Timely response to customer needs is extremely important, at the same time, the internal and external qualities of the product need to be ensured. A good process design and control technique could lower the production cost, at the same time, reduce the scrap rate. However, the non-linear nature of the manufacturing, the close coupling between the thermal, mechanical and material phenomena, and the multi-step nature of optimization make the process difficult to formulate and solve. Advances in the numerical analysis tools have made it possible to model the metal flow in non-isothermal metal forming processes. However, the calculations often become time and computer resource intensive. Process optimizations become very difficult and unstable limiting the numerical tools to trial and error approaches.; In this dissertation, interdisciplinary approaches to optimization of multi-stage metal forming are investigated for application to different aspects of the metal forming processes. These approaches utilize the numerical efficiency and accuracy of the finite element method, fast processing and decision making ability of AI (Artificial Intelligence) techniques, and the strong planning function of the Design of Experiment (DOE) method. The inverse technique is examined for its efficiency in determining the thermomechanical processing history of the rolling mill for the desired final product attributes in roll pass and mill design. Moreover, a virtual soft sensor is developed for the metal forming process, and is applied to the hot forging of a wheel hub process. Finally, a roll pass design approach with minimum sensitivity is presented to accommodate the variance in the rolling process. In this dissertation, the interdisciplinary approaches are investigated, including Finite Element Method (FEM), ANN, Simulated Annealing (SA), and DOE and other statistical techniques.
展开▼
