Page 67 - 《中国医疗器械杂志》2025年第6期

Chinese Journal of Medical Instrumentation 2025年 第49卷 第6期

临 床 医 学 工 程

文章编号：1671-7104(2025)06-0653-07

基于粒子群优化算法的BP神经网络在CT扫描仪

球管故障诊断中的应用研究

1
【作 者】 薛张华 ，王杰 ，顾成云 2
1
1 上海市胸科医院（上海交通大学医学院附属胸科医院）后勤保障部，上海市，200030
2 上海市第六人民医院 基建办公室，上海市，200233
【摘 要】 目的 提出一种基于粒子群优化（particle swarm optimization, PSO）算法优化BP神经网络的CT扫描仪球管
故障诊断方法，用于识别灯丝开路、灯丝半开路、扫描噪声和电弧放电等典型故障。方法 选取阳极电压、
电流、球管温度、扫描时间、电流波动幅度等关键参数，基于357组故障样本构建数据集，将其划分为训练
集与测试集。采用PSO优化BP神经网络的初始权重与阈值，构建PSO-BP模型进行故障分类。结果 该模型
在训练集和测试集上的分类准确率分别为96.25%和92.31%，各故障类型的皮尔逊相关系数介于
0.894~0.971，表明故障特征与类别间存在较强线性关系。与传统方法相比，PSO-BP模型在准确性和鲁棒
性方面具有明显优势。结论 基于PSO的BP神经网络能有效提升CT球管典型故障的诊断性能，具有良好的
应用前景，为CT设备智能诊断提供了新的方法。
【关 键 词】 PSO-BP神经网络；球管故障；计算机断层扫描术；灯丝故障；电弧放电
【中图分类号】 R197.39; TH77
【文献标志码】 A doi: 10.12455/j.issn.1671-7104.250320
Application of BP Neural Network Based on Particle Swarm Optimization
Algorithm in Fault Diagnosis of CT Scanner Bulb Tube

1
1
【 Authors 】 XUE Zhanghua , WANG Jie , GU Chengyun 2
1 Logistics Support Department, Shanghai Chest Hospital, Shanghai Jiao Tong University School of
Medicine, Shanghai, 200030
2 Infrastructure Construction Office, Shanghai Sixth People’s Hospital, Shanghai, 200233
【 Abstract 】 Objective To propose a CT scanner tube fault diagnosis method based on a BP neural network optimized
by particle swarm optimization (PSO) algorithm for identifying typical faults such as filament open circuit,
filament half open circuit, scanning noise and arc discharge. Methods Key parameters such as anode
voltage, current, tube temperature, scanning time, and current fluctuation amplitude are selected, and the
data set is constructed based on 357 sets of fault samples, which are divided into training set and test set.
PSO is used to optimize the initial weights and thresholds of BP neural network, and PSO-BP model is
constructed for fault classification. Results The classification accuracies of the model on the training and
test sets are 96.25% and 92.31%, respectively, and the Pearson correlation coefficients of each fault type
range from 0.894 to 0.971, indicating a strong linear relationship between the fault features and
categories. Compared with the traditional methods, the PSO-BP model has obvious advantages in terms
of accuracy and robustness. Conclusion PSO-based BP neural network can effectively improve the
diagnostic performance of typical faults of CT bulb tube, which has a good application prospect and
provides a new method for the intelligent diagnosis of CT equipment.
【Key words】 PSO-BP neural network, tube fault, CT, filament fault, arc discharge

收稿日期：2025-05-14
基金项目：浙江省教育厅科研项目（Y202557995 ）基于人工智能 0 引言
的医疗设备预算论证评估模型；医疗设备社会化服务
的增效路径及设备配置优化研究；2025年度上海交通
大学中国医院发展研究院决策咨询课题（CHDI-2025- 近年来，医学影像技术的迅猛发展推动了计算
Z-16） 机断层扫描术（computer tomography, CT）在临床
作者简介：薛张华，E-mail: shsxkyyzwk@163.com
通信作者：顾成云，E-mail: 297515348@qq.com 疾病检测与诊断中的广泛应用，作为一种重要的无

653