficiently studied. The applied methods of analysis and synthesis of IVCS do not pay sufficient attention to the multi-criteria nature of the emerging optimization problems. Methods of adapting control to variable external operating conditions are not effective enough. These circumstances do not allow to fully realize the potential of IVCS for power plants of electric and hybrid vehicles. Goal. Substantiation and implementation of a comprehensive methodology for building a highly efficient system of control over technological operations and measurement of information in various types of power units of modern electric vehicles. The developed system allows for the operational synthesis of the optimal control effect and the formation of control influences in real time in accordance with the specified energy and quality efficiency criteria, with mandatory consideration of the dynamic change in external operating conditions and environmental parameters. Methodology. The methodological basis of this scientific work is a rational and balanced combination of fundamental theoretical provisions and applied experimental research. The work uses a comprehensive systemic approach to the design of an information and measurement control system, which is invariant to the specifics of the design of various power units of electric vehicles. This approach provides the opportunity to quickly and flexibly solve complex tasks of coordination and control of operating modes according to a set of quality, energy and other operational criteria. Results. During the study, a mathematically based optimal control was obtained, which can be directly used in the development of clear logical rules for choosing a strategy for adaptive vehicle control. In addition, the results provide a reliable scientific justification of the key parameters, operating characteristics and functional relationships of modern systems and individual units of an electric vehicle. Scientific novelty. An innovative concept of mathematical modeling and multi-criteria optimization of analytical models of complex physical processes in power plants, which are critically difficult to formalize by classical methods, was formulated and proposed by presenting and approximating them in the form of artificial neural networks. Practical significance. Further development and practical implementation of the results of this study has broad potential for significant improvement of adaptive control systems for passenger and freight electric trains. The practical use of neural network adaptive criticism methods enables effective overcoming of the chronic lack of a priori information about the key parameters of the real driving cycle and changing external operating conditions, as well as compensating for the low accuracy of traditional deterministic mathematical models.