CN-122006060-A - Sleep-aiding music playing method and system based on multi-sensor fusion

CN122006060ACN 122006060 ACN122006060 ACN 122006060ACN-122006060-A

Abstract

The invention relates to the technical field of sleep management and discloses a sleep-aiding music playing method and system based on multi-sensor fusion, wherein the sleep-aiding music playing method comprises the steps of collecting sleep sound signals of a user and determining sound characteristic coefficients according to the sleep sound signals; the method comprises the steps of collecting sleep acceleration signals of a user, determining acceleration characteristic coefficients according to the sleep acceleration signals, collecting sleep eye movement data of the user, and determining sleep characteristics of the user according to the sleep eye movement data, the sound characteristic coefficients and the acceleration characteristic coefficients. The multi-dimensional feature extraction is carried out on the sleep signal of the user, so that more accurate sleep features of the user are extracted, and accurate recommendation of sleep-aiding music is realized.

Inventors

WANG WEI
WU HE

Assignees

北京广吉象达传媒集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260304

Claims (10)

1. A sleep-aiding music playing method based on multi-sensor fusion is characterized by comprising the following steps: Collecting a sleep sound signal of a user, and determining a sound characteristic coefficient according to the sleep sound signal; Collecting a sleep acceleration signal of a user, and determining an acceleration characteristic coefficient according to the sleep acceleration signal; collecting sleep eye movement data of a user, and determining sleep characteristics of the user according to the sleep eye movement data, the sound characteristic coefficient and the acceleration characteristic coefficient; And determining sleep quality parameters according to the sleep characteristics of the user, and playing sleep-aiding music of the user according to the sleep quality parameters.
2. The sleep-aiding music playing method based on multi-sensor fusion according to claim 1, wherein determining the sound characteristic coefficient from the sleep sound signal comprises: preprocessing the sleep sound signal, and performing spectrum analysis on the preprocessed sleep sound signal to obtain a sleep sound signal spectrogram; Determining a breathing period according to a sleep sound signal spectrogram, establishing a preset time window, calculating the standard deviation and the average value of the breathing period in the preset time window, and calculating the variation coefficient of the time window according to the standard deviation and the average value of the breathing period in the preset time window; Establishing a variation coefficient sequence according to the variation coefficient of each time window in time sequence, and calculating the degree of confusion of sleeping sounds according to the variation coefficient sequence; And determining the sound characteristic coefficient according to the degree of disorder and the variation coefficient of the sleeping sound.
3. The sleep-aiding music playing method based on multi-sensor fusion according to claim 2, wherein calculating the degree of confusion of sleep sounds according to the variation coefficient sequence comprises: Obtaining a preset neighborhood, and calculating the difference value between any variation coefficient in the variation coefficient sequence and the average value of the rest variation coefficients in the preset neighborhood; And counting the average value of the difference values of all the variation coefficients in the variation coefficient sequence and the average value of the rest variation coefficients in the preset neighborhood of the variation coefficient sequence to obtain the degree of disorder of sleeping sounds.
4. The sleep-aiding music playing method based on multi-sensor fusion according to claim 2, wherein determining the sound characteristic coefficient according to the degree of confusion and the variation coefficient of the sleep sound comprises: And counting the average variation coefficients of all the time windows, setting a correction weight according to the disordered degree of the sleeping sounds, and correcting the average variation coefficients according to the correction weight to obtain the sound characteristic coefficients.
5. The sleep-aiding music playing method based on multi-sensor fusion according to claim 1, wherein determining an acceleration characteristic coefficient according to a sleep acceleration signal comprises: extracting time domain features and frequency domain features of the sleep acceleration signals to obtain a sleep acceleration feature set; determining a body movement event based on the trained machine learning model and the sleep acceleration feature set; And counting the body movement type of each body movement event, extracting key body movement events according to the body movement type, and determining an acceleration characteristic coefficient according to the occurrence frequency of the key body movement events.
6. The sleep-aiding music playing method based on multi-sensor fusion according to claim 1, wherein determining the sleep characteristics of the user according to the sleep eye movement data, the sound characteristic coefficient and the acceleration characteristic coefficient comprises: Determining sleep characteristic parameters according to the sound characteristic coefficients and the acceleration characteristic coefficients, determining sleep stages according to the sleep eye movement data, and counting the variation trend of the sleep characteristic parameters of each sleep stage; And setting a feature tolerance threshold according to the sleep stages, and determining sleep features according to the sleep feature parameter change trend of all the sleep stages and the corresponding feature tolerance threshold.
7. The method for playing sleep-aiding music based on multi-sensor fusion according to claim 6, wherein determining sleep characteristics according to sleep characteristic parameter variation trends and corresponding characteristic tolerance thresholds of all sleep stages comprises: drawing a sleep characteristic parameter change curve according to the sleep characteristic parameter change trend, and performing curve fitting on the sleep characteristic parameter change curve to obtain a sleep characteristic parameter prediction curve; and predicting the time required for the sleep characteristic parameters to reach the corresponding characteristic tolerance threshold according to the sleep characteristic parameter prediction curve, and determining the sleep characteristic according to the time required for the sleep characteristic parameters of all the sleep characteristic parameter prediction curves to reach the corresponding characteristic tolerance threshold.
8. The method for playing sleep-aiding music based on multi-sensor fusion according to claim 7, wherein playing the user sleep-aiding music according to the sleep quality parameter comprises: acquiring a current sleep stage of a user, and determining a basic music category according to the current sleep stage; And determining a specific music list in the basic music category according to the sleep quality parameter, and determining the sleep-aiding music of the user according to the specific music list.
9. The sleep-aiding music playing method based on the multi-sensor fusion according to claim 8, wherein determining a specific music list in a basic music category according to sleep quality parameters comprises: Acquiring a sleep quality parameter change trend of a user during sleep-aiding music playing, and determining a music adaptation degree according to the sleep quality parameter change trend of the user during sleep-aiding music playing; Collecting the adaptation degree of the user to all the sleep aiding music, and sequencing the sleep aiding music according to the adaptation degree from high to low; and screening out the adaptive sleep aiding music with the rank less than the preset rank threshold according to the ranking result, and determining the adaptive sleep aiding music as a specific music list.
10. A sleep-aiding music playing system based on multi-sensor fusion, comprising: the sound module is used for collecting sleep sound signals of the user and determining sound characteristic coefficients according to the sleep sound signals; The acceleration module is used for collecting sleep acceleration signals of the user and determining acceleration characteristic coefficients according to the sleep acceleration signals; The fusion module is used for collecting sleep eye movement data of the user and determining sleep characteristics of the user according to the sleep eye movement data, the sound characteristic coefficient and the acceleration characteristic coefficient; and the playing module is used for determining sleep quality parameters according to the sleep characteristics of the user and playing the sleep-aiding music of the user according to the sleep quality parameters.

Description

Sleep-aiding music playing method and system based on multi-sensor fusion Technical Field The invention relates to the technical field of sleep management, in particular to a sleep-aiding music playing method and system based on multi-sensor fusion. Background Sleep is an indispensable physiological process of the human body, and its quality is directly related to physical and mental health, cognitive function and daytime behavioral state. The accurate extraction of sleep characteristics through sleep signals is a core premise of evaluating sleep quality parameters and recommending sleep-aiding music for users. The prior art mainly relies on manual analysis of Polysomnography (PSG) to identify sleep signals and extract features, and has the advantages of time and labor consumption, strong subjectivity, low feature extraction accuracy, lack of pertinence in recommending sleep-aiding music and difficulty in adapting to the sleep-aiding demands of users. Disclosure of Invention The invention provides a sleep-aiding music playing method and system based on multi-sensor fusion, which are used for improving the accuracy of sleep feature extraction of a user. In one aspect, the invention provides a sleep-aiding music playing method based on multi-sensor fusion, which comprises the following steps: The method comprises the steps of collecting sleep sound signals of a user, determining sound characteristic coefficients according to the sleep sound signals, collecting sleep acceleration signals of the user, determining acceleration characteristic coefficients according to the sleep acceleration signals, collecting sleep eye movement data of the user, determining sleep characteristics of the user according to the sleep eye movement data, the sound characteristic coefficients and the acceleration characteristic coefficients, determining sleep quality parameters according to the sleep characteristics of the user, and playing sleep-aiding music of the user according to the sleep quality parameters. The method comprises the steps of determining a sound characteristic coefficient according to a sleep sound signal, preprocessing the sleep sound signal, carrying out frequency spectrum analysis on the preprocessed sleep sound signal to obtain a sleep sound signal spectrogram, determining a breathing period according to the sleep sound signal spectrogram, establishing a preset time window, calculating standard deviation and average value of the breathing period in the preset time window, calculating variation coefficients of the time window according to the standard deviation and average value of the breathing period in the preset time window, establishing a variation coefficient sequence according to the variation coefficients of the time windows in a time sequence, calculating the degree of confusion of the sleep sound according to the variation coefficient sequence, and determining the sound characteristic coefficient according to the degree of confusion of the sleep sound and the variation coefficients. Further, the method for calculating the sleep sound confusion degree according to the variation coefficient sequence comprises the steps of obtaining a preset neighborhood, calculating the difference value of any variation coefficient in the variation coefficient sequence and the average value of the rest variation coefficients in the preset neighborhood, and calculating the average value of the difference values of all variation coefficients in the variation coefficient sequence and the average value of the rest variation coefficients in the preset neighborhood to obtain the sleep sound confusion degree. Further, determining the sound characteristic coefficient according to the sleep sound confusion degree and the variation coefficient comprises the steps of counting the average variation coefficient of all time windows, setting a correction weight according to the sleep sound confusion degree, and correcting the average variation coefficient according to the correction weight to obtain the sound characteristic coefficient. The method comprises the steps of determining an acceleration characteristic coefficient according to a sleep acceleration signal, extracting time domain characteristics and frequency domain characteristics of the sleep acceleration signal to obtain a sleep acceleration characteristic set, determining body movement events based on a trained machine learning model and the sleep acceleration characteristic set, counting body movement types of all body movement events, extracting key body movement events according to the body movement types, and determining the acceleration characteristic coefficient according to the occurrence frequency of the key body movement events. Further, determining the sleep characteristics of the user according to the sleep eye movement data, the sound characteristic coefficient and the acceleration characteristic coefficient comprises determining sleep characteristic parameters