Abstract
The study objective was to construct models of multimass electromechanical systems using neural nets, fuzzy inference systems and hybrid networks by means of MATLAB tools. A model of a system in a form of a neural net or a neuro-fuzzy inference system was constructed on the basis of known input signals and signals measured at the system output. Methods of the theory of artificial neural nets and methods of the fuzzy modeling technology were used in the study.A neural net for solving the problem of identification of the electromechanical systems with complex kinematic connections was synthesized using the Neural Network Toolbox application package of the MATLAB system. A possibility of solving the identification problem using an approximating fuzzy system using the Fuzzy Logic Toolbox package was considered. A hybrid network was synthesized and implemented in a form of an adaptive neuro-fuzzy inference system using the ANFIS editor. Recommendations for choosing parameters that have the most significant effect on identification accuracy when applying the methods under consideration were given. It was shown that the use of neural nets and adaptive neuro-fuzzy inference systems makes it possible to identify systems with accuracy of 2 to 4%.As a result of the conducted studies, efficiency of application of neural nets, fuzzy inference systems and hybrid nets to identification of systems with complex kinematic connections in the presence of input-output information was shown. The neural-network, fuzzy and neuro-fuzzy models of two-mass electromechanical systems were synthesized with the use of modern software tools.The considered approach to using artificial intelligence technologies, that is neural nets and fuzzy logic is a promising line of construction of appropriate neural-network and neuro-fuzzy models of technical objects and systems. The study results can be used in synthesis of regulators for the systems with complex kinematic connections to ensure their high performance.
Highlights
When modern control systems are synthesized, the problem of identification persists to be extremely important
The problem of parametric identification of an asynchronous squirrel-cage induction motor by the transient curve of the direct start process at a specified load is considered. Application of this method to identification of multimass electromechanical systems will lead to significant errors because, as it was noted, the state equations can be written only for an equivalent model with concentrated loads which does not take into consideration all the factors influencing dynamic processes in multimass systems
The study objective is to construct models of multimass electromechanical systems using neural nets, fuzzy inference systems and hybrid networks based on measured input and output signals with the help of MATLAB tools
Summary
When modern control systems are synthesized, the problem of identification persists to be extremely important. The lack of complete information on operating conditions, properties and parameters of objects and systems necessitates the use of an adaptive approach to control which tolerates the use of simplified, in particular, linear models This approach makes it possible, in some cases, to significantly reduce a priori uncertainty and implement rather effective control, constriction with linear models does not always ensure obtaining of desired results. In addition to neural nets, methods based on the theory of fuzzy sets and fuzzy logic, namely the technologies of fuzzy modeling have found wide application in solving problems of identification These methods are effective when information about the subject under study is incomplete or inaccurate. Current studies on application of the neural-network technologies and the fuzzy modeling technology to improve accuracy of identification of electromechanical systems with complex kinematic connections are urgent in the absence of complete information about their structure and parameters
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Eastern-European Journal of Enterprise Technologies
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
