Train wheel sets must be periodically inspected for possible or actual premature failures and it is very significant to record the wear history for the full life of utilization of wheel sets. This means that an online measuring system could be of great benefit to overall process control. An online non-contact method for measuring a wheel set’s geometric parameters based on the opto-electronic measuring technique is presented in this paper. A charge coupled device (CCD) camera with a selected optical lens and a frame grabber was used to capture the image of the light profile of the wheel set illuminated by a linear laser. The analogue signals of the image were transformed into corresponding digital grey level values. The ‘mapping function method’ is used to transform an image pixel coordinate to a space coordinate. The images of wheel sets were captured when the train passed through the measuring system. The rim inside thickness and flange thickness were measured and analyzed. The spatial resolution of the whole image capturing system is about 0.33 mm. Theoretic and experimental results show that the online measurement system based on computer vision can meet wheel set measurement requirements.

A speckle interferometer which can measure whole field three-dimensional displacements continuously and dynamically has been built. Three different wavelength lasers are used to produce the speckle interferograms of the two in-plane displacements (displacements in the x- and y-direction) and one out-of-plane displacement (displacement in the z-direction), respectively. One color CCD camera is employed to collect these mixed speckle interferograms simultaneously. The mixed interferograms are separated by the Red, Green and Blue channels of the color CCD camera, and then are processed by the wavelet transform technique to extract the phase information of the measured object. The preliminary experiment is carried out to demonstrate the performance of this new device.

A Fourier-transform imaging spectrometer, believed to be novel, based on the Savart polariscope is presented. There is no slit in this instrument, which means that it has a high throughput. The principle and the system configuration are described. Several preliminary experimental results are shown.

The three-dimensional measurement of structured light is commonly used and has widespread applications in many industries. In this study, machine learning is used for structured light 3D measurement to recover the phase distribution of the measured object by employing two machine learning models. Without phase shift, the measurement operational complexity and computation time decline renders real-time measurement possible. Finally, a grating-based structured light measurement system is constructed, and machine learning is used to recover the phase. The calculated phase of distribution is wrapped in only one dimension and not in two dimensions, as in other methods. The measurement error is observed to be under 1%.

Many vision-based bolt looseness detection methods that directly observe the bolts have been developed. However, these methods have many limitations in terms of the conditions and processes of their implementation. To address these problems, this paper proposed a fully automated vision-based bolt looseness detection method for a rigid multi-bolt connection. The proposed method combines digital shearing speckle pattern interferometry (DSSPI) and recurrent neural network (RNN) and involves capturing speckle fringe patterns under various looseness cases using the DSSPI system and classifying these patterns with an RNN model to detect the loose bolts. The proposed method can detect all the bolts within the measured surface at one time, which is efficient. On the other hand, it eliminates the need for prior information such as the initial angle and position of each bolt. It can even detect unseen bolts in multi-bolt connections, making it applicable for connections in complex structures in which occlusion often occurs. Additionally, the method eliminates the complex process of distortion rectification. These features make the method achieve a single-judgment time (four bolts at one detection) of only 4.70 millisecond with a detection accuracy over 99%, which has potential for the real-time detection of loose bolts in multi-bolt connections.