An improved density-based spatial clustering of applications with noise (IDBSCAN) analysis approach based on kurtosis and sample entropy (SE) is presented for the identification of operational state in order to provide accurate monitoring of spindle operation condition. This is because of the low strength of the shock signal created by bearing of precision spindle of misalignment or imbalanced load, and the difficulties in extracting shock features. Wavelet noise reduction begins by dividing the recorded vibration data into equal lengths. Features like kurtosis and entropy in the frequency domain are used to generate feature vectors that indicate the bearing operation state. IDBSCAN cluster analysis is then utilized to establish the ideal neighborhood radius (Eps) and the minimum number of objects contained within the neighborhood radius (MinPts) of the vector set, which are combined to identify the bearing operating condition features. Finally, utilizing data from the University of Cincinnati, the approach was validated and assessed, attaining a condition detection accuracy of 99.2%. As a follow-up, the spindle’s vibration characteristics were studied utilizing an unbalanced bearing’s load bench. Bearing state recognition accuracy was 98.4%, 98.4%, and 96.7%, respectively, under mild, medium, and overload circumstances, according to the results of the experimental investigation. Moreover, it shows that conditions of bearings under various unbalanced loads can be precisely monitored using the proposed method without picking up on specific sorts of failures.

Skiving is an efficient gear cutting technology with relatively high machining accuracy, especially for internal gears. Nevertheless, the machining accuracy can hardly be satisfied if the skived gear is modified, so we try to extend the definition and hypotheses domain of the previous works. In this paper, fundamental research is conducted to further improve the skiving accuracy. Firstly, a new type of skiving tool with double rake faces is proposed for flexibly meeting different modification requirements of the two flanks of the work gear teeth. Secondly, the mathematical model of curve-surface conjugated cutting edges of the skiving tool for enveloping the tooth flanks of the working gear with profile modification is established. Then, the algorithm of the skiving tool path with alterable shaft angle in the cutting process is proposed for eliminating the twist of gear tooth flanks with lead modification. Finally, machining simulations are carried out to verify the feasibility of the proposed improved skiving methods.

In this paper, a quasi-static angular contact ball bearing model, considering assembly accuracy is constructed, while a numerical solution method for bearing stiffness under bad assembly state is established. A 7014C angular contact ball bearing is used as the research object and five groups of different spacer inclinations are designed to imitate the installation error of the spindle bearing. The bearing stiffness performance was comparatively analyzed, according to the five spacers. The effect of preload and rotation speed on bearing stiffness are systematically investigated, considering different parallelism errors, as induced by the spacers. The influence mechanism of the badly assembled bearing on the respective stiffness anisotropy is studied based on the proposed model. The results show that the variations of the inclination between the inner and outer rings of the bearing exhibit a very weak effect on the axial stiffness, while the influence on the radial and angular stiffness is more significant.

In machining operations, the misalignment of the bearing assembly or imbalanced load often leads to deflection and failure of the tool spindle. The use of single feature information does not accurately monitor the complex working conditions. Considering this, this paper proposes a rolling bearing running condition monitoring method which is based on multiple feature information. Firstly, a multi-dimensional feature matrix is obtained by extracting the features of a single type of raw data in the time domain, frequency domain, and time-frequency domain, and then the dimensionality of the matrix is reduced by principal component analysis (PCA). An entropy weight improved the D-S(EWID-S) evidence theory is proposed. By updating the initial evidence source, and applying the Euclidean distance of the spatial centroid, the fusion results were evaluated. Finally, a test rig for eccentric bearing load operation is developed to obtain the vibration signals at two distinct locations and to confirm the proposed method. The test results show that the condition monitoring method based on the PCA and EWID-S evidence theory can effectively identify the bearing operating at different degrees of deflection. At the same time, by comparing with other improved D-S evidence theory methods, it is verified that this method has more advantages in information fusion and bearing condition monitoring.

In order to solve the problem that it is difficult for a single sensor to accurately characterize the running state of rotating bearings under complex working conditions, this paper proposes a data-level fusion method based on multi-source isomorphic sensors to monitor spindle bearings. First, new vibration signals in the X,Y,Z direction were obtained through the process of decomposing, de-noise, and reconstructing. Second, the PCA algorithm was used to select the time-domain and frequency-domain features of the vibration signals, construct the feature matrix, and perform dimensionality reduction in the feature matrix. Finally, the entropy weight method was introduced to obtain the initial weights of the three directions as the inputs of the adaptive function. The chaotic particle swarm optimization algorithm proposed in this paper helps particles jump out of the local optimum. Chaotic mapping is used to initialize the velocity and position of the particles, which calculates globally optimal weights in three directions. In order to extract bearing signal features more accurately and efficiently, a DenseNet and Transformer (DAT) feature extraction model is proposed to deal with the complex changes and noise interference of bearing signals. Through the open data set of Jiangnan University and the data collected by our own experimental platform, the maximum accuracy of the DAT model was verified to be 100%.

Misalignment or unbalanced loading of machine tool spindle bearings often results in skewed bearing operation, which makes the spindle more susceptible to failure. In addition, due to the weak impact signal of the bearing in skewed operation, a single feature information cannot accurately characterize the operation status of the bearing. To address the above problems, this paper proposes a method to monitor the uneven running state of bearing load based on a dual-channel fusion improved dense connection (DenseNet) network. First, the original signal is pre-processed by overlapping sampling method, and the dual-channel experimental data are obtained by frequency-domain and time-frequency-domain algorithms; then the processed data are input into the improved 1D-DenseNet and 2D-DenseNet models respectively for feature extraction; then the frequency-domain and time-frequency-domain features are fused by concat splicing operation, and the output belongs to each category The probability distribution is used to characterize the operating state of the bearings. Finally, the validity of the algorithm model is verified by using the Case Western Reserve University public rolling bearing data set, and an experimental bench is designed and built for experimental verification of the uneven bearing load operation. The comparative analysis of the experimental results in this paper shows that the algorithm can extract the features of the input signal more comprehensively and finally achieve 100% recognition accuracy.