Tianjin University
Assembly Line
Ultrasonic Extruded Weld-Riveting
Traditionally, either fasteners or adhesives have been used to join metal parts to carbon fiber-reinforced thermoplastic (CFRTP). Alternatively, welding is being used to join metal and CFRTP. The mainstream method of welding metal and CFRTP is first to modify the metal part by creating macro- or microstructures on its surface. These structures help to increase the effective connection area and improve the mechanical interlocking between the two parts. Next, a thermal technique, such as laser welding, friction stir welding or hot-press welding, is used to heat the metal part above the melting point of the CFRTP. Heat from the metal then melts the CFRTP, and the molten resin flows into the surface structures of the metal part to form a joint.
The problem with this method is that the heat can adversely affect the metal part, especially for metals with poor high-temperature resistance, such as magnesium alloys. We have developed a new technique, ultrasonic extruded weld-riveting, that shows promise for joining metal to CFRTP without a third component and without damaging either part. In our method, prefabricated through-holes are machined on the metal part. The CFRTP part is placed on top of the metal one. Ultrasonic welding is used to melt the CFRTP, and the molten plastic is squeezed into the holes in the metal. A riveted joint will then form after the plastic cools and solidifies.
Phase partition and online monitoring for batch processes based on Harris hawks optimization
Most industrial batch processes exhibit significantly different characteristics at different manufacturing steps, and it is advantageous to partition batch processes reasonably and establish phase models separately for online monitoring. In the present work, a novel phase partition method is developed based on Harris hawks optimization (HHO) with hard sequentiality constraint, which seeks for the optimal phase partition results under the specific target phase number in inner loop and automatically determines the optimal phase number in outer loop by making a trade-off between modeling complexity and partition performance. First, a new definition of the sum of quadratic error (SQE) is designed as the fitness function to evaluate the within-phase compactness, which makes the time-slice matrices with similar process variable correlations stay in the same phase. Then, an optimization refinement scheme (ORS) is developed to find the potential local minimum of the SQE by inspecting and reallocating the phase-adjacent samples. Afterwards, the percentage of performance improvement indicator (PPII) is proposed to determine the optimal phase numbers by quantifying the improvement of minimum SQE. For each subphase, canonical variate analysis (CVA) is performed to build statistical models for dynamic process monitoring. The effectiveness of the proposed method is illustrated by a numerical example with some outliers and an injection molding process.