Decomposition of overlapping peaks in X‐ray fluorescence using improved crow searching algorithm based on opposite learning. (16th November 2021)

• Record Type: Journal Article
• Authors: Zhao, Xueliang; Sun, Shicheng; Li, Meijie; Zhang, Jie; Li, Kang; Pang, Lili; Shi, Yanxin; Chen, Ying
• Is Part Of: Journal of chemometrics. Volume 35:Number 12(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-11-16
• Subjects: heavy metal -- improved CSA -- opposite learning -- overlapping peaks decomposition -- X‐ray fluorescence spectroscopy; Chemistry -- Mathematics -- Periodicals; Chemistry -- Statistical methods -- Periodicals; 542.85
• DOI: 10.1002/cem.3377
• Languages: English
• ISSNs: 0886-9383

Abstract

Abstract: To overcome the difficulties in overlapping peaks decomposition in X‐ray fluorescence (XRF) spectrum of heavy metal Pb and As, an improved crow searching algorithm (CSA) (ICSA) method was proposed. Compared with the CSA, three main improvements were performed: the opposite learning (OL) mechanism was introduced to expand the diversity of population and the parameter TF was introduced to control the convergence, which could avoid falling into the local minimum during the initialization process. Furthermore, this method has made some adjustments on the global optimization strategy, which can make the convergence more stable. When using the ICSA to decompose the overlapping peaks, the resolution accuracy improved by 2.23%, and the mean square error is better than other four algorithms of the same type, showing the feasibility of ICSA in dealing with the overlapping peaks. Abstract : To overcome the difficulties in overlapping peaks decomposition in XRF spectrum of Pb and As, an improved crow searching algorithm (ICSA) method was proposed. The OL mechanism was introduced to expand the diversity of population. The new parameter TF was introduced to control the convergence. The global optimization strategy was adjusted to make the convergence more stable. The resolution accuracy improved by 2.23%, and the mean square error is better than other similar algorithms, showing the feasibility of ICSA.