Research On The Fault Diagnosis Method Of Ball Screw Sub Based On SSA-VMD

Ball screw sub is the most accurate,reasonable and economical mechanical transmission device to realize the conversion of rotary motion and linear motion in the feeding system,which is widely used in the manufacturing industry such as aerospace,CNC machine tools and automobile industry.The vibration,impact,overload and other factors in the manufacturing process will lead to a series of failures such as wear,pitting,deformation and so on,which will directly affect the processing accuracy and processing quality of manufacturing.Therefore,theoretical and experimental research on the performance state identification of ball screw subsets is of great practical significance to ensure the quality of manufacturing production,improve the efficiency of processing and production,and reduce the economic loss of enterprises.The key to the study of ball screw sub performance state identification is:The first step is to understand the various types of failure forms of ball screw subs and the causes of the failures.Determine the main forms of failure to be studied in conjunction with the production reality.The second is to carry out the ball screw sub fault information acquisition experiment.The experimental design and the type of fault signal to be acquired are determined for the studied failure form of the ball screw sub.Experimental acquisition of signal data for different performance states of the ball screw sub.Then the study of signal analysis and processing methods is carried out.The data obtained from the experiments are analyzed and processed by the adopted signal processing methods.Extracting the characteristic parameters corresponding to each failure form to construct a fault feature set.Using the feature set to carry out training and testing of the condition recognition model.Finally,the validityj of the ball screw sub fault state recognition model is verified.In this paper,a fault state recognition method based on SSA-VMD for ball screw subsets is proposed from the actual production requirements,and the details of the research are as follows:(1)Analysis of ball screw subsystem structure and main failure forms.The main structural forms of the ball screw sub are introduced,the parameters of the ball screw sub model studied in this paper are given,the possible forms of failure,causes of failure and performance of failure of each component part of the ball screw sub are analyzed,and the performance states of the ball screw sub studied in this paper are determined as screw bending failure,screw raceway pitting failure and normal state.(2)Signal acquisition experimental platform design and construction and signal acquisition.The signal acquisition requirements for ball screw sub fault diagnosis are analyzed,and the type of signal acquired is determined to be vibration signal.The ball screw sub is arranged in the working form of reciprocating motion of the table connected with the screw nut driven by the rotation of the screw by a servo motor.The experimental platform for vibration signal acquisition of the ball screw sub was designed according to the experimental requirements,and the selection of the main material and each component of the drive control system and signal acquisition system was confirmed.Then the pitting failure and bending failure of the screw were produced by using artificial destruction,and the vibration signal acquisition experiments of each performance state of the ball screw sub were designed and conducted,and finally the Finally,40 sets of vibration signals were obtained for each performance state.(3)SSA-VMD based ball screw sub fault feature extraction method.After the pre-processing operation of the original signal,the vibration signal of the stable running part of the original signal is processed by using the variable modal decomposition algorithm.The sparrow optimization algorithm is introduced for the parameter selection problem of the variational modal decomposition algorithm to perform adaptive optimization of the parameters of the variational modal decomposition algorithm.The algorithm principle and algorithm flow of the variational modal decomposition algorithm and the sparrow optimization algorithm are introduced.The decomposed signal is further filtered and reconstructed to extract the time and frequency domain feature parameters in the signal as the feature vector set for state identification.(4)SVM-based ball screw sub fault state identification.3/4 of the feature dataset is selected for training the SVM for ball screw vice condition recognition model,and the remaining 1/4 of the dataset is used for verification of the recognition model accuracy.The EMD and VMD are used to extract the feature dataset and build the fault recognition model for training and testing,and then compared with the method in this paper for verification.