Operation monitoring system and monitoring method for online trajectory prediction of industrial robot

Abstract

The invention discloses an operation monitoring system and a monitoring method for online trajectory prediction of an industrial robot. The invention establishes a long-term and short-term memory network model for the historical pose data of the industrial robot, and focuses on the online trajectory prediction and operation status monitoring of the industrial robot. The established prediction model can effectively solve the problem of nonlinearity and dynamics of the running pose data of the industrial robot, so that the results of the running state monitoring are more accurate and reliable; the present invention takes the position and orientation coordinates of the industrial robot in the future as the prediction output of the model , to realize long-term and short-term memory network modeling and nonlinear dynamic trajectory prediction, which is used to monitor the real-time operation of industrial robots.

Description

technical field [0001] The invention relates to the technical field of industrial robot control, in particular to an operation monitoring system for online trajectory prediction of an industrial robot based on a long short-term memory network and a monitoring method thereof. Background technique [0002] With the continuous development of production automation and intelligent technology, the proportion of industrial robots used in the manufacturing industry is increasing. In the application process of industrial robots, how to ensure the reliability and safety of their operation has become a research hotspot in related fields; at present, the operation status monitoring of industrial robots is mainly based on data-driven fault detection and fault diagnosis. For example, Chen Shijian of South China University of Technology proposed an expert system for fault diagnosis of industrial robots based on fuzzy theory; however, such methods mainly consider the monitoring of the runni...

AI Technical Summary

Problems solved by technology

In the application process of industrial robots, how to ensure the reliability and safety of their operation has become a research hotspot in related fields; at present, the monitoring of the operating status of industrial robots is mainly based on data-driven fault detection and fault diagnosis For example, Chen Shijian of South China University of Technology proposed an expert system for fault diagnosis of industrial robots based on fuzzy theory; however, such methods mainly consider the monitoring of the operating status of industrial robots themselves, and pay less attention to the real-time status and monitoring of operating trajectories.

[0003] In order to incorporate the online running trajectory into the scope of monitoring the running status of industrial robots, the position and orientation coordinates of the future time can be predicted as the output indicators of the running status of industrial robots, and the reliability of running status monitoring of industrial robots can be improved through online regression prediction; among them, based on data-driven The online regression prediction method is currently the most commonly used partial least squares method (Partial Least Squares, PLS) and its improved method; this type of method uses offline historical data to establish a partial least squares model between general operating variables and quality variables for online operation However, such methods are often based on ideal assumptions such as linear and non-dynamic operating data, or only solve the impact of one of the characteristics on modeling, and it is difficult to solve the problem of nonlinear dynamic data of industrial robots. Modeling and Monitoring Issues

Images