Research on the accuracy of dieless single point incremental forming based on machine vision

Dieless single point incremental forming is a manufacturing process that has gained popularity due to its high flexibility and low cost, making it suitable for individual and small batch production. However, ensuring optimal geometric accuracy of the formed parts during this process remains challenging. This paper presents a novel high-accuracy incremental forming model by leveraging machine vision technology that acquires, processes, and analyzes contour data of the part online. And a genetic algorithm backpropagation neural network is developed to learn the relationship between the toolpath and the actual contour using data from both finite element analysis results and real parts. The network then generates an optimized tool path based on the desired contour of the next forming pass. Additionally, we use a ring gasket to constrain material deformation and minimize errors in part geometry. Our experimental results demonstrate that this approach effectively limits deviations to within 0.5 mm and eliminates defects, thereby significantly increasing the efficiency and accuracy of the incremental forming process.