Instantaneous Rotational Speed Measurement Of Rotating Machinery Through Image Processing

The rotational speed is an essential parameter to monitor and control the condition of rotating equipment such as generators,motors,centrifuges,etc.Real-time speed measurement is essential for precision machining and fault warning.Therefore,accurate and reliable speed measurement is required in various industries.With the advent of high-performance GPUs and the improvement of image processing speed by system scientists,image processing(especially in the case of image capture,recognition,and classification)has achieved a huge leap.The image class classifies images by training a supervised model,helping to solve problems that can be solved using a large number of annotated data sets.This dissertation also highlights the limitations of the existing techniques such as contact-based or contactless tachometers.A few may suffer from wear and tear,and a few do not perform well in tough working environments.There are several commercially available tachometers.but because of their high installation costs and high maintenance costs,they are not considered customer-friendly for large-scale industrial experiments.Therefore,commercially there is a need to incorporate an approach based on real-time online data experiments;and,merge this approach with the existing technology used for early fault diagnosis for accurate and efficient experimental results.Accurate and dynamic measurement of instantaneous rotational speed is the requirement for intelligent monitoring and diagnosis of the condition of rotating equipment.This thesis proposes a real-time speed measurement system based on an inexpensive imaging device with a marker on the rotor.The marker used in our study is QR code.The sequential image is pre-processed and converted from RGB to grayscale,and then processed through the similarity evaluation method to obtain a certain degree of similarity with respect to the reference image.Finally,an algorithm for determining the periodicity in the reconstructed signal is used to obtain the rotational speed of the rotor.A schematic diagram of our proposed system,its structure,and performance evaluation are given.The influence of various influencing factors such as the frame rate.the algorithm for evaluating image similarity,and the algorithm for determining the frequency on the performance of the measurement system are quantified and discussed in a series of experimental tests on a working system.Experimental results of our proposed system show that it can continuously measure speed in the range of 100 to 1800 rpm when 10 to 500 frames are given per second,proving that our system can accurately measure the instantaneous rotational speed(IRS).