A large model-based data center operation and maintenance robot control method

Abstract

The application discloses a kind of data center operation and maintenance robot control method and system based on large model, it is related to intelligent robot control technical field, including, using extended Kalman filter generates current state estimation vector, and generates prediction state by discrete Kalman filter, combine joint filtering mechanism to fuse current state and prediction state, obtain the global state estimation vector of robot, input Gaussian mixture model to carry out abnormal probability prediction with global state estimation vector, according to the generation of the predicted result decision robot target state, based on target state and global state estimation vector optimization KFOSMC parameter, using the parameter of optimization generation robot control signal.The joint filtering mechanism fuses current state estimation and prediction state, ensures that robot is efficiently and stably operated in complex data environment, abnormal probability prediction is carried out by Gaussian mixture model, accurately predict the possibility of failure occurrence in variable environment, avoid the defect of traditional method relying on threshold judgment.

Images