Abstract

Info

Title:: Anomalous pressure detection in sampling systems based on Gramian angular field and parallel KConvNeXt

Author(s):: ZHANG Qi1; XIAO Shenping2; NIE Libo1; PENG Yuangang3; SONG Yongbo3; 1. School of Life Science and Chemistry, Hunan University of Technology, Zhuzhou 412007, China 2. School of Electrical and Information Engineering, Hunan University of Technology, Zhuzhou 412007, China 3. Shenzhen YHLO Biotech Co., Ltd., Shenzhen 518116, China

Keywords:: Keywords: sampling system anomaly detection Gramian angular field ConvNeXt network

Abstract:: Abstract: A detection model based on Gramian angular field (GAF) and parallel KConvNeXt network is proposed for accurately detecting the abnormal conditions caused by sample needle blockage in the sampling system during the sampling, thus improving the testing accuracy and detection efficiency of automated biochemical analyzers. GAF-based method is employed to transform the time series of one-dimensional pressure signals into two-dimensional image representations. Subsequently, an improved attention mechanism integrated with a parallel dual-channel KConvNeXt network is used to classify the pressure signals, and achieves a final classification accuracy of 94.58%. The experimental results show that the proposed method can effectively capture the key characteristics of the pressure signals, offering an efficient solution for the anomalous pressure detection in biochemical analyzer sampling system and exhibiting important practical significance.