Font Size: a A A

Design Of Sleep Pillow System Based On Sleep Detection

Posted on:2022-11-02Degree:MasterType:Thesis
Country:ChinaCandidate:P C WangFull Text:PDF
GTID:2481306749462744Subject:Photoelectric detection and control technology
Abstract/Summary:PDF Full Text Request
Sleep is vital to human health,and pillows play an important role in people's sleep.Studies have found that people's demands for pillow height vary widely between the supine and side sleeping positions.Inappropriate pillow height will affect the sleep quality,and even cause cervical spondylosis,headache,obstructive sleep apnea and other diseases.At the same time,the quality of sleep depends not only on the duration of sleep and the frequency of body movements,but also on the proportion of Light Sleep,Deep Sleep,and Rapid Eye Movement.Based on the above problems,the thesis designs a sleeping pillow system.At the end of the sleeping pillow,the sleeping position is detected in real time through pressure sensor,infrared photothermal radiation sensor and infrared photopyroelectric sensor.The height of the sleeping pillow is automatically adjusted by controlling the air pressure of the sleeping pillow air bag,supplemented by neck heating and music playing functions to relieve and help sleep.On the server side,the Attention-LSTM model that has been deployed and trained is saved.The sleep state is detected and classified through the sign information collected at the pillow end,and the sleep quality is analyzed according to the proportion of different sleep states.On the mobile terminal,users can view the sleep quality report through the front end of the web page,and use the We Chat applet to automatically and manually control and adjust the supine and lateral lying height of the pillow,neck heating,music playing,etc.Main content and innovations:1.Sleep quality detection.Based on the LSTM model,a two-layer attention mechanism is introduced to design the Attention-LSTM model to solve the problem of classifying the current sleep state and improve the accuracy of the model classification.The sleep quality was evaluated by the proportion of different sleep states in the whole sleep process.2.Sleeping posture detection.The movement of the head is detected by the infrared photothermal radiation sensor and the infrared photopyroelectric sensor;detecting the shoulder area through a plurality of longitudinal film strip pressure switch strip;the shoulder pressure value is detected by a transverse flexible film strip pressure sensor with a length of 40 cm.Based on the above three characteristics,a large number of sleeping postures are statistically classified from the perspective of whether the height of the sleeping pillow is comfortable,and the sleeping postures are divided into supine sleeping positions and side sleeping positions.When performing sleeping posture detection,output the sleeping posture result according to the classification corresponding to the eigenvalues,and the height of the sleeping pillow is automatically adjusted.3.According to the functional requirements,the system is built and tested.The test results show that the comprehensive accuracy of sleeping posture recognition and the classification model were 93.8% and 78.7%,respectively,and the system can run normally and stably,basically meeting the needs.
Keywords/Search Tags:Sleeping position detection, Sleep state, LSTM, Attention mechanism
PDF Full Text Request
Related items