Research On Fault Diagnosis Method Of Manipulator Transmission System Based On Slow Feature Analysis

Industrial robots have a wide range of applications in the industrial production process.At the same time,only by ensuring the safe,stable and efficient operation of industrial robots can we provide protection for industrial production.As an important system of industrial robots,the transmission system is the part with the highest work intensity in the operation of industrial robots.Once a failure occurs,it will affect the performance of the entire robot.Therefore,monitoring its operating status,discovering the faults of the transmission system of industrial robots,and repairing them in time can greatly improve the reliability of the robot and reduce the losses caused by faults.However,due to the complex working environment of industrial robots and many influencing factors such as noise,its operating data cannot directly reflect the status information of the transmission system.Incipient faults and minor fault signals are often covered by noise signals,resulting in difficulties in fault diagnosis.Therefore,starting from practical engineering problems,this thesis mainly focuses on the following points:(1)The key components and fault mechanism of the transmission system are analyzed,and the commonly used statistics for fault detection are introduced.For the incipient faults and minor faults in the transmission system that are difficult to be found intuitively through data information,combined with the data-driven method,a new fault detection statistic is designed to realize fault detection.(2)Aiming at the problem of incipient fault detection of the transmission system,a datadriven method based on KL divergence slow feature analysis is proposed.This method considers the problem of incipient fault detection of multiple bearings,and combines the characteristics of slow change of incipient faults,and uses the slow feature analysis method to effectively extract information reflecting system state changes from noise.The KL divergence evaluation function of univariate and multivariate forms is given through mathematical theoretical analysis.By analyzing the evaluation function between different states of each bearing,the incipient fault information with small amplitude is extracted,and the incipient fault information with slow change is improved.detection rate.Finally,set the threshold and use the judgment rule to detect the fault.Experiments with numerical cases and bearing operating data verify the effectiveness of the method and prove that the proposed algorithm has accurate fault detection capabilities.(3)Aiming at the small fault detection problem of the transmission system,a fault detection method based on Hellinger distance slow feature analysis is proposed.Considering the fault characteristics of gears under actual working conditions,a mathematical model is established.By calculating the distance of slow features in different states,the hidden probability information in the fault data set is fully utilized.This method has high sensitivity and can effectively detect the small faults of the gears in the transmission system.Finally,the feasibility of the method is verified by the gear operation data.