Acoustic Diagnostics of Lever Mechanisms with Subsequent Processing of Data on Neural Networks

Abstract

The technique of acoustic diagnostics for machine tools - robots is developed. A neural network reference model has been constructed that allows to diagnose the current characteristics of the state of objects under different conditions, namely, the configuration of the mechanism, the geometric parameters of the mechanism with the motor-spindle running, the dynamics of the movement of the nodes of the experimental stand mechanism with variable speed and load on the drive, and the temperature of the object. Experiments have been carried out to investigate the relationship between the parameters of the spectrum of an acoustic signal with a given discreteness, excited by a perturbing effect in the form of “white noise.” The possibility of using the proposed approach to the management of complex technological machines, such as machines with mechanisms based on parallel kinematics, is shown to improve the accuracy of the positioning of the actuators, to ensure their dynamic tuning and to optimize the trajectories of the movements of the working organs of the equipment.