Heating mechanisms of ultrasonic welding for thermoplastics were studied via numerical simulation and experiment. Relationship between dynamic modulus and static relaxation modulus of thermoplastics was theoretically analyzed. Segment time—temperature equivalency equation for different temperature ranges and a method for directly separating viscoelastic heat from strain energy were used in the simulation. Temperature measurement tests with poly(methyl methacrylate) specimens were carried out to verify the simulation results. Results of simulation and experiment reveal that interfacial friction rather than viscoelastic heat initially start the whole welding process. Viscoelastic heating becomes dominant when temperature reaches T_g (glass transition temperature) of the material. And viscoelastic heat provides most required heat during welding.

The machine vision measurement module is indispensable for the Modularized Flexible Precision Assembly Station (MFPAS), which is a fully automatic assembly system being developed at Dalian University of Technology (DUT). MFPAS consists of basic and additional modules, and are expected to be flexible, expandable, and re-configurable to adapt to a variety of parts with a large size range, requiring the machine vision measurement module to be able to achieve accurate measurement of position, as well as orientation of the parts with different size scale. An automatic zooming vision system was set up for evaluation and final integration in MFPAS. Pixel equivalent, principal point and orientation deviation of images were analyzed and experimentally studied using different magnifications of the lens. A new template with circular patterns of different diameters was designed for zoom-lens calibration. The experiments show that the measurement error caused by the variation of the pixel equivalent, principal point and orientation is estimated under 10 μm without online calibration. When high accuracy is required, online calibration can be employed during assembly. The evaluation results of the vision system with or without on-line calibration were given for a better trade-off between accuracy and efficiency during assembly.

Eccentricity measurement of annular parts with millimeter scale and micrometer precision requirements is widely used in mechanical engineering applications. To realize accurate eccentricity measurement for large-scaled annular parts, a vision-based and sub-pixel dimensional measurement method is proposed. First, to facilitate the eccentricity measurement, an improved auto focus algorithm is introduced to provide better focused images of the measured parts. Then the traditional Canny operator is modified in gradient direction calculation and a double threshold process to locate the pixel edge more accurately. Next, a model-based sub-pixel edge detection method is studied to extract the sub-pixel edge coordinates. Finally, the eccentricity is calculated according to these sub-pixel edge coordinates. To guarantee measurement accuracy, the pixel equivalent and manual installation error of three degree of freedom (DOF) stages are calibrated, and the verification experiments indicate that the measurement error of the proposed method is better than 1.0 µm.