Research On Outlier Detection Algorithms And Their Applications In Passive Localization

With the arrival of the Internet era,massive data is now ubiquitous,and traditional data processing methods struggle to handle large data sets,giving rise to data mining algorithms.Data mining refers to the process of automatically exploring and analyzing potential patterns and rules in large data sets,one important area of which is outlier detection.Outlier detection is the task of identifying data that is significantly different from other data in the dataset.This article summarizes the background,significance,and research status of outlier detection both domestically and internationally,and analyzes the nearest-neighbor-based outlier detection method.In response to current problems in outlier algorithms,the article proposes improvements while studying the practical application of these improvements to non-source positioning.The research content of this article is as follows:(1)Analyzing existing outlier detection algorithms,comparing the advantages and disadvantages of commonly used outlier detection algorithms,and researching and analyzing common non-source bearing cross-location algorithms and data preprocessing technologies.(2)In response to the sensitivity of the nearest-neighbor-based outlier detection algorithm to the parameter k(neighbor quantity),the article proposes an improved outlier detection algorithm based on relative density accumulation.This algorithm defines a relative density accumulation factor to measure the outlier degree of data objects,and experiments show that the improved algorithm is more stable than other nearest-neighbor-based algorithms for any parameter k.(3)The proposed improved algorithm is applied to non-source positioning to address the intersection set formed by bearing lines in non-source bearing cross-location.First,the improved algorithm is used to exclude most of the false intersection points in the intersection set,and then the minimum least-square algorithm based on point-to-point distance is used to locate the target position.By comparing the accuracy of traditional positioning algorithms,it is confirmed that the proposed algorithm is more effective.In the outlier detection experiments,synthetic datasets with different shapes,densities,distances,and sizes are used.Similarly,in non-source positioning experiments,datasets produced by different measurement stations and bearing angles are used.The experimental results of both indicate that the proposed algorithm outperforms certain classical algorithms.