| Sleep Apnea Hypopnea Syndrome has a high prevalence with regard to other sleep disorders,which will not only affect the normal study and life,but also induce arterial hypertension and other chronic cardiovascular disease.Because the current sleep detection means are not only time-consuming and laborious,the wear of a large number of sensors will also seriously affect the use comfort.Therefore,the use of non-contact radar system for sleep monitoring has a very important value.Based on the deep study of radar signal processing and sleep detection technology,this paper focuses on the study of the SAHS detection method and Inter Beat interval(IBI)extraction algorithm based on radar signal,the main works are as follows:(1)In view of the problem that the existing sleep monitoring system will affect the use of comfort,the non-contact sleep monitoring technology based on radar signal has been studied in this paper.Some important radar parameters has been well discussed and determined while taking into account the system requirement of monitoring the weak signal of respiratory and heartbeat during sleep,and then a complete sleep detection system and a SAHS detection software has been designed.(2)Traditional SAHS detection methods based on thresholds need to take into account the physiological signals such as respiratory signals and oxygen saturation,while the radar system can only obtain the respiratory signal,and with a single physiological signal using the threshold based method will often cause great error.In order to solve this problem,this paper firstly analyzes the detection performance based on the threshold discriminant method,and then carries on the simulation experiment in WEKA software and verifies that the method based on the threshold discrimination method is applicable to the system,and the SAHS detection scheme using the machine learning method is determined.Then a scheme based on support vector machine(SVM)is proposed.In addition,this paper improves the parameter optimization process of SVM.Through the experimental comparison,the double base algorithm is better than the original algorithm,and the classification speed can be improved a lot.(3)In order to solve the problem that the PQRST wave in the heartbeat signal cannot be accurately detected by the radar system,the traditional peak detection algorithm is no longer applicable and considering the fact that the heartbeat signal has different pattern for different person and varying with time,the dynamic time wrapping algorithm which is widely used in audio signal processing has been learned to segment the heartbeat waveform and an IBI extraction algorithm based on dynamic time wrapping has been proposed in this paper.The experimental results show that the average absolute error of the algorithm is 0.134 s.Using IBI as a feature in the SVM based SAHS detection scheme can improve the accuracy.(4)The SAHS detection hardware system has been set up and SAHS detection software are developed,with the IBI extraction algorithm and the improved SVM based SAHS detection algorithm are applied to the software.Using the system to monitor the SAHS patients to conduct sleep monitoring experiments.Compared with the PSG test results,the system can achieve more than 80% of the detection accuracy,which meets the system requirements indicators. |
PDF Full Text Request