Zhihua Liu

The multiparameter vibration sensors, which are composed of a multiaxial accelerometer and gyroscope, are widely used in pose estimation, vibration monitoring, and other engineering fields because of their ability to simultaneously measure the basic linear and angular motions. To ensure the performance of these engineering, it is necessary to test the sensitivities of these sensors regularly. However, the current dynamic testing methods for the sensitivities inevitably have problems such as high cost, low efficiency, and poor traceability, which are difficult to meet the practical requirements. In this study, a new 6-DOF motion platform-based dynamic testing method is investigated so as to improve the economy, efficiency, and traceability. This method converts the input spatial motion excitation of the sensors to 2-D planar motions of three laser spots by a physical dimensionality reduction way. Whereafter, the monocular vision method with subpixel accuracy is adopted to locate and track these spots, and the spot subpixel coordinates are converted into excitation displacements by the decoupling model to achieve the excitation measurement. Finally, the sensitivities can be calculated by using the relation between their outputs and the measured input excitation. The comparison experiments with the shaker- and turntable-based testing methods have been conducted, and their corresponding maximum relative deviations are about 1.5%, which confirmed that the investigated method can effectively and accurately determine the sensitivities. Additionally, the main error sources of the investigated method are analyzed to calculate the corresponding uncertainties.

The six-degree-of-freedom parallel robots is crucial for intelligent manufacturing, motion simulation, aerospace and other fields. Their trajectory performance level directly affects the reliable application of high-precision operation scenarios. However, dynamic trajectory errors under motion conditions remain a challenge. To address this, to improve the motion trajectory accuracy of parallel robots, a CNN-GRU model-based trajectory error prediction and compensation method is proposed. The novelty of this method lies in the hybrid deep learning architecture that combines CNN for spatial feature extraction and GRU for temporal dependency modeling. This method accurately predicts the trajectory error of parallel robots by constructing a deep learning model that integrates CNN and GRU, and compensates for the amplitude and bias of the trajectory error at the control command end, thereby improving the trajectory accuracy of parallel robots. The simulation and the 6-UPS parallel robot experiment verified the effectiveness of the proposed trajectory error prediction and compensation method. Key findings showed that the accuracy of the sinusoidal trajectory and circular trajectory of the parallel robot after error compensation was improved by about 90%.

Vibration sensors, as critical components in motion control and measurement systems, have dynamic characteristics that directly affect measurement accuracy. However, existing sensitivity models, due to structural simplifications and parameter uncertainties, hinder conventional vibration and shock calibration methods from fully characterizing their dynamic performance. In addition, traditional parameter identification approaches are often noise-sensitive and lack interpretability, making them inadequate for high-precision applications. To address these challenges, this study proposes an Algorithm-Unrolled Interpretable Physics-Informed Neural Network (IPNN) for parameter identification of a vibration sensor sensitivity model. By integrating the physical characteristics of the sensors with vibration calibration data, the method enables high-precision parameter identification and interpretable dynamic modeling. Comparative experimental results show that the proposed IPNN reduces the RMSE of sensor voltage predictions by over 60% compared with GRU and LSTM and decreases the average full-frequency relative deviation from laser interferometry calibration results by approximately 65% relative to LSM.

Kinematic parameter errors of parallel robots are affected by manufacturing errors, assembly errors, and shape errors caused by heavy loads, resulting in a gradual decrease of their kinematic accuracy. This study proposes an online learning method for the kinematic parameter errors based on a six-degree-of-freedom (6-DOF) in-situ tracking system, which achieves their online identification. This method uses six high-precision measurement legs that are embedded in the parallel robot to achieve in-situ data measurement and adopts an online learning method to identify the kinematic parameter errors. Experimental results compared with the least squares method demonstrated that the proposed method effectively achieves online identification of the kinematic parameter errors, with position and orientation accuracy improved by 85.3% and 79.2%, respectively. Moreover, it can also maintain small position deviations even under varying loads, thus sustaining the high-accuracy motion of the parallel robot.

In recent years, the monocular vision method has been gradually applied in the precision vibration calibration. To ensure its measurement performance, a highly efficient and accurate traceability strategy for the monocular vision measurement has become especially important. In this study, a novel digital metrological traceability method for the monocular vision-based linear and angular vibration measurement is proposed. To facilitate unification and promotion, a standard digital vibration video database, along with a universal uncertainty evaluation model, has been established. This method achieves standard linear and angular vibration measurements by combining the sub-pixel feature edge and corner extraction techniques as well as an accurate rounding error correction model, whose frequency range can be extended to 0.01-10 Hz. The numerical experimental results show that the maximum relative deviations of the monocular vision-based linear and angular vibration measurements from the standard values are less than 0.070% and 0.069%, respectively, while the expanded measurement uncertainty is less than 0.093%. This method promotes the improvement and unification of the MV-based linear and angular vibration metrological traceability system and achieves a flat traceability chain. Additionally, it only requires a high refresh rate display and a series of standard digital vibration videos to achieve the highly efficient traceability, which enables MV-based in-situ calibration and traceability of vibration sensors and advances the digital transformation of vibration metrology.

To ensure that multi-axis vibration sensors provide accurate and reliable measurements, a hexapod-based calibration device (HBCD) with high-bandwidth and large stroke needs to be developed. The high-bandwidth motion provided by the HBCD can reach up to hundreds of mm at low frequencies and drop to only μm level at high frequencies. The decline of the feedback quality with the increasing frequency severely deteriorates the control performance of the HBCD. This study presents the design of an observer-based complementary filter (OBCF) to estimate the high-bandwidth motion state of the HBCD, addressing both low-frequency, large-stroke and high-frequency, small-stroke motion. A model-based nonlinear observer is constructed to remedy the position measurement by employing the control force. And a complementary filter is formed to combine the best frequency ranges of the control force and the position measurement. Stability of the feedback control system when implementing the OBCF is analyzed. Comparison experiments are conducted to evaluate the effectiveness of the proposed OBCF in achieving the high-bandwidth motion. The experimental results demonstrate that the testing capabilities of the HBCD are enhanced at high frequencies. The repeatability of the circular motion at 100 Hz and the harmonic distortion of the sine motion at 100 Hz can be enhanced from 0.46 m/s² to 0.22 m/s² and 25.2% to 6.2%, respectively.

The linear and angular accelerometers as the most basic motion measurement components, which have been widely utilized in various engineering fields. However, the existing calibration methods are difficult to determine their sensitivity magnitude and phase simultaneously with high accuracy and efficiency. This study proposes a novel monocular vision method to calibrate low-frequency linear and angular accelerometers, which has ability to determine the sensitivity magnitude and phase accurately by using spatiotemporal synchronization technology. This method improves the calibration performance by adopting a zero-encoder pulse to efficiently algin the spatial excitation and time signal, and it can be easily implemented on a flexible vision measurement system. Experiments compared with the comparison method and laser interferometry show the proposed method can obtain the calibration accuracies of sensitivity magnitude and phase about 0.3% and 0.3° with a considerable efficiency in the 0.01-10 Hz.

The most common forms of target multi-degree-of-freedom (MDOF) motion include plane and space motions, whose accurate measurement is crucial for the fields of pose estimation, inertial navigation, and structural health monitoring of buildings. Currently, the primary methods used for the MDOF motion measurement are laser interferometry, grating ruler-based method, and sensor-based method. However, the laser interferometry is operationally complex, costly, and it requires strict environmental conditions. Although the grating ruler-based method is low-cost and accurate, its applicable frequency range is limited. Moreover, the sensor-based method suffers from limited accuracy and dynamic range. To achieve accurate measurement of MDOF motion across a wide frequency range, a new monocular vision (MV)-based decoupling measurement method is investigated. This method utilizes a specially designed measurement mark, combined with an improved Harris corner positioning algorithm with sub-pixel accuracy and decoupling models for plane and space motions, which has the advantages of high measurement accuracy, simple algorithm, and strong robustness. Comparison experiments with the current methods demonstrated that the investigated method can simultaneously accomplish the decoupling measurements of plane and space motions, which achieved the accuracy of 0.66% and 1.29% in the range of 0.1-2 Hz.

Micro-acceleration generation during ultra-low-frequency micro-vibration calibration is a sensitive issue. There are issues of traditional pendulum tables being unable to change the pendulum radius and direction to produce micro-accelerations of different magnitudes, and the line shakers having a low signal-to-noise ratio when the vibration amplitude is the same as that of the pendulum tables. Therefore, a novel ultra-low-frequency micro-vibration calibration method is proposed to solve the above issues based on virtual pendulum motion trajectories of the Stewart platform. The micro-accelerations of 10–5 to 10–3 m/s2 can be generated by the trajectories with the radius of up to 12 m, the displacement amplitudes of up to 11.636 mm and the frequencies between 0.01 and 0.1 Hz. In the virtual pendulum motion, the maximum acceleration can be 2481 times greater than the acceleration of linear motion at the same frequency and displacement amplitude. In a comparison experiment with the current rotating platform, the maximum relative deviation of sensitivity amplitude calibration for pendulum motion around the x- and y-axis based on the Stewart platform are 0.411% and 0.295% respectively. The above results demonstrate the validity and reliability of this kind of method.