Issue 78

D. S. Lobanov et alii, Fracture and Structural Integrity, 78 (2026) 207-219; DOI: 10.3221/IGF-ESIS.78.12

between two data sets of equal size to be assessed in cases where the relationship is not linear, or the data exhibit significant variation. The essence of the method lies in converting absolute values into their numerical order (ranks). If the ranks are ordered in ascending or descending order and, in most cases, coincide, this indicates a direct relationship. In cases where the order is reversed, an inverse relationship is inferred. To determine the monotonic relationship between parameters, the following formula is used:

( ) 2 i 2 6 Σd

p=1

n n -1

where: d i 2 - Rank difference, n – Number of data points. Values range from +1 to -1, where a positive value indicates a perfect linear relationship, a negative value indicates an inverse relationship, and 0 indicates the absence of a monotonic relationship. To analyze the results obtained, the standard correlation strength scale is used, where a value above 0.8 is considered a very strong correlation, between 0.8 and 0.6 a strong correlation, between 0.6 and 0.4 a moderate correlation, and 0.2 and below a weak correlation. Based on the data from the correlation analysis, parameters were selected for further data clustering. As the original data exhibits a spread of values across several orders of magnitude, the Silhouette Index and the Davies-Bouldin Index were used to determine the optimal number of clusters. Clustering was performed using two methods: K-means and the Gaussian Mixture Model (GMM). K-means was selected as the primary method for determining the initial clusters, their centroids and boundaries. GMM was selected as a supplementary method, allowing for the possible elliptical shape of clusters and transitions between types of AE signals. R ESULTS AND DISCUSSION nalyzing the defective structure after preliminary defect testing on fiberglass specimens under controlled loading parameters, it's worth noting that a sphere-shaped indentation causes damage associated with matrix fracture and fiber rupture around the sphere's imprint. A cylinder dent indentation results in brittle matrix fracture between the surface layers without fiber breakage. A scratch defect results in cutting of the upper layers of the material with localized delamination of the surface layer around the indentation area. A total of 32 samples were tested, 4 samples in each group. The test results have yielded new data on the effect of external surface defects in fiberglass composites on mechanical properties in the warp and weft directions; the load-displacement curves are shown in Fig.5, and the compressive strength results were presented in Tab. 1 and Fig. 6. In Fig.6, the test results were presented as a histogram for clarity. A

a b Figure 5: Typical load-displacement diagrams for glass-fiber-reinforced plastic specimens in the weft direction (a) and in the warp direction (b): no defect – black line; indentation caused by a cylinder – red line; indentation caused by a sphere – blue line; scratch – green line

211