Topological Measures for Pattern quantification of Impact Centers in Piezo Vibration Striking Treatment (PVST)

TL;DR

This paper applies Topological Data Analysis to quantify the pattern of impact centers in surfaces treated by Piezo Vibration Striking Treatment, offering a new, automated way to assess surface pattern order and deviations.

Contribution

It introduces a TDA-based framework for analyzing impact center patterns in PVST-treated surfaces, overcoming limitations of manual and expert-dependent methods.

Findings

TDA effectively characterizes impact pattern types.

The method quantifies deviations from nominal impact grids.

TDA simplifies surface pattern analysis.

Abstract

Surface texture influences wear and tribological properties of manufactured parts, and it plays a critical role in end-user products. Therefore, quantifying the order or structure of a manufactured surface provides important information on the quality and life expectancy of the product. Although texture can be intentionally introduced to enhance aesthetics or to satisfy a design function, sometimes it is an inevitable byproduct of surface treatment processes such as Piezo Vibration Striking Treatment (PVST). Measures of order for surfaces have been characterized using statistical, spectral, and geometric approaches. For nearly hexagonal lattices, topological tools have also been used to measure the surface order. This paper utilizes tools from Topological Data Analysis for quantifying the impact centers' pattern in PVST. We compute measures of order based on optical digital microscope images of surfaces treated using PVST. These measures are applied to the grid obtained from estimating the centers of tool impacts, and they quantify the grid's deviations from the nominal one. Our results show that TDA provides a convenient framework for the characterization of pattern type that bypasses some limitations of existing tools such as difficult manual processing of the data and the need for an expert user to analyze and interpret the surface images.