Sleep monitoring method and system based on millimeter wave radar and pressure sensor
By combining pressure sensors and millimeter-wave radar sensors, deep learning algorithms are used to accurately divide the sleep stages, solving the shortcomings of millimeter-wave radar in positioning accuracy and micro-physiological activity detection, and achieving efficient and accurate sleep monitoring, suitable for Home and medical environment.
Patent Information
- Application Number
- CN202411008671.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-26
- Publication Date
- 2024-11-15
AI Technical Summary
Existing millimeter-wave radar sleep monitoring technology may not maintain monitoring accuracy when user location changes and there are shortcomings in detection of micro-physiological activity.
Combining pressure sensors and millimeter-wave radar sensors, the sleep phase is accurately divided through deep learning algorithms, the data input ratio is adjusted, and the overall reliability and accuracy of sleep monitoring is improved.
It significantly improves the accuracy and comprehensiveness of sleep monitoring, provides deeper sleep quality analysis, adapts to specific sleep patterns of different users, provides personalized sleep improvement suggestions, and maintains user comfort and natural sleep.
Smart Images

Figure CN118965265A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of sleep monitoring, and in particular to a sleep monitoring method and system based on millimeter wave radar and a pressure sensor. Background Art
[0002] In the field of sleep monitoring, traditional methods rely on contact sensors. Although these methods are more accurate, they affect the naturalness of sleep. In recent years, non-contact sleep monitoring technology has developed rapidly, among which millimeter-wave radar has attracted much attention due to the advantages of non-contact and continuous monitoring it provides. However, existing millimeter-wave radar monitoring technology may not be able to maintain monitoring accuracy when the user's position changes. In order to solve the above problems, the present invention combines the method of pressure sensors to make up for the shortcomings of millimeter-wave radar in positioning accuracy and detection of small physiological activities. The pressure sensor can provide pressure change information at different positions on the bed, help determine the user's specific sleeping posture and movement, thereby complementing the data of the millimeter-wave radar to achieve a more comprehensive and accurate sleep quality assessment. Summary of the invention
[0003] In view of this, the present invention provides a sleep monitoring method and system based on millimeter-wave radar and pressure sensor, aiming to provide an efficient and accurate sleep monitoring means, by combining the data of millimeter-wave radar sensor and pressure sensor and using deep learning algorithm to accurately divide the sleep stages, thereby improving the overall reliability and accuracy of sleep monitoring.
[0004] To achieve the above objectives, the present invention provides a sleep monitoring method and system based on millimeter wave radar and pressure sensor, comprising the following steps:
[0005] S1: Collects data from millimeter-wave radar sensors deployed on the ceiling and pressure data from pressure sensors arrayed under the mattress;
[0006] S2: preprocessing the collected millimeter-wave radar sensor data and pressure data to obtain preprocessed data, and fusing the millimeter-wave radar sensor data and the pressure data at an initial ratio of 1:1 to obtain fused data;
[0007] S3: Input the fused data into the SleepStage-CNN algorithm to obtain the probability of sleep stage division;
[0008] S4: calculating an entropy value according to the probability of the sleep stage division, and adjusting an input ratio of millimeter wave radar sensor data to pressure data;
[0009] S5: Re-input the adjusted input ratio into the SleepStage-CNN algorithm to obtain a sleep stage division result.
[0010] As a further improvement method of the present invention:
[0011] Optionally, the step S1 collects the data of the millimeter wave radar sensor deployed on the ceiling and the pressure data of the pressure sensor array deployed under the mattress, including:
[0012] The millimeter wave radar sensor can capture subtle body movements such as breathing and heartbeat, while the pressure sensor array accurately monitors the pressure distribution of the body on the mattress. The combination of these two sensors can non-invasively monitor various physiological and behavioral parameters during sleep, thereby providing data support for high-precision sleep quality analysis and evaluation, greatly enhancing the practicality and accuracy of the monitoring system.
[0013] Optionally, in step S2, preprocessing the collected millimeter wave radar sensor data and pressure data to obtain preprocessed data, and fusing the millimeter wave radar sensor data and pressure data at an initial ratio of 1:1 to obtain fused data, including:
[0014] S21: De-noising the millimeter-wave radar data:
[0015] The millimeter wave radar sensor collects the heart rate and the respiratory rate, removes the low frequency noise through the IIR high pass filter, retains the high frequency components in the heart rate and the respiratory rate, and obtains the millimeter wave radar sensor data O1 after denoising;
[0016] This step strengthens the high-frequency characteristics of heartbeat and breathing, making the key characteristics of these physiological signals prominent, improving the accuracy and reliability of the data, making the data more reliable, enhancing the identifiability of the signal, and reducing the subsequent computational burden;
[0017] S22: After the millimeter wave radar sensor data is denoised, a Z-score normalization method is used to perform normalization processing to obtain converted normalized millimeter wave radar sensor data O1';
[0018] This step not only makes the data more normalized, but also has a certain inhibitory effect on extreme values and outliers in the data, because it converts the data through the mean and standard deviation, thereby reducing the impact of extreme values on the overall data analysis;
[0019] S23: performing spatial normalization on the pressure data O2 using a minimum-maximum normalization method to obtain normalized pressure data O2′;
[0020] This step adjusts the pressure data to the range of 0 to 1, achieves consistency in data scale, enhances the model's sensitivity to data, and accelerates the convergence of the algorithm;
[0021] S24: reconstructing the pressure data O2' into a two-dimensional image format according to the pressure sensor array layout to obtain pressure data O2";
[0022] This step reconstructs the pressure data into a format consistent with the millimeter-wave radar sensor data, simplifying the subsequent data fusion process;
[0023] S25: The millimeter wave radar sensor data O1' and the pressure data O2" are fused at an initial ratio of 1:1 to obtain fused data R:
[0024]
[0025] This step enhances the comprehensiveness and multi-dimensional information of the data and reduces the errors and biases that may arise from relying on a single data source.
[0026] Optionally, in step S3, the fused data is input into a SleepStage-CNN algorithm to obtain a sleep stage division probability, including:
[0027] S31: Input the fused data R into the convolution layer of the SleepStage-CNN algorithm to obtain a feature vector;
[0028] S32: Performing maximum pooling layer processing on the feature vector to obtain pooling features;
[0029] S33: The pooled features are processed by a fully connected layer to obtain initial sleep stage division probabilities P1, P2, P3, and P4, which correspond to the awake, light sleep, deep sleep, and REM sleep stages, respectively, as follows:
[0030]
[0031] Among them, α1~α4 are the weights corresponding to each sleep stage, β1~β4 are the corresponding biases, and xy α·H xy +βAfter linear transformation, the probability of sleep stage division is obtained through softmax function;
[0032] Step S3 adaptively identifies key features from complex fused data, reducing human intervention and subjective errors while improving the speed and efficiency of data processing.
[0033] Optionally, in step S4, calculating the entropy value according to the sleep stage division probability and adjusting the input ratio of the millimeter wave radar sensor data to the pressure data include:
[0034] S41: Calculate the entropy value T according to the initial sleep stage division probability:
[0035] T=-P1ln(P1)-P2ln(P2)-P3ln(P3)-P4ln(P4)
[0036] This step quantifies the uncertainty of the sleep stage classification. This metric helps identify how well the current model classifies sleep stages and whether further adjustments are needed.
[0037] S42: Adjust the input ratio K of the millimeter wave radar sensor data and the pressure data, specifically:
[0038]
[0039] Among them, T max 、T min and T target are the maximum, minimum and target values of the entropy value respectively; the adjustment coefficient v is the normalized deviation between the current entropy value and the target entropy value, which is obtained after being processed by the logarithmic function and reflects the magnitude of the deviation in a nonlinear way; the intermediate variable u is v multiplied by a reverse normalization proportional factor to bring the value of u back to the original range;
[0040] This step optimizes the classification accuracy by changing the data input ratio, using the logarithmic function processing and the method of reversing the normalization scale factor to fine-tune the data weight in a nonlinear manner, which can more effectively cope with a wide range of entropy value changes and ensure the adaptability and flexibility of the system;
[0041] Optionally, in step S5, the adjusted input ratio is re-input into the SleepStage-CNN algorithm to obtain a sleep stage division result, including:
[0042] S51: adjusting the input ratio K of the millimeter wave radar sensor data and the pressure data to obtain adjusted fusion data R':
[0043] R′=K×O1′+(1-K)O2″
[0044] S52: Re-input the adjusted fusion data R' into the SleepStage-CNN algorithm, and repeat steps S41-S42 until the entropy value T is within the target range [T target -Δ,T target +Δ], Δ is the preset tolerance;
[0045] S53: When the entropy value T reaches the target range, the SleepStage-CNN algorithm outputs the final sleep stage division probabilities P'1, P'2, P'3, and P'4, which correspond to the awake, light sleep, deep sleep, and REM sleep stages, respectively. The sleep stage division result A is the sleep stage corresponding to the maximum value of the four probabilities, specifically:
[0046] A=max(P1′,P2′,P3′,P4′)
[0047] Step S5 enhances the adaptive ability of the SleepStage-CNN algorithm by continuously adjusting the input data ratio and feedback iterative optimization, so that the algorithm can continuously optimize its parameters according to the performance of actual data and improve the accuracy of classification. The present invention also provides a sleep monitoring system based on millimeter wave radar and pressure sensor, including:
[0048] Data collection module: used to obtain data from the millimeter wave radar module and the pressure sensor array to ensure that sufficient physiological signals are obtained for subsequent processing;
[0049] Data processing module: used to pre-process the collected data and fuse the millimeter-wave radar sensor data and pressure data in an initial ratio of 1:1 to obtain fused data;
[0050] Algorithm module: fusion data input to SleepStage-CNN algorithm;
[0051] Feedback module: Calculates the entropy value based on the output of the algorithm module, dynamically adjusts the input ratio of the millimeter-wave radar sensor data and the pressure data, and inputs it into the SleepStage-CNN algorithm again. This process is repeated until the entropy value reaches the preset target range;
[0052] Sleep monitoring module: When the entropy value of the feedback module reaches the target range, the final sleep stage division result is output;
[0053] The beneficial effects of the present invention are:
[0054] The present invention proposes a sleep monitoring method and system based on millimeter wave radar and pressure sensor. By combining millimeter wave radar and pressure sensor, compared with the prior art, the millimeter wave radar makes up for the shortcomings of capturing the precise position and small-scale movements on the bed, thereby significantly improving the overall monitoring accuracy of the system. By combining two complementary sensing technologies, the present invention not only improves the accuracy of monitoring data, but also can provide a deeper sleep quality analysis by analyzing richer and more multidimensional data. This comprehensive data fusion enables the present invention to adapt to the specific sleep patterns of different users and provide personalized sleep improvement suggestions. The non-contact monitoring method of the present invention not only maintains the user's comfort and natural sleep, but also ensures the high reliability and accuracy of the monitoring results through precise data fusion technology. It is not only suitable for home environments, but also suitable for use in medical and nursing facilities, enhancing its market applicability. In addition, the SleepStage-CNN algorithm adopted in combination with a real-time feedback mechanism can adjust the data input ratio according to the entropy value, realize real-time adaptation to changes in sleep patterns, and further improve the flexibility of the algorithm and the personalized accuracy of the monitoring results.
[0055] In summary, the present invention significantly improves the performance of sleep monitoring technology, overcomes the limitations of traditional technology in terms of contact, accuracy, and comprehensiveness, and provides users and medical providers with a more effective and comfortable sleep monitoring solution. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] Figure 1 The present invention is a flowchart of a sleep monitoring method and system based on millimeter wave radar and pressure sensor according to an embodiment of the present invention. DETAILED DESCRIPTION
[0057] The present invention is further described below in conjunction with the accompanying drawings, but the present invention is not limited in any way. Any changes or substitutions made based on the teachings of the present invention belong to the protection scope of the present invention.
[0058] Embodiment 1: In order to achieve the above-mentioned object, the present invention provides a sleep monitoring method and system based on millimeter wave radar and pressure sensor, such as Figure 1 The following steps are shown:
[0059] S1: Collecting data from the millimeter wave radar sensor deployed on the ceiling and the pressure data from the pressure sensor array deployed under the mattress:
[0060] The millimeter wave radar sensor can capture subtle body movements such as breathing and heartbeat, while the pressure sensor array accurately monitors the pressure distribution of the body on the mattress. The use of millimeter wave radar sensors or pressure sensor arrays for non-contact sleep monitoring has its own limitations, which may affect the accuracy and comprehensiveness of sleep monitoring. Although the use of millimeter wave radar alone can effectively monitor parameters such as breathing and heart rate, it is less sensitive to capturing body position changes and precise body position. Although the use of pressure sensors alone can provide data about body position and pressure, it is not as effective as millimeter wave radar sensors in monitoring tiny physiological activities such as breathing or heartbeat. In addition, if the pressure sensor coverage is uneven or the sleeper leaves the area covered by the pressure sensor, the accuracy and integrity of the data may also be affected. Therefore, the combination of these two sensors can non-invasively monitor various physiological and behavioral parameters during sleep, thereby providing data support for high-precision sleep quality analysis and evaluation, greatly enhancing the practicality and accuracy of the monitoring system. This embodiment uses a 24GHz millimeter wave radar sensor, and a 64×64 grid pressure sensor array is arranged under the mattress.
[0061] S2: preprocessing the collected millimeter-wave radar sensor data and pressure data to obtain preprocessed data, and fusing the millimeter-wave radar sensor data and pressure data at an initial ratio of 1:1 to obtain fused data:
[0062] S21: De-noising the millimeter-wave radar data:
[0063] The millimeter wave radar sensor collects the heart rate and the breathing rate, removes the low frequency noise through the IIR high pass filter, retains the high frequency components in the heart rate and the breathing rate, and obtains the millimeter wave radar sensor data O1 after denoising; in this embodiment, the sampling frequency is set to 100 Hz and the cutoff frequency is set to 0.5 Hz. This is because the important frequency components of the heart rate and the breathing rate are usually higher than this frequency, which is suitable for monitoring physiological changes during sleep;
[0064] S22: After the millimeter wave radar sensor data is denoised, it is normalized using a Z-score normalization method and then converted into a form with zero mean and unit variance to obtain converted standardized millimeter wave radar sensor data O1';
[0065] S23: performing spatial normalization on the pressure data O2 using a minimum-maximum normalization method to obtain normalized pressure data O2'; the sampling frequency of the pressure sensor array in this embodiment is set to 100 Hz, which is the same as that of the millimeter wave radar, to ensure time synchronization of the two sensor data, which is convenient for subsequent data fusion processing;
[0066] S24: reconstructing the pressure data O2' into a two-dimensional image format according to the pressure sensor array layout to obtain pressure data O2". In this embodiment, a complete 64×64 matrix is constructed. This method intuitively represents the pressure distribution of the sleeper on the mattress, and provides an ideal data format for further feature analysis and sleep stage recognition using the SleepStage-CNN algorithm.
[0067] S25: The millimeter wave radar sensor data O1' and the pressure data O2" are fused at an initial ratio of 1:1 to obtain fused data R:
[0068]
[0069] S3: Input the fused data into the SleepStage-CNN algorithm to obtain the probability of sleep stage division:
[0070] S31: Input the fused data R into the convolution layer of the SleepStage-CNN algorithm to obtain a feature vector; this embodiment uses two convolution layers, two convolution layers, the first layer uses 32 convolution kernels, the second layer uses 64 convolution kernels, the convolution kernel size is 3×3, the step length r is 1, and no padding is used; the weight initialization adopts the He normal distribution method, and the bias is initialized to 0;
[0071] S33: performing maximum pooling layer processing on the feature vector to obtain pooled features; in this embodiment, two maximum pooling layers are used, and each layer uses a 2×2 pooling window;
[0072] S34: The pooled features are processed by a fully connected layer to obtain initial sleep stage division probabilities P1, P2, P3, and P4, which correspond to the awake, light sleep, deep sleep, and REM sleep stages, respectively, specifically:
[0073]
[0074] Among them, α1~α4 are the weights corresponding to each sleep stage, β1~β4 are the corresponding biases, and xy α·H xy +β linear transformation is performed and the probability of sleep stage division is obtained by softmax function; in this embodiment, the weight initialization adopts He normal distribution method, the bias is initialized to 0, two fully connected layers are used, the first fully connected layer has 1024 neurons, and the second fully connected layer uses softmax activation function to output the probabilities of four sleep stages (awake, light sleep, deep sleep, REM sleep);
[0075] In the SleepStage-CNN algorithm of this embodiment, the optimizer adopts Adam, the learning rate is set to 0.001, and the decay rate is set to 1e-6; the loss function is set to cross entropy loss for multi-classification problems; this model can effectively learn and extract useful features from the fused sensor data to accurately analyze and predict sleep quality and stages.
[0076] S4: Calculate the entropy value according to the probability of the sleep stage division, and adjust the input ratio of the millimeter wave radar sensor data and the pressure data:
[0077] S41: Calculate the entropy value T according to the initial sleep stage division probability:
[0078] T=-P1ln(P1)-P2ln(P2)-P3ln(P3)-P4ln(P4)
[0079] S42: Adjust the input ratio K of the millimeter wave radar sensor data and the pressure data, specifically:
[0080]
[0081] Among them, T max , T min and T target are the maximum value, minimum value and target value of the entropy value, v is the adjustment coefficient, and u is the intermediate variable; T of this embodiment max , T min and T target They are 1.2, 0.5 and 0.85 respectively;
[0082] S5: Re-input the adjusted input ratio into the SleepStage-CNN algorithm to obtain the sleep stage division result:
[0083] S51: adjusting the input ratio K of the millimeter wave radar sensor data and the pressure data to obtain adjusted fusion data R':
[0084] R′=K×O1′+(1-K)O2″
[0085] S52: Re-input the adjusted fusion data R' into the SleepStage-CNN algorithm, and repeat steps S41-S42 until the entropy value T is within the target range [T target -Δ,T target +Δ], Δ is a preset tolerance; Δ in this embodiment is 0.13;
[0086] S53: When the entropy value T reaches the target range, the SleepStage-CNN algorithm outputs the final sleep stage division probabilities P'1, P'2, P'3, and P'4, which correspond to the awake, light sleep, deep sleep, and REM sleep stages, respectively. The sleep stage division result A is the sleep stage corresponding to the maximum value of the four probabilities, specifically:
[0087] A=max(P1′,P2′,P3′,P4′)
[0088] Embodiment 2:
[0089] The present invention also provides a sleep monitoring system based on millimeter wave radar and pressure sensor, the system comprising:
[0090] Data collection module: used to obtain data from the millimeter wave radar module and the pressure sensor array to ensure that sufficient physiological signals are obtained for subsequent processing;
[0091] Data processing module: used to pre-process the collected data and fuse the millimeter-wave radar sensor data and pressure data in an initial ratio of 1:1 to obtain fused data;
[0092] Algorithm module: fusion data input to SleepStage-CNN algorithm;
[0093] Feedback module: Calculates the entropy value based on the output of the algorithm module, dynamically adjusts the input ratio of the millimeter-wave radar sensor data and the pressure data, and inputs it into the SleepStage-CNN algorithm again. This process is repeated until the entropy value reaches the preset target range;
[0094] Sleep monitoring module: When the entropy value of the feedback module reaches the target range, the final sleep stage division result is output;
[0095] In detail, each module in the sleep monitoring system based on millimeter wave radar and pressure sensor in the embodiment of the present invention is used in the same manner as above. Figure 1 The sleep monitoring method based on millimeter wave radar and pressure sensor described in the invention has the same technical means and can produce the same technical effects, so it will not be repeated here.
[0096] It should be understood that the embodiments are for illustration purposes only and are not limited in the scope of the patent application. In particular, the values of ambient temperature, humidity, relative wind speed and atmospheric pressure are only used as examples and do not serve as specific threshold limits for the present invention.
[0097] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.
[0098] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.
[0099] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.
[0100] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A sleep monitoring method based on millimeter wave radar and pressure sensor, It is characterized in that The following steps are involved: S1: Collects data from millimeter-wave radar sensors deployed on the ceiling and pressure data from pressure sensors arrayed under the mattress; S2: preprocessing the collected millimeter-wave radar sensor data and pressure data to obtain preprocessed data, and fusing the millimeter-wave radar sensor data and the pressure data at an initial ratio of 1:1 to obtain fused data; S3: Input the fused data into the SleepStage-CNN algorithm to obtain the probability of sleep stage division; S4: calculating an entropy value according to the probability of the sleep stage division, and adjusting an input ratio of millimeter wave radar sensor data to pressure data; S5: Re-input the adjusted input ratio into the SleepStage-CNN algorithm to obtain a sleep stage division result.
2. The sleep monitoring method based on millimeter wave radar and pressure sensor according to claim 1, It is characterized in that The step S2 comprises: S21: De-noising the millimeter-wave radar data: The millimeter wave radar sensor collects heart rate and respiratory rate, removes low frequency noise through an IIR high pass filter, retains high frequency components in the heart rate and respiratory rate, and obtains millimeter wave radar sensor data after denoising. 1 ; S22: After the millimeter wave radar sensor data is denoised, the Z-score normalization method is used to perform normalization processing to obtain converted standardized millimeter wave radar sensor data. 1 '; S23: The pressure data O 2 Use the minimum-maximum normalization method to perform spatial normalization to obtain the normalized pressure data O 2 '; S24: The pressure data O 2 'According to the layout of the pressure sensor array, reconstruct it into a two-dimensional image format to obtain pressure data. 2 ”; S25: The millimeter wave radar sensor data O 1 'With pressure data O 2 "According to the initial ratio of 1:1, the fused data R is obtained:
3. The sleep monitoring method based on millimeter wave radar and pressure sensor according to claim 2, It is characterized in that The step S3 comprises: S31: Input the fused data R into the convolution layer of the SleepStage-CNN algorithm to obtain a feature vector; S32: Performing maximum pooling layer processing on the feature vector to obtain pooling features; S33: Process the pooled features through a fully connected layer to obtain the sleep stage division probability P 1 , P 2 , P 3 , P 4 , corresponding to the wakefulness, light sleep, deep sleep, and REM sleep stages, specifically: Among them, α 1 ~α 4 is the weight sum corresponding to each sleep stage, β 1 ~β 4 is the corresponding bias, for H xy α·H xy +βAfter linear transformation, the probability of sleep stage division is obtained through softmax function.
4. The sleep monitoring method based on millimeter wave radar and pressure sensor according to claim 3, It is characterized in that The step S4 includes: S41: Calculate the entropy value T according to the sleep stage division probability: T=-P 1 ln(P 1 )-P 2 ln(P 2 )-P 3 ln(P 3 )-P 4 ln(P 4 ) S42: Adjust the input ratio K of the millimeter wave radar sensor data and the pressure data, specifically: Among them, T max , T min and T target are the maximum, minimum and target values of the entropy value, v is the adjustment coefficient, and u is the intermediate variable.
5. The sleep monitoring method based on millimeter wave radar and pressure sensor according to claim 4, It is characterized in that The step S5 includes: S51: adjusting the input ratio K of the millimeter wave radar sensor data and the pressure data to obtain adjusted fusion data R': R′=K×O 1 ′+(1-K)O 2 ″ S52: Re-input the adjusted fusion data R' into the SleepStage-CNN algorithm, and repeat steps S41-S42 until the entropy value T is within the target range [T target -Δ,T target +Δ], Δ is the preset tolerance; S53: When the entropy value T reaches the target range, the SleepStage-CNN algorithm outputs the final sleep stage division probability P' 1 , P' 2 , P' 3 , P' 4 , corresponding to the awake, light sleep, deep sleep, and REM sleep stages respectively. The sleep stage division result A is the sleep stage corresponding to the maximum value of the four probabilities, specifically: A=max(P 1 ′,P 2 ′,P 3 ′,P 4 ′)。 6. A sleep monitoring system based on millimeter wave radar and pressure sensor, It is characterized in that include: Data collection module: used to obtain data from the millimeter wave radar module and the pressure sensor array; Data processing module: used to pre-process the collected data and fuse the millimeter-wave radar sensor data and pressure data in an initial ratio of 1:1 to obtain fused data; Algorithm module: fusion data input to SleepStage-CNN algorithm; Feedback module: Calculates the entropy value based on the output of the algorithm module, dynamically adjusts the input ratio of the millimeter-wave radar sensor data and the pressure data, and inputs it into the SleepStage-CNN algorithm again. This process is repeated until the entropy value reaches the preset target range; Sleep monitoring module: When the entropy value of the feedback module reaches the target range, the final sleep stage division result is output; To realize a sleep monitoring method based on millimeter wave radar and pressure sensor as described in any one of claims 1-5.
Citation Information
Cited By
Infant sign monitoring system
CN119655723A
Sleep monitoring device and method based on optical fiber Michelson interferometer
CN119924824A
A Sleep Monitoring Device and Method Based on a Fiber Optic Michelson Interferometer
CN119924824B
Sleep state monitoring and analyzing system based on multi-sensor fusion
CN120859427A
A sleep state monitoring and analyzing system based on multi-sensor fusion
CN120859427B
