研究目的
To investigate vibration control on a two-connected piezoelectric flexible hinged plate using a fuzzy neural network algorithm based on binocular vision measurement, focusing on acquiring vibration signals and damping out residual vibrations quickly.
研究成果
The binocular stereo vision vibration detection method and FNN controller are feasible for detecting and attenuating vibrations in a hinged flexible plate. The FNN controller outperforms PD control in suppressing both large and small residual vibrations more effectively, demonstrating adaptiveness and robustness through online parameter adjustment.
研究不足
The frame rate of the cameras is limited to 50 Hz, restricting measurement to dominant low-frequency modes. The use of pasted marks affects the vibrating object's characteristics, and the method does not fully reflect non-contact vision measurement. Future work should consider spot projection for marks and other control algorithms like model predictive control.
1：Experimental Design and Method Selection:
The study involves constructing an experimental setup with a piezoelectric flexible hinged plate, binocular stereo vision sensors, and piezoelectric transducers. A fuzzy neural network (FNN) controller is designed for adaptive vibration suppression, compared with proportional and derivative (PD) control. The binocular stereo vision model uses a pin-hole camera model for 3D displacement measurement.
2：Sample Selection and Data Sources:
A hinged plate apparatus with bonded piezoelectric patches and circular marks is used. Vibration signals are acquired from images captured by two industrial cameras and from piezoelectric sensors.
3：List of Experimental Equipment and Materials:
Includes industrial cameras (BASLER acA1920-150uc), camera lenses (COMPUTAR M1214-MP2), motion control board (GTS-400-PV-PCI), voltage amplifiers (APEX PA240CX chips), charge amplifiers, piezoelectric transducers, and a computer. Circular marks are pasted on the plate for vision detection.
4：Experimental Procedures and Operational Workflow:
Calibration of the stereo vision system using Zhang's chessboard method. Image processing with OpenCV for contour extraction and stereo matching. Vibration excitation by hitting the plate at specific points. Control application after 1 second, with FNN and PD controllers tested for bending and torsional vibrations. Data collection at 50 Hz frame rate.
5：Data Analysis Methods:
Vibration signals are analyzed in time domain and frequency domain using power spectrum density. FNN parameters are updated using backpropagation learning algorithm. Performance is compared between FNN and PD control.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
