CN-121445363-B - Sleep body movement monitoring method and system based on flexible sensor

CN121445363BCN 121445363 BCN121445363 BCN 121445363BCN-121445363-B

Abstract

The invention discloses a sleep body movement monitoring method and system based on a flexible sensor. And calculating time constants and steady-state pressure values of the sensor points by acquiring gravity strain signals of the flexible sensors on the surface of the mattress under different bearing conditions, determining time delay coefficients and pressure difference change coefficients according to the time constants and steady-state pressure values, and establishing a pressure delay compensation model. And compensating the real-time strain signal during monitoring to obtain a sleep strain differential sequence, constructing a pressure signal distribution change graph according to the sleep strain differential sequence, and repairing the missing signal to form a sleep body movement monitoring graph. By identifying the monitoring graph, the sleeping body movement state of the user can be accurately judged. The invention improves the body movement identification precision and the signal reliability of the flexible sensor in sleep monitoring, and is suitable for intelligent mattresses and sleep monitoring equipment.

Inventors

WU E
JIN YUAN
XIE YU

Assignees

爱梦睡眠(珠海)智能科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260106

Claims (6)

1. The sleep body movement monitoring method based on the flexible sensor is characterized by comprising the following steps of: Acquiring a gravity strain signal sequence of each sensing channel of flexible sensors arranged on the surface of the mattress in different weight bearing states, and analyzing and calculating a time constant and a steady-state pressure value thereof required by the pressure value of each sensor point in different weight pressure states to reach the steady state from the initial state according to the gravity strain signal sequence; Determining a time delay coefficient and a pressure difference change coefficient for each sensor point location to reach a steady-state pressure value based on the time constant and the steady-state pressure value under different weight pressure states, and determining a regional pressure delay compensation model of each sensor point location according to the time delay coefficient and the pressure difference change coefficient, wherein the regional pressure delay compensation model specifically comprises the following steps: Determining a time delay coefficient of each sensor point position from a sensed pressure value to a steady-state pressure value in different pressure states based on the pressure stabilizing time constant and the steady-state pressure value, and determining a pressure difference change coefficient between the measured pressure value of the sensor and the steady-state pressure value in a time period when the sensed pressure value reaches the steady-state pressure value; Based on pressure stability time constant and steady-state pressure value of each sensor channel under different weight pressures, extracting a pressure rising curve of each sensor point under different pressure states, and carrying out normalization processing on the pressure rising curve to obtain a normalized pressure rising curve; aiming at the normalized pressure rising curve of each sensor point position, introducing a Gaussian process regression algorithm, taking the applied pressure weight as an input variable, taking a pressure stability time constant and a steady-state pressure value as an observation target, setting a kernel function of the Gaussian process regression algorithm as a radial basis function, and calculating covariance matrixes of the pressure stability time constant and the steady-state pressure value under different pressure weights according to the radial basis function; constructing a Gaussian process regression model based on the covariance matrix, and optimizing the hyper-parameters of the Gaussian process regression model by using a maximum likelihood estimation method to obtain an optimized Gaussian process regression model; Inputting continuous pressure weight values into the optimized Gaussian process regression model, predicting a pressure stability time constant and a steady-state pressure value of each sensor point under any pressure weight, and calculating a time delay coefficient according to the predicted pressure stability time constant, wherein the time delay coefficient is the ratio of time required by the pressure value to reach the steady-state pressure value from an initial state to a standard time constant; Calculating a pressure difference change coefficient according to the predicted steady-state pressure value and the pressure rising curve, wherein the pressure difference change coefficient is the slope of the time change of the difference value between the instantaneous pressure value and the steady-state pressure value in the pressure rising process; Constructing a pressure delay compensation function of each sensor point location based on the time delay coefficient and the pressure difference change coefficient, taking the pressure delay compensation function as a regional pressure delay compensation model, wherein the input of the regional pressure delay compensation model is a pressure value acquired in real time and pressure application time, and the input of the regional pressure delay compensation model is a compensated pressure value; Acquiring real-time gravity strain signal data of each flexible sensor during sleeping of a user, and performing pressure delay compensation on the real-time gravity strain signal data according to the regional pressure delay compensation model to obtain a sleeping strain differential sequence; Constructing a pressure signal distribution change chart according to the sleep strain differential sequence, analyzing signal integrity of the pressure signal distribution change chart, and repairing incomplete pressure signals to obtain a sleep body movement monitoring chart; And identifying the sleeping body movement state of the user according to the sleeping body movement monitoring graph.
2. The sleep body movement monitoring method based on the flexible sensor according to claim 1, wherein the method is characterized in that the gravity strain signal sequences of the sensing channels of the flexible sensor arranged on the surface of the mattress are obtained under different weight bearing states, and the time constant and the steady-state pressure value thereof required for the pressure value of each sensor point to reach the steady state from the initial state under different weight pressure states are analyzed and calculated according to the gravity strain signal sequences, specifically: Acquiring gravity strain signals of each sensing channel of a flexible sensing array distributed on the surface of the mattress in different weight bearing states at a preset acquisition frequency to form an original gravity strain signal sequence of each sensor channel changing along with time; performing moving average filtering processing on the original gravity strain signal sequence of each sensing channel to obtain a smoothed gravity strain signal sequence, calculating a difference value of signal amplitude between adjacent sampling points of the gravity strain signal sequence of each sensing channel from a signal starting point, and setting a stability judging threshold value; And when absolute values of signal amplitude difference values in a plurality of continuous sampling periods are smaller than the stability judging threshold value, judging that the pressure value of the sensor channel reaches a stable state, and recording the time from the signal starting point to the stable state and the pressure value reaching the stable state to obtain the pressure stability time constant and the steady pressure value of each sensor channel under different weight pressures.
3. The sleep body movement monitoring method based on flexible sensors according to claim 1, wherein the method is characterized in that the method obtains real-time gravity strain signal data of each flexible sensor during the sleeping process of a user, performs pressure delay compensation on the real-time gravity strain signal data according to the area pressure delay compensation model, and obtains a sleep strain differential sequence, and specifically comprises the following steps: acquiring real-time gravity strain signal data of each flexible sensor during sleeping of a user, and extracting a predicted time delay coefficient and a pressure difference change coefficient corresponding to a current pressure value according to a regional pressure delay compensation model; Determining a pressure delay time window based on the predicted time delay coefficient, and performing pressure delay compensation on the real-time gravity strain signal data at each sampling point of the pressure delay time window according to a pressure compensation function corresponding to the pressure difference change coefficient in the time window; and carrying out space-time alignment on the compensated pressure sequences of the sensor points, calculating the pressure variation of the sensor points in adjacent sampling periods, and generating a sleep strain differential sequence.
4. The sleep body movement monitoring method based on the flexible sensor according to claim 1, wherein the step of constructing a pressure signal distribution change chart according to the sleep strain differential sequence, performing signal integrity analysis on the pressure signal distribution change chart, and repairing a malformed pressure signal to obtain a sleep body movement monitoring chart comprises the following specific steps: Constructing a pressure signal distribution change diagram of the sensor array at continuous time points according to the sleep strain differential sequence, and extracting signal amplitude variance and signal peak value of each sensor node in a preset time window based on the pressure signal distribution change diagram; comparing the signal amplitude variance and the signal peak value of each sensor node with a reference variance range and a reference peak value range under the historical normal working condition, and if the current signal amplitude variance of the sensor node is continuously lower than the reference variance range and the current signal peak value is continuously lower than the reference peak value range, judging that the sensor node is a suspected failure node and marking the space position of the sensor node; Modeling the flexible sensor array as a graph structure, wherein each sensor node is used as a graph node, edge connection is established between the sensor nodes which are adjacent in space, and the dynamic time regular distance between the sensor nodes and all adjacent normal nodes of the sensor nodes in the sleep strain differential sequence within the preset time window is calculated based on the marked suspected failure node positions; Determining signal association degree weights between suspected failure nodes and each adjacent normal node according to the dynamic time warping distance, and carrying out weighted fusion on sleep strain differential sequences of the adjacent normal nodes based on the signal association degree weights to generate a repair signal sequence of the suspected failure nodes; and replacing the repair signal sequence with the incomplete sleep strain differential sequence of the original suspected failure node, and updating the pressure signal distribution change graph to obtain a complete and continuous sleep body movement monitoring graph.
5. The sleep body movement monitoring method based on the flexible sensor according to claim 1, wherein the step of identifying the sleep body movement state of the user according to the sleep body movement monitoring chart is specifically as follows: extracting pressure distribution change data of a sensor array in a preset time window based on the sleep body movement monitoring graph, and calculating a movement track and a movement speed of the pressure center of gravity in the pressure distribution change data; Constructing a pressure gravity center motion feature vector according to the movement track and the movement speed of the pressure gravity center, extracting the area change rate and the shape change rate of a pressure distribution area in the sleep body motion monitoring graph, constructing a pressure distribution form feature vector, fusing the pressure gravity center motion feature vector and the pressure distribution form feature vector, and constructing a sleep body motion multidimensional feature vector; Obtaining standard body movement characteristic vectors under different body movement states in historical sleep body movement data, wherein the different body movement states comprise turning, limb movement, sitting up and body jogging, a body movement state classification model is constructed based on a support vector machine algorithm, and the standard body movement characteristic vectors are used as training samples to train the body movement state classification model; Inputting the sleep body movement multidimensional feature vector into a body movement state classification model after training to classify, and outputting the current sleep body movement state type of the user; and counting the occurrence frequency and duration of various body movement states in a preset time period according to the sleep body movement state types, evaluating the sleep body movement intensity level of the user based on the occurrence frequency and duration, and generating a sleep body movement state report according to the sleep body movement intensity level.
6. The sleep body movement monitoring system based on the flexible sensor is characterized by comprising a memory and a processor, wherein the memory comprises a sleep body movement monitoring method program based on the flexible sensor, and when the sleep body movement monitoring method program based on the flexible sensor is executed by the processor, the following steps are realized: Acquiring a gravity strain signal sequence of each sensing channel of flexible sensors arranged on the surface of the mattress in different weight bearing states, and analyzing and calculating a time constant and a steady-state pressure value thereof required by the pressure value of each sensor point in different weight pressure states to reach the steady state from the initial state according to the gravity strain signal sequence; Determining a time delay coefficient and a pressure difference change coefficient for each sensor point location to reach a steady-state pressure value based on the time constant and the steady-state pressure value under different weight pressure states, and determining a regional pressure delay compensation model of each sensor point location according to the time delay coefficient and the pressure difference change coefficient, wherein the regional pressure delay compensation model specifically comprises the following steps: Determining a time delay coefficient of each sensor point position from a sensed pressure value to a steady-state pressure value in different pressure states based on the pressure stabilizing time constant and the steady-state pressure value, and determining a pressure difference change coefficient between the measured pressure value of the sensor and the steady-state pressure value in a time period when the sensed pressure value reaches the steady-state pressure value; Based on pressure stability time constant and steady-state pressure value of each sensor channel under different weight pressures, extracting a pressure rising curve of each sensor point under different pressure states, and carrying out normalization processing on the pressure rising curve to obtain a normalized pressure rising curve; aiming at the normalized pressure rising curve of each sensor point position, introducing a Gaussian process regression algorithm, taking the applied pressure weight as an input variable, taking a pressure stability time constant and a steady-state pressure value as an observation target, setting a kernel function of the Gaussian process regression algorithm as a radial basis function, and calculating covariance matrixes of the pressure stability time constant and the steady-state pressure value under different pressure weights according to the radial basis function; constructing a Gaussian process regression model based on the covariance matrix, and optimizing the hyper-parameters of the Gaussian process regression model by using a maximum likelihood estimation method to obtain an optimized Gaussian process regression model; Inputting continuous pressure weight values into the optimized Gaussian process regression model, predicting a pressure stability time constant and a steady-state pressure value of each sensor point under any pressure weight, and calculating a time delay coefficient according to the predicted pressure stability time constant, wherein the time delay coefficient is the ratio of time required by the pressure value to reach the steady-state pressure value from an initial state to a standard time constant; Calculating a pressure difference change coefficient according to the predicted steady-state pressure value and the pressure rising curve, wherein the pressure difference change coefficient is the slope of the time change of the difference value between the instantaneous pressure value and the steady-state pressure value in the pressure rising process; Constructing a pressure delay compensation function of each sensor point location based on the time delay coefficient and the pressure difference change coefficient, taking the pressure delay compensation function as a regional pressure delay compensation model, wherein the input of the regional pressure delay compensation model is a pressure value acquired in real time and pressure application time, and the input of the regional pressure delay compensation model is a compensated pressure value; Acquiring real-time gravity strain signal data of each flexible sensor during sleeping of a user, and performing pressure delay compensation on the real-time gravity strain signal data according to the regional pressure delay compensation model to obtain a sleeping strain differential sequence; Constructing a pressure signal distribution change chart according to the sleep strain differential sequence, analyzing signal integrity of the pressure signal distribution change chart, and repairing incomplete pressure signals to obtain a sleep body movement monitoring chart; And identifying the sleeping body movement state of the user according to the sleeping body movement monitoring graph.

Description

Sleep body movement monitoring method and system based on flexible sensor Technical Field The invention relates to the technical field of sleep monitoring, in particular to a sleep body movement monitoring method and system based on a flexible sensor. Background With the acceleration of modern life pace, sleep quality problems are increasingly concerned. Sleep body movement is an important physiological index reflecting the sleep state of a human body, and the change of the sleep body movement can reflect the sleeping depth, the turnover frequency and the potential sleep disorder condition. The traditional sleep monitoring method mainly depends on Polysomnography (PSG) or wearable equipment, but has the problems of limited monitoring environment, inconvenient use, poor comfort, difficult continuous monitoring for a long time and the like. In recent years, the development of flexible sensor technology has provided new solutions for intelligent mattresses and sleep monitoring. By arranging the flexible sensor array on the surface of the mattress, the pressure distribution and body movement information of the user in the sleeping process can be acquired in a non-contact and continuous mode. However, due to differences in the elasticity of the mattress material, local non-uniformity in rebound performance, and differences in the physical characteristics of the sensor placement locations, there is a significant lag in the pressure response of each sensor point when the user lies down or changes sleeping positions, and the signal may undergo a dynamic transition phase from initial pressure to steady state pressure, resulting in time delays and amplitude deviations in the real-time acquisition of data. In addition, the flexible sensor may have local failure or abnormal signal during long-term use, so that the pressure distribution diagram is incomplete, and the continuity and accuracy of sleep body movement analysis are affected. Therefore, a method and a system for effectively correcting the pressure delay of a flexible sensor, repairing the incomplete signals and realizing high-precision and continuous monitoring of the sleep state of a user are needed to improve the reliability and the data availability of sleep monitoring and provide accurate support for sleep health assessment. Disclosure of Invention In order to solve at least one technical problem, the invention provides a sleep body movement monitoring method and system based on a flexible sensor. The first aspect of the invention provides a sleep body movement monitoring method based on a flexible sensor, which comprises the following steps: Acquiring a gravity strain signal sequence of each sensing channel of flexible sensors arranged on the surface of the mattress in different weight bearing states, and analyzing and calculating a time constant and a steady-state pressure value thereof required by the pressure value of each sensor point in different weight pressure states to reach the steady state from the initial state according to the gravity strain signal sequence; determining a time delay coefficient and a pressure difference change coefficient for each sensor point position to reach a steady-state pressure value based on the time constant and the steady-state pressure value under different weight pressure states, and determining a regional pressure delay compensation model of each sensor point position according to the time delay coefficient and the pressure difference change coefficient; Acquiring real-time gravity strain signal data of each flexible sensor during sleeping of a user, and performing pressure delay compensation on the real-time gravity strain signal data according to the regional pressure delay compensation model to obtain a sleeping strain differential sequence; Constructing a pressure signal distribution change chart according to the sleep strain differential sequence, analyzing signal integrity of the pressure signal distribution change chart, and repairing incomplete pressure signals to obtain a sleep body movement monitoring chart; And identifying the sleeping body movement state of the user according to the sleeping body movement monitoring graph. In this scheme, obtain the flexible sensor that is laid on mattress surface and put the weight strain signal sequence of each sensing channel under different weight bearing states, according to the analysis of weight strain signal sequence and calculate every sensor point position and be in the different weight pressure state pressure value from initial state reach steady state required time constant and steady state pressure value, specifically do: Acquiring gravity strain signals of each sensing channel of a flexible sensing array distributed on the surface of the mattress in different weight bearing states at a preset acquisition frequency to form an original gravity strain signal sequence of each sensor channel changing along with time; performing moving average filtering process