Harmonic drive is susceptible to strength failure and performance degradation failure during operation. Given the long life test cycles, limited sample size, and incomplete understanding of degradation laws, this study conducted comprehensive life test and performance degradation test research to enable future failure prediction and reliability assessment for harmonic drive. Building upon established test rigs for a HD life test and performance degradation test, a step-down accelerated life test methodology was developed. Life tests under step-down accelerated conditions were executed, with a concurrent performance degradation test throughout the life test. Key datasets acquired include vibration signal histories, degradation data for critical performance indicators such as stiffness, precision, transmission efficiency, and backlash. Test results show that the predominant strength failure in the atmospheric environment is flexspline fatigue fracture, while significant degradation occurred in stiffness, precision, and backlash across all test conditions; transmission efficiency showed gradual degradation before strength failure followed by a marked decline post-failure. Finally, the failure mechanism of strength and performance degradation is analyzed, and the mechanism consistency of the two failures is verified by Hilbert envelope spectrum analysis and degradation trajectory shape consistency, respectively. The results from this paper provide critical data support and a reference foundation for the proactive maintenance of the harmonic drive.

In this article, the residual vibration signal of the worktable of the machine tool is used to identify its dynamic parameters. By analyzing the time-domain characteristics of the residual vibration signal, it is found that the signal has the similar attenuation characteristics to the impulse response of the second-order underdamped system, so a new method for identifying the dynamic parameters of the worktable is proposed. In this method, the impulse response model of the second-order underdamped system is used to describe the residual vibration signal, so the parameter identification problem based on the time-domain signal is transformed into the optimization problem of the model parameters. Then, the dynamic parameters can be determined by using the genetic algorithm. The results of the genetic algorithm are compared with the results of the autoregressive moving average method, which is a commonly used parameter identification method based on the time-domain signal. It was found that the results of the model-based method proposed in this article are more accurate than the autoregressive moving average method.

In order to suppress the spindle vibration with high efficiency and high precision, a no without trial weight spray online balance method is proposed in this paper. By analyzing the relationship between the unbalanced excitation and the unbalanced response of the spindle, the relationship between the dynamic influence coefficient and the system model is studied. A high‐speed spindle finite element analysis model was established, and the dynamic influence coefficient matrix was identified. A no trial weight spray online dynamic balancing system was developed, which has the advantages of without trial weight and high‐precision loading. A new type of integrated balancing terminal that was formed using 3D printing technology was first proposed by our research group, and its advantages in various aspects are significantly higher than traditional assembly balanced terminals. The experimental verification of the without trial weight spray online dynamic balancing system was performed on a high‐speed spindle test stand. Experiments show that the no trial weight spray online balancing method proposed in this paper can achieve high‐efficiency and high‐precision vibration suppression, greatly reducing balance time and cost of the spindle. At the same time, the online balance test also verified the reliability of the integrated balanced terminal.

To improve the machining accuracy and production efficiency of precision components with deep hole structures, an online prediction method of the inner hole roundness error, which cannot be directly measured in real time during the machining process, is proposed in this paper. For online prediction of the workpiece roundness error (WRE) during machining, a predictive model based on correlation analysis and a proportional method is proposed according to the spindle synchronous error motion (SSEM) by three‐probe method testing. To improve the prediction accuracy of the WRE, a particle swarm optimization (PSO) algorithm is introduced for optimizing a probe mounting angle of a three‐probe method, and a harmonic wavelet method for SSEM feature extraction is proposed. Using the PSO algorithm, the optimal probe mounting angle of the three‐probe method is obtained, the influence of spindle surface roundness on SSEM is eliminated, and the higher‐order harmonic suppression of the three‐probe method is avoided effectively. By the harmonic wavelet method, the accurate SSEM extraction is enhanced and the WRE prediction accuracy is promoted. The experiments show that the inner hole roundness error online prediction method proposed in this paper has high prediction accuracy.