Dynamic Performance Identification And Condition Monitoring Of NC Machine Tool Feeding System

The research on dynamic performance identification and condition monitoring of NC machine tool feeding system is concentrated in this dissertation for its great importance to improve performance and reliability of NC machine tool. With the support of the Key Project of Chinese National Programs for Fundamental Research and Development item "Dynamic Performance Variation Regularity Analysis And High Precision Control of High Speed NC Machine Tool" and state lab of digital manufacturing equipment & technology item "Modeling And Analysis Of NC Machine Tool's Dynamic Performance". The dissertation intends to make an intensive study of dynamic performance identification and condition monitoring of NC machine tool feeding system to develop the method to state change process monitoring and analysis. Especially the stiffness and damping identification is focused, such as stiffness and damping of support bearing, as well as those of nut, guide way and so on.To provide the theoretical basis of the research on dynamic performance variation regularity of NC machine, the dissertation explores the internal relationship between the stiffness and damping of support point and the amplitude of ball screw. Meanwhile, an identification method of the stiffness and damping of support point is put forward in this dissertation. The identification models of stiffness and damping are built by analyzing the amplitude changes of ball screw under some load affection employing the knowledge of mechanical vibration and material mechanics. The influence of assembly accuracy on the stiffness and damping is considered while coupling stiffness is ignored.The identification problem is transformed to an optimal problem using the damping ratio and frequency identification method based on Partical Swarm Optimization. Global optimal solution can be obtained by taking the advantage of the Partical Swarm Optimization. The optimal solution is namely modal parameter to be identified.The identification method of dynamic stiffness and coupling stiffness in different directions, that includes x axis, y axis, z axis, rotation direction of y axis and z axis, is studied. The identification model of bearing dynamic stiffness matrix is established by employing substructure synthesis method, modality testing method and finite element method.When the feature change of states is unobvious, the existing monitoring and analysis methods of states are incapable. Aiming at this problem, an intelligent analysis method of machine performance degradation based on vibration signal is presented in this dissertation. A lot of time-domain features, frequency-domain features, wavelet features and fractal features are extracted from the vibration signal, which form original feature sets. The most sensitive features to state change form a new feature vectors by means of feature optimization. The preliminary analysis is carried out through the PNN. The ratio of association cells overlap in CMAC neural networks acts as new features to analyze machine performance degradation.A lot of experimental researches are accomplished and the experimental results show that the above models and methods are effective and practical.