Application and Evaluation of Trinocular Photographic Technique in Vibration Analysis of Rotating Machinery
In response to the needs for vibration analysis and stability assessment of rotating machinery,a modal identification measurement method based on trinocular photography was proposed.This method integrated data from three cameras,character encoding marker design.By introducing convolutional layers into the traditional multi-layer perceptron neural network structure,increasing net-work depth & width,and integrating the'Squeeze-and-Excitation'(SE)module,a new network structure named HybridNetwork was formed.The detection accuracy of vibration data in rotating machinery was significantly improved by HybridNetwork.Utilizing the binocu-lar stereo vision principle and information fusion technology,the 3D coordinates of the target were calculated accurately,and the time-domain vibration response of the mechanical structure was effectively reflected.By analyzing the time-domain response data,the modal frequencies of the rotor system were successfully extracted and compared with the results from a laser Doppler vibrometer.The results demonstrate that the relative error of this method is controlled within 0.853％,and it offers a more comprehensive perspective on spatial displacement and vibration patterns,effectively proving its feasibility and reliability in the field of vibration analysis of rotating machinery.
万方数据
维普
