Optimization analysis of cold extrusion process parameters for valve screws

Abstract

This study focuses on valve screws and establishes a cold extrusion simulation model using DEFORM-3D software to analyze the deformation characteristics and influencing factors during the extrusion process. Supported by the theory of rigid-plastic finite element, an orthogonal experimental design is conducted to establish a four-factor three-level simulation scheme with general fillet radius, die clearance, extrusion speed, and friction coefficient as the simulation variables, and maximum equivalent stress on the convex die, forming damage of the blank, and maximum wear depth on the concave die as the objective values. The prediction model based on BP neural network algorithm and orthogonal experimental data shows high accuracy. The optimization problem is simplified into a dimensionless numerical analysis problem using GC grey correlation analysis, and the optimal solutions within each variable range are determined using range analysis. The simulation results after optimization show a 20 5% reduction in maximum equivalent stress on the convex die compared to the original scheme, a 4.9% reduction in forming damage of the extruded part, and a 24.24% reduction in maximum wear depth on the concave die compared to before optimization. The actual verification confirms the good forming effect of valve screw extrusion, with well-formed contours and a uniform surface without wrinkles or defects. Finite element simulation provides valuable guidance for practical production.

Keywords

• valve screw
• extrusion forming
• orthogonal experiment
• neural network
• grey analysis