论文部分内容阅读
采用小波变换法和一种改进的基于状态机的极值对过零点检测算法在FPGA(现场可编程逻辑门阵列)硬件平台上实现了心电信号QRS波的检测和ST段特征点的提取,然后采用BP(反向传播)神经网络的方法在FPGA嵌入的软核niosII的软件平台上实现了对ST段的三种形态识别.通过小波变换和极值对过零点检测算法提取的ST段的数据作为BP神经网络的测试输入.用MIT-BIH数据库中的心电信号数据作为数据来源,选取Altera公司的FPGA开发板DE-2(EP2C35F672C6)作为验证平台,结果 R波峰检测和ST段的形态识别的正确率分别达到97.4%和96.8%,硬件资源共消耗4 309个逻辑单元,系统的整体延时时间为0.52s.结果表明该设计方案可以满足心电信号的实时分析.
Using wavelet transform and an improved state-based extreme value pair detection algorithm of zero-crossing to detect QRS wave and extract ST-segment feature points on FPGA (Field Programmable Logic Array) hardware platform, Then three kinds of morphological recognition of ST segment are implemented on the software platform of soft niosII embedded in FPGA by using BP (Back Propagation) neural network method.At the same time, the ST segment extracted by wavelet transform and extreme value zero crossing detection algorithm The data is used as the test input of BP neural network.The ECG data of MIT-BIH database is used as the data source, and the Altera FPGA development board DE-2 (EP2C35F672C6) is selected as the verification platform. The results of R wave detection and ST segment morphology The recognition accuracy rate is 97.4% and 96.8%, respectively, and the hardware resources consume a total of 4 309 logic elements, the overall delay time of the system is 0.52s. The results show that this design scheme can meet the real-time analysis of ECG signals.