Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Volumetric Error-Based Condition and Health Monitoring System for Machine-Tools
Ort / Verlag
ProQuest Dissertations & Theses
Erscheinungsjahr
2019
Quelle
ProQuest Dissertations & Theses A&I
Beschreibungen/Notizen
Unexpected or undetected machine tool failures or deterioration results in production and quality losses, hence proactive and prescriptive maintenance using machine tool condition monitoring is sought. This research presents the methodology and solutions developed to monitor the accuracy state of five-axis machine tools by analyzing the machine tool volumetric errors which are defined as the Cartesian error vector of the deviation of the actual tool position compared to its expected position relative to the workpiece frame and projected into the foundation frame. The scale and master ball artefact (SAMBA) method has been used for the measurement of volumetric errors of the experimental five-axis machine tool. The acquired volumetric errors containing machine tool normal and faulty states provide the database for this research. In addition, pseudo-faults and the simulated gradual and sudden faults have also been used. Volumetric error vector features extracted by vector similarity measures are used as the input for the exponential weight moving average control chart where the abnormal change of the single volumetric error vector can be detected. To comprehensively monitor the machine tool accuracy state, a combined vector similarity measure array containing all acquired volumetric errors features has been proposed and processed by the exponential weight moving average control chart. Towards the same faults, the above two data processing can all automatically detect the exact fault occurrence time. Based on the logic of comprehensive monitoring of volumetric errors, fractal analysis of volumetric error coordinates has also been explored. The testing results reveal that it is an effective tool for volumetric errors features representing. To understand the change process of the machine tool state, the acquired historical volumetric errors have been processed by principal component analysis and K-means. For one thing, the proposed methods separate the normal and faulty states of the machine tool (Nearly 100%), for another thing, the designated machine tools provide the references for machine tools state recognition when processing new volumetric errors data. In summary, this research contributed to the development of an efficient solution for machine tool accuracy state monitoring using machine tools volumetric errors based on feature extraction, change recognition and state classification methods. The developed system can recognize the exact change points of real C-axis encoder faults, pseudo-faults EXX and EYX. In addition, it achieves close to 100% accuracy in machine tool faulty and normal state classification.