基于音频识别的血透机泵状态预警系统设计

目的设计一种基于音频识别的预警系统,用以监测血透机泵运行状态,为血透机的稳定工作提供保障。方法首先通过麦克风采集泵在故障和正常工作状态下的音频数据;然后对其进行快速傅里叶变换,分别从时域和频域对两种状态的数据进行对比分析,找出特征差异,设计相关判断算法;最后通过实验来验证判断泵故障与否条件的准确性。结果选取了8个血透机泵进行测试,测试结果与实际情况一致,准确地判断出3个泵处于正常状态,其余5个泵故障或存在故障风险,且该方法对同一对象不同声音样本的检测具有稳定性。结论本文设计的预警系统,利用音频识别实现了对血透机泵状态的实时监测,便于临床工程人员管理血透设备,具有较高的推广使用价值。

Design of Early Warning System for Hemodialysis Pump State Based on Audio Recognition

Objective An early warning system based on audio recognition is designed to monitor the running state of the hemodialysis pump,so as to ensure the stable operation of the hemodialysis machine.Methods Firstly,the microphone was used to collect the audio data of the pump under fault and normal working conditions.Then,the fast Fourier transform was carried out to compare and analyze the data of the two states from the time domain and the frequency domain respectively,in order to find out the characteristic differences and design the relevant judgment algorithm.Finally,the accuracy of the condition to judge whether the pump failure was verified by experiments.Results Eight hemodialysis machine pumps were selected for testing,and the test results were consistent with the actual situation.It was accurately judged that three pumps were in normal state,and the remaining five pumps were in failure or risk of failure,and the method was stable for the detection of different sound samples of the same object.Conclusion The early warning system designed in this paper realizes the real-time monitoring of the status of hemodialysis pump by using audio recognition,which is convenient for clinical engineers to manage hemodialysis equipment,and has high practical value.