Partition KMNN-DBSCAN Algorithm and Its Application in Extraction of Rail Damage Data

Abstract

In order to realize intelligent identification of rail damage, this paper studies the extraction method of complete damage ultrasonic B-scan data based on the density-based spatial clustering of applications with noise algorithm (DBSCAN). Aiming at the problem that the traditional DBSCAN algorithm needs to manually set the Eps and Minpts parameters, a KMNN-DBSCAN (K-median nearest neighbor DBSCAN) algorithm is proposed. The algorithm first uses the dataset’s own distribution characteristics to generate a list of Eps and Minpts parameters and then determines the optimal Eps and Minpts through an optimization strategy to achieve complete self-adaptation of the two parameters of Eps and Minpts. In order to further improve the clustering performance of the algorithm, the partition idea is introduced, and the partition KMNN-DBSCAN algorithm is proposed to solve the problem that the clustering results of the DBSCAN algorithm are inconsistent with the actual categories on datasets with uneven density. The experimental results show that the KMNN-DBSCAN algorithm has higher clustering accuracy and silhouette coefficient (SC) for the D037 dataset ultrasound information group (UIG) division; compared with the KMNN-DBSCAN algorithm, the proposed partition KMNN-DBSCAN algorithm has higher clustering accuracy, F-Measure, and SC values. The partition KMNN-DBSCAN algorithm achieves accurate division of all damage UIG on the damaged B-scan data with large density differences, and completes the effective extraction of complete damage data.

WOA-DBSCAN: Application of Whale Optimization Algorithm in DBSCAN Parameter Adaption

“…For each object in the datasets, the classification result is deterministic, allowing us to evaluate the clustering performance using ACC, AMI, and ARI metrics. WOA-DBSCAN algorithm, which we proposed, combines the K-Medians Nearest Neighbor algorithm [16] to determine the value ranges of Eps and MinPts. It also incorporates the concept of density peak clustering to identify the optimal number of clusters.…”

“…The optimal Eps parameter is identified when the generated clusters are consistent for three consecutive times, and the corresponding MinPts is obtained using the mathematical expectation method. Li et al [16] proposed a Partition KMNN-DBSCAN Algorithm, which constitutes a K-median nearest neighbor set as a list of Eps by calculating the K-nearest neighbor distance matrix of the input dataset and then finding the median of the K-nearest neighbor distances of all elemental points. The median method and the given list of Eps parameter values are then used to generate a list of Minpts parameter values.…”

Performance of K-Means and DBSCAN Algorithm in Clustering Gross Regional Domestic Product

“…Zhang (2022) applied DBSCAN in information security detection and also combine BIRCH with DBSCAN. Li, Yang, Jiao, and Li (2022) From some studies, the accuracy of K-Means and DBSCAN is depending on the data. In DBSCAN algorithm, determine MinPts and 𝜀 can affects the result too.…”

Interaction of Crime Risk across Crime Types in Hotspot Areas

“…The traditional DBSCAN algorithm can produce errors when clustering datasets with large density differences. Therefore, to determine the experimental parameters, the K-means nearest neighbor method based on the self-decay term was used to generate the optimal value of Eps, and the mathematical expectation method based on the self-decay term was used to generate the optimal value of Minpts, while five was assigned as the number of continuous and stable clusters for use as the parameter with which to determine the stability interval [38]. The steps of the DBSCAN algorithm are included in the Supplementary Materials (Algorithm S1: DBSCAN algorithm).…”