Research On Fault Diagnosis Technology Of Disc-type Tool Changer Based On CS-SVM

As a key functional component of the machining center,the disc-type tool changer have features of complex structure,fast tool changing and high requirement for position accuracy and so on,which make it a functional component with high probability of fault in the machining center,and the high failure rate directly affects the reliability of the machining center.Maintenance facility personnel can't find out the failure cause and rapidly repair the disc-type tool changer due to its multiple failure modes and complicated failure causes.With many experiments and further study of failure mode and failure mechanism,research on fault diagnosis technology of disc-type tool changer developed by lab was proposed in the paper,which aimed at making a decision of the fault location and the fault cause rapidly and accurately,and the research can shorten the maintenance time greatly which improves the utilization of the machining center.This paper took a domestic disc-type tool changer as the research subject,and the main work and innovations of this paper are as follows:(1)The sensitive signal which represented the operating condition of disc-type tool changer was determined.The common fault modes were analyzed clearly by the reliability test data of the machining center.After the fault mechanism were analyzed,the sensitive signal(vibration signal)which represented the operating condition that the disc-type tool changer was normal or failed was determined.The vibration signal could be collected with the use of condition monitoring of the disc-type tool changer.(2)A fault classifier of disc-type tool changer based on CS-SVM was proposed in this paper.First of all,the wavelet transform as a method of feature extraction was used to extract feature values of characteristic signals,and the extracted feature was normalized to be a new initial feature set which could reflect the corresponding fault of the disc-type tool changer.And then,the cuckoo search was used to optimize the support vector machine.Finally,a fault classifier of disc-type tool changer based on CS-SVM was developed.(3)A dimension reduction method was proposed in this paper in the case of guaranteeing the accuracy of the given classifier.Firstly,after the vibration feature signals were decomposed into multiple wavelet coefficients by five layers' wavelet packet transform,energies of these wavelet coefficients were normalized as a feature set.Then,the importance evaluation of the characteristics of the initial feature set was carried out by taking account of the subsequent classifier.Finally,using the forward sequential search algorithm,the classifier was trained and tested,and the accuracy of the classifier was evaluated,and then the feature subset which could make the classifier achieve optimal performance and contain the minimum number of fault features was the optimal fault feature subset according to the evaluation results.So during the subsequent diagnosis,using the optimal feature subset corresponding features directly on the classifier to train and test,could not only make the classifier achieve the best performance,but reduce computational complexity significantly.(4)Experiments and analysis of the result.Three kinds of stage of the disc-type tool changer were classified successfully with high accuracy by the use of the fault classifier of disc-type tool changer proposed in this paper,which proved the accuracy and feasibility of the methods proposed in the paper.