Research On The Online Intelligent Detection And Suppression Of Cutting Chatter

Aviation manufacturing is one of the most important parts in manufacturing industry.Machining of aeronautical thin-walled parts have always been key factors which constrain the development of the aviation industry.It is very prone to cause chatter in the processing of thin-walled parts,thereby affecting production efficiency and machining accuracy.There is no doubt that chatter is harmful to work-piece,cutting tool and machine tools.Thus,it is very critical to develop chatter detection and suppression method for aviation parts manufacturing.This paper studies and designs a closed-loop machining system which integrates online intelligent chatter detection and suppression.By detecting the cutting force signals,cutting state is real-time monitored and chatter will be online suppressed,achieving the goal of high quality processing of thin-walled parts.First,conditions and characteristics of cutting chatter is studied theoretically and experimentally.It is showing that chatter usually occurs in the weak point of machine tools,with increasing energy and amplitude vibration of its nature frequencies.According to the chatter characteristics,wavelet packet node energy(WPNE)is presented,using as feature extraction method.The effect of decomposition level of WPNE is studied experimentally.It is found that the accuracy of the detecting systems is enhanced by using the high-level WPNE.According to the characteristics of LSSVM classifier,a feature selection method,LSSVM-RFE is presented to eliminate redundant or irrelevant features in the high-level WPNE and reduce execution time.It is shown that,by LSSVM-RFE,the accuracy of chatter detection is enhanced from 97.41% to 98.90%.In addition,the execution time is reduced from 46.3 ms to 9.5 ms.Second,LS-OC-SVM,the least squares form of one class support vector machine is studied and used to chatter detection.An online evolution detection model is proposed based on LS-OC-SVM.In which,partitioned matrix inversion is used to implement online incremental solution of LS-OC-SVM.The sparse solution of LS-OC-SVM is presented based on dataset dictionary which is constructed by coherence criterion.Thus,the dataset information is stored in the dictionary.The detection model is online evolved by online updating the dataset dictionary.The experimental results show that the online evolution model performs better than offline model in the cutting chatter detection.Finally,through stability analysis of cutting processes,the mechanism of chatter suppression by spindle speed variation(SSV)cutting is studied.By using synchronized actions,which is one of 840 D advanced programming,spindle speed variation cutting is implemented in machine tools.Constant speed cutting and SSV cutting is conducted respectively.The result comparison shows that SSV cutting can suppress chatter efficiently.By modifying the machine tools,an intelligent closed-loop machining system is designed which combined the online intelligent chatter detection and suppression function.In the machining process,cutting force is real-time detected in order to monitor the cutting states.When chatter is gestated,SSV cutting is enabled to suppress chatter.Thin-walled parts turning experiment shows that the intelligent machining system proposed in this paper can conduct non-chatter and high-quality processing efficiently.