Neural network technologies in control systems of cargo movement mechanisms

Abstract

THE PURPOSE. Development and analysis of control systems for cargo movement mechanisms that do not contain a speed sensor in their structure. The use of intelligent devices in the implementation of sensorless control systems. The study of the proposed solutions in closed-type systems in order to identify the most optimal option that provides the best performance according to the criteria, in this case, the accuracy of speed testing.METHODS. It is possible to achieve these goals through the use of mathematical modeling carried out in the Matlab Simulink simulation environment.RESULTS. In the study, the analysis of systems containing various kinds of velocity observers in their structure was carried out. The stability of the work of the observers under consideration was evaluated taking into account external disturbing influences – the inter-turn closure mode was considered.CONCLUSION. The use of control systems that do not have sensors in their structure is in demand on mechanisms installed in rooms with a small area, on objects with elevated ambient temperatures and with increased pollution. The study compared systems with a speed sensor, a system containing a non-adaptive observer and systems with neural network observers. Optimal indicators were obtained in a system containing a NARMA-L2 neurocontroller. A combined structure is proposed containing several neuroregulators that are trained for dynamic engine parameters and monitored dangerous modes that may occur in dynamics.