Abstract
Assembly is the final process of manufacturing, and a good assembly plan reduces the effect of the tolerance generated in the early stages by the tolerance elimination. In the current assembly lines, the assemblers pick up the workpieces and install them together by the assembly instructions. When the workpieces are oversize or undersize, the product can not be installed correctly. Therefore, the assembler considers the secondary processing to fix the tolerance and then installs them together again. The product could be installed, but the product quality may be reduced by the secondary process. So, we formulate the assembly process as a combinatorial optimization problem, named by the dimensional chain assembly (DCA) problem. Given some workpieces with the corresponding actual size, computing the assembly guidance is the goal of the DCA problem, and the product quality is applied to represent the solution quality. The assemblers follow the assembly guidance to install the products. We firstly prove that the DCA problem is NP-complete and collect the requirements of solving the DCA problem from the implementation perspective: the sustainability, the minimization of computation time, and the guarantee of product quality. We consider solution refinement and the solution property inheritance of the single-solution evolution approach to discover and refine the quality of the assembly guidance. Based on the above strategies, we propose the assembly guidance optimizer (AGO) based on the simulated annealing algorithm to compute the assembly guidance. From the simulation results, the AGO reaches all requirements of the DCA problem. The variance of the computation time and the solution quality is related to the problem scale linearly, so the computation time and the solution quality can be estimated by the problem scale. Moreover, increasing the search breadth is unnecessary for improving the solution quality. In summary, the proposed AGO satisfies with the necessaries of the sustainability, the minimization of computation time, and the guarantee of product quality for the requirements of the DCA, and it can be considered in the real-world applications.
Highlights
The automatic and intelligent manufacturing is the vision in Industry 4.0
We prove that the dimensional chain assembly (DCA) problem is a NP-complete problem, and the search algorithms [7,8] requires huge computation time to find the optimal solution
To prove that the DCA problem is NP-complete, we reduce the subset of the DCA from the exact weight perfect matching problem [25]
Summary
The automatic and intelligent manufacturing is the vision in Industry 4.0. The automation means that the manufacturing process can be started and finish without human’s operation. The AG computation must cover all kinds of product assembly processes that includes different number of workpieces and parts, the assembly sequence, the final size, etc. To guarantee of product quality, we modify the simulated annealing (SA) algorithm to design the assembly guidance optimizer (AGO) to calculate the AG. The sustainability: the assemblers obtain the AG from the proposed AGO in the different DCA problems (to install various products). 2. The minimization of computation time: the AGO outputs the AG for installing thousands of workpieces in few seconds, and the computation is finish before all workpieces arrive at the assembly line. Given an assembly configuration of the AGO, the increase of the computation time and the decrease of the solution quality is linear to the problem scale. The AGO can be applied to the real-world assembly lines for computing the AGs of various products
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
