Theory Of Scale Fractal Decomposition And Applications In Machine&Tool Health Monitoring

Taking the implementation of the "Made in China 2025" plan as a symbol,China has entered a crucial stage of the structural adjustment of the manufacturing industry and the upgrading of production capacity.Intelligent manufacturing technology is a key pillar of the digital workshop/smart factory.In this paper,relying on the intelligent manufacturing special project of the Ministry of Industry and Information Technology and the National Natural Science Foundation of China,the key technologies for condition monitoring and health assessment of machine tools in the machining process are studied,and the digitized basis for tool life prediction and maintenance decision-making is provided.This paper first reviews the research progress of machine tool and tool health monitoring,and summarizes several key technologies.The classical processing method of dynamic signals is summarized,which lays a solid theoretical foundation for the development of the subject.Aiming at the problem of identification of non-stationary anomalous components in dynamic data processing of machine tools,a new multiscale theory of fractal fractal decomposition was proposed.Based on the approximate analytical discrete wavelet,the construction method of the hidden scale wavelet packet intermediate scale is given,and the approximate translation invariance and perfect reconstruction characteristics are maintained without increasing the computational complexity.The center frequency is kept fixed in the process of layer-by-layer refinement of the resolution,and the center analysis frequency is also at the edge of the classical wavelet theory,which greatly improves the feature extraction effect.The concept of "central nested subspace clusters" is proposed to reveal the fractal-set characteristics of the"time-scale" grid of this new type of multiresolution analysis,and it is proved mathematically that the fractal fractal decomposition is a more general multiresolution analysis theory.Based on the proposed fractal fractal decomposition theory,a transient transient feature extraction method is proposed.The characteristics of strong shock cycle components were used to measure the impact characteristics of dynamic processes.The feature extraction method was applied to the dynamic and static collision tests of the rotor,and the strong impact component was extracted.A single mode component with more physical meaning was found near the optimal analysis scale.In the project,the proposed method is applied to the analysis of spindle box torque signal of a three-axis semi-numerical machining center.According to the optimal scale and the fault characteristics of its sub-subspace,the gear failure on a drive shaft of the spindle box is identified.This further validates the practicality of the method.Aiming at the common fragmentation and poor synchronization of a large number of monitoring data in the digital manufacturing process,a new method of process isomorphism monitoring in manufacturing process was proposed.Combining LabVIEW and SQL Server database platform,the upper and lower position monitoring system of the monitoring system was developed.A large number of data were completely collected during the machining process of the machine tool,and then the data features were referenced to the process steps for intelligent data segmentation and storage.The classical stationary process analysis method and the new fractal fractal decomposition algorithm are used to intelligently process the stored isomorphic data.From the sparsity of characteristic components,the wear failure process of the tool is revealed.Acoustic emission signals of several typical states broken by collecting tools in the laboratory verify the effectiveness of the intelligent algorithm.In the machining center,45#steel parts were used for multi-stage tool life test.Collect the microscopic images of tool vibration acceleration signals and tool wear.The analysis results show that the evolution of the sparse feature of the vibration acceleration signal and the actual wear process of the tool have a good corresponding relationship,indicating the effectiveness of the proposed monitoring system.