Introduction to the Problem
In vacuum forming, achieving consistent product quality is heavily dependent on selecting the appropriate process parameters, such as heating time, pressure, and material thickness. Improper parameter settings can lead to common defects like incomplete forming, warping, thinning, or uneven surface finishes, resulting in suboptimal products and material waste. Given the variety of materials, geometries, and production conditions involved in vacuum forming, identifying the ideal set of process parameters for each scenario is both challenging and time-consuming. Manual trial-and-error approaches often introduce variability and increase production costs. Therefore, developing accurate methods for automatically generating optimal process parameters is essential to minimize errors and defects, ensure high-quality results, and streamline the vacuum forming process across diverse materials and designs.
Solution
We have successfully tested the performance of our CNN-based vision model, which suggests optimal process parameter adjustments for the vacuum forming process by analyzing images of the material. The model has proven effective in improving forming quality by providing accurate suggestions. To assess the model’s generalizability, we conducted experimental validations across various conditions. These experiments included forming a hemisphere with sheets of different colors but the same thickness (1 mm), forming hemispheres with varying thicknesses (1.5 mm and 2 mm) and colors, and forming different geometrical shapes, such as a tapered box and a circular frustum, using 1 mm thick sheets. The results showed that the model performs well across different colors, accurately recommending process parameter changes. For varying material thicknesses, the model correctly suggested the direction of adjustments, although the precision decreased slightly, which is expected since thicker materials require more heat or time. A scaling factor based on material thickness may be necessary. Additionally, the model provided reliable suggestions for parameter changes when forming different geometrical shapes, demonstrating its versatility and robustness across a range of forming scenarios. The data collection and image processing is shown as figure 1. Vacuum forming processes for different shapes is shown in figure 2.
Figure 1. Image collection and processing
Figure 2. Vacuum forming processes for different shapes