Non-contact vibration frequency measurement method based on deep learning and image processing

Abstract

The invention relates to a non-contact vibration frequency measurement method based on deep learning and image processing. The method comprises the following steps: 1) selecting a to-be-measured areaof a measured object and extracting an image sequence of the to-be-measured area; 2) inputting a first frame of the image sequence into a deep convolutional neural network, and dividing all pixels inthe image into zero-value pixels and non-zero pixels; 3) selecting all non-zero pixels as effective pixels, and storing the coordinates of the non-zero pixels in a list for vibration signal extraction; 4) for each image in the image sequence, converting the image into a brightness signal by adopting an optical flow method, and then extracting a speed signal time history of each effective pixel; 5)averaging and normalizing the speed signal time history of all effective pixels; and 6) performing power spectral density estimation through Fourier transform to obtain power spectral density, and obtaining frequency composition through a peak value pickup method to obtain a non-contact vibration frequency measurement result. The method is beneficial to improving the efficiency and accuracy of non-contact vibration frequency measurement.

Description

technical field [0001] The invention belongs to the technical field of vibration frequency measurement, and in particular relates to a non-contact vibration frequency measurement method based on deep learning and image processing. Background technique [0002] Vibration frequency measurement is an important part of many engineering practices. Vibration frequency measurement is widely used in the fields of mechanical engineering and civil engineering. For example, the health monitoring system of large civil structures, the non-destructive testing system, the operation and maintenance monitoring of mechanical equipment, etc. Nowadays, most vibration measurement methods require the sensor to be attached to the target object to be measured. The development of contact measurement has been widely used and has high precision, such as acceleration sensors, speed sensors, strain gauges, etc. However, the installation process of traditional contactors is complicated, requiring the i...

AI Technical Summary

Problems solved by technology

However, the installation process of traditional contactors is complicated, requiring the installation and deployment of power supply lines and signal lines, and it will be difficult to install and measure at some difficult-to-reach objects.

In addition, the contact sensor will introduce extra mass to those measured objects with small mass or very thin mass, which will introduce measurement errors and even make it impossible to measure

Images