声磁EAS系统设计及标签检测算法研究

发布时间：2018-03-21 12:09

本文选题：声磁EAS系统　切入点：标签识别算法　出处：《郑州大学》2017年硕士论文　论文类型：学位论文

【摘要】：电子商品防盗系统(Electronic Article Surveillance,EAS)是一种应用于现代零售业中的防盗设备,通过减少商品失窃,来提高销售利润。随着科学技术的发展,EAS系统的工作环境日趋复杂,各种电磁干扰严重影响了系统的检测性能。声磁技术为EAS系统的发展提供了一条新的道路,声磁EAS系统根据音叉原理而设计,其检测性能远超其他类型的产品。本文设计一种基于DSP的声磁EAS系统,并根据声磁标签信号特征研究了一种基于支持向量机的检测方法。本文所设计的系统检测器的核心处理器采用TMS320F28335,系统工作频率为58kHz,系统天线采用收发天线共用的时分复用一体式收发方式;系统通过引入50Hz的网络同步信号,可以有效地提高系统稳定性和抗干扰能力;发射信号通过整形电路、线性放大电路、功率放大电路处理后馈送至发射天线上,在天线周围形成了标签检测区域。标签信号一般都很微弱,通过放大电路、滤波电路、运算电路处理后输入DSP中进行AD转化和处理,处理器将分析接收信号特征以决定是否触发报警系统。针对传统声磁EAS系统标签识别算法导致系统检测率低和抗干扰性差等问题,本文研究了一种改进型人工鱼群算法(IAFSA)与支持向量机(SVM)相结合的声磁标签信号检测算法(IAFSA-SVM)。首先详细地分析了支持向量机和传统人工鱼群算法的原理和优势,根据算法的不足提出了优化方案,并建立了SVM检测模型;实验收集了足够的测试样本,提取并分析了标签信号和系统环境噪声的特征,综合考虑算法复杂度和标签信号的特点,选择了信号幅度、频率和信噪比作为支持向量机的输入向量,最后设计了实验仿真流程图。根据本文所提出的系统方案,设计了收发一体式声磁EAS系统和标签识别算法,对系统硬件和算法的性能进行了测试:系统硬件各模块工作正常,与预设参数相同;基于SVM的标签识别算法检测距离可达1.5m,检测率为99.75%,算法平均耗时4.6ms,可满足系统实时检测要求。
[Abstract]:Electronic Article Survey system (EAS) is a kind of anti-theft equipment used in modern retail industry, which can increase the sales profit by reducing the theft of goods. With the development of science and technology, the working environment of EAS system is becoming more and more complex. All kinds of electromagnetic interference seriously affect the detection performance of the system. Acoustic and magnetic technology provides a new way for the development of EAS system. The acoustic magnetic EAS system is designed according to the principle of tuning fork. Its detection performance is far superior to other types of products. In this paper, a sound magnetic EAS system based on DSP is designed. A detection method based on support vector machine (SVM) is studied according to the characteristics of acoustomagnetic label signal. The core processor of the system detector is TMS320F28335, the system frequency is 58kHz, and the antenna of the system is shared by transceiver antenna. Time division multiplexing integrated transceiver mode; By introducing 50Hz network synchronous signal, the system can effectively improve the stability and anti-jamming ability of the system, and the transmitting signal is fed to the transmitting antenna through shaping circuit, linear amplifier circuit and power amplifier circuit. The tag detection area is formed around the antenna. The tag signal is usually very weak. After processing the amplifier circuit, filter circuit and operation circuit, it is input into DSP for AD conversion and processing. The processor will analyze the characteristics of the received signal to determine whether to trigger the alarm system. The traditional acoustomagnetic EAS system label recognition algorithm leads to low detection rate and poor anti-jamming. In this paper, an improved artificial fish swarm algorithm (IAFSA) and support vector machine (SVM) are studied. Firstly, the principle and advantages of support vector machine and traditional artificial fish swarm algorithm are analyzed in detail. According to the deficiency of the algorithm, the optimization scheme is put forward, and the SVM detection model is established, and enough test samples are collected, the characteristics of tag signal and system environment noise are extracted and analyzed, and the complexity of the algorithm and the characteristics of label signal are considered synthetically. The signal amplitude, frequency and signal-to-noise ratio are selected as input vectors of support vector machine. Finally, an experimental simulation flow chart is designed. According to the proposed system, a transceiver acoustic-magnetic EAS system and label recognition algorithm are designed. The performance of the system hardware and algorithm is tested: the system hardware modules work normally and are the same as the preset parameters; The detection distance of tag recognition algorithm based on SVM is up to 1.5 m, the detection rate is 99.75 and the average time of the algorithm is 4.6 ms.It can meet the requirement of real-time detection of the system.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【相似文献】