Non-inductive fatigue pressure monitoring cushion system and method based on multi-mode optical fiber sensing

Abstract

The invention provides a non-inductive fatigue pressure monitoring cushion system and method based on multi-mode optical fiber sensing. Relates to the technical field of health monitoring and medical instruments, and comprises an optical fiber multi-mode sensing layer which is embedded in a cushion foaming elastomer, forms a two-dimensional sensing grid by staggered Bragg grating array optical fibers and micro-bend loss plastic optical fibers, and is used for collecting heart and lung micro vibration signals and body pressure distribution-micro body movement signals; and the optical signal conditioning and synchronous acquisition unit is coupled with an optical path of the optical fiber multi-mode sensing layer and is used for carrying out amplitude-phase demodulation and temperature drift compensation on optical fiber reflection / transmission light and digitally outputting multichannel original data. The non-inductive fatigue pressure monitoring cushion system and method based on multi-mode optical fiber sensing have the advantages of noninvasive comfort, high-precision recognition, self-adaptive intervention, low power consumption, long endurance and the like, and an accurate and continuous fatigue and pressure management solution is provided for sedentary people.

Description

Technical Field

[0001] The present invention relates to the technical field of health monitoring and medical equipment, and specifically to a non-sensing fatigue pressure monitoring seat cushion system and method based on multimodal optical fiber sensing. Background Art

[0002] Existing methods for monitoring fatigue and stress in sedentary individuals primarily rely on contact-type mechanical sensors such as flexible piezoresistive / piezoelectric films, strain gauges, and capacitive pressure arrays, or on surface-attached photoplethysmography and electrocardiogram (ECG) patches to acquire physiological signals. Some office chairs, car seats, and nursing chairs have these components embedded in the seat cushions or armrests. These components estimate sitting posture and weight distribution through pressure distribution, calculate heart rate through PPG / ECG, and then infer fatigue levels through simple indicators. Other studies have also employed single fiber Bragg gratings or plastic optical fibers within seat cushions, using microbend loss to detect body movement and respiration. Overall, current solutions primarily rely on single-modal measurement, focusing on posture recognition, heart rate statistics, and rough fatigue assessment, and have already formed a prototype of a mass-producible smart chair.

[0003] However, these solutions still have obvious defects: patches, electrodes or handheld sensors require users to keep in contact or wear them, which reduces comfort and compliance; a single modality cannot simultaneously obtain high-fidelity heart and lung waveforms, micro-body movements and fine pressure distribution in the same seat cushion, resulting in insufficient resolution for fatigue-pressure determination; existing systems lack optical fiber, multimodal sensor fusion and real-time algorithm iteration mechanisms, making it difficult to build a closed-loop service from data collection, state recognition to intelligent intervention, and cannot meet the urgent needs of schools, offices, vehicles and homes for long-term, imperceptible and high-precision fatigue and stress monitoring. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a non-sensing fatigue and pressure monitoring seat cushion system and method based on multi-modal fiber optic sensing. It solves the problem of how to achieve multi-modal non-sensing acquisition by implanting interwoven Bragg gratings and slightly bent plastic optical fibers in the seat cushion, and combines edge deep fusion and reinforcement learning algorithms to evaluate fatigue and pressure in real time, providing high-precision, low-power integrated sedentary health management.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a non-sensing fatigue pressure monitoring seat cushion system based on multimodal optical fiber sensing, comprising:

[0006] The optical fiber multimodal sensing layer is embedded in the foam elastic body of the seat cushion. It is composed of a two-dimensional sensing grid composed of interlaced Bragg grating array optical fibers and micro-bend loss plastic optical fibers. It is used to collect cardiopulmonary micro-vibration signals and body pressure distribution-micro-body motion signals.

[0007] An optical signal conditioning and synchronous acquisition unit, coupled to the optical path of the optical fiber multimodal sensing layer, is used to perform amplitude-phase demodulation and temperature drift compensation on the optical fiber reflected / transmitted light and digitally output multi-channel raw data. The optical signal conditioning and synchronous acquisition unit includes a narrow linewidth adjustable laser light source, a temperature compensated photodetector, and a unified time-base synchronous sampling circuit;

[0008] a multimodal feature fusion processing unit, connected to the optical signal conditioning and synchronization acquisition unit, for fusing the cardiopulmonary microvibration signal with the body pressure distribution-micro motion signal in a unified time domain and extracting fatigue-stress characterization features through an edge computing processor and a transformer-long short-term memory hybrid network model;

[0009] a fatigue-stress assessment unit, connected to the multimodal feature fusion processing unit, which uses a reinforcement learning algorithm to output fatigue level, stress risk index, and posture stability in real time according to preset evaluation indicators;

[0010] The intervention execution and cloud-based collaborative learning unit includes a vibration-heating integrated actuator array arranged corresponding to the seat cushion and a cloud-based model management platform. The vibration-heating integrated actuator array implements rhythmic vibration, zoned heating or sitting posture prompt intervention according to the fatigue level and stress risk index. The cloud-based model management platform receives anonymized model gradients and updates global parameters through a federated learning mechanism, and then transmits them downstream to the fatigue-stress assessment unit to achieve end-cloud closed-loop performance iteration.

[0011] Preferably, the Bragg grating array optical fibers are arranged in a serpentine staggered manner with a grid spacing of 15 mm to 25 mm, and the microbend loss plastic optical fibers are located between adjacent Bragg grating optical fibers and are arranged in a 0° / 90° weave pattern.

[0012] Preferably, the optical signal conditioning and synchronous acquisition unit adopts a bilateral reference achromatic dispersion modulation-demodulation algorithm to perform temperature drift compensation, and its strain calculation model is:

[0013]

[0014] Among them, ε c (i, j, t) is the strain at position (i, j) after compensation, Δλ B (i, j, t) is the real-time Bragg wavelength shift, λ B0 (i, j) is the initial wavelength of the installation, T(t) is the real-time temperature, T0 is the calibration temperature, and κ is the thermal sensitivity coefficient, so that the temperature-strain cross-sensitivity error is ≤1με.

[0015] Preferably, the multimodal feature fusion processing unit performs wavelet packet energy spectrum decomposition on the cardiopulmonary micro-vibration signal at the input end, performs third-order tensor high-order singular value decomposition on the body pressure distribution-micro-body motion signal, and splices the obtained 128-dimensional cardiopulmonary features with the 256-dimensional body pressure-body motion features and inputs them into the transformer-long short-term memory hybrid network model.

[0016] Preferably, the instantaneous reward function of the reinforcement learning algorithm is defined as:

[0017] R t =α(1-|F t -F * |)-βE t +γP t

[0018] Among them, F t is the current fatigue factor, F * is the target safety threshold, E t is the energy consumption of the actuator, P t is the attitude stability, α, β, and γ are positive weight coefficients, which are obtained through global search of genetic algorithm to minimize fatigue error and energy consumption and maximize attitude stability simultaneously.

[0019] Preferably, the vibration-heating integrated actuator array constructs a drive map based on the seat cushion pressure matrix and fatigue hotspot map, automatically adjusts the local amplitude and temperature rise, achieves a surface temperature rise of ≤6°C and an amplitude of ≤1.2mm, and maintains an intervention error of ≤5% through closed-loop feedback.

[0020] Preferably, the cloud-based model management platform is based on a hybrid strategy of federated averaging and knowledge distillation, performs a global aggregation of client model weights in each time period, and adopts a temperature-adaptive distillation coefficient to ensure that the cross-user generalization accuracy is improved by ≥5% and the communication overhead is controlled within 1MB / round.

[0021] A non-sensing fatigue stress monitoring method based on multimodal optical fiber sensing, comprising:

[0022] S1. A Bragg grating array of optical fibers is embedded in the seat cushion foam at a preset grid spacing in a serpentine pattern. Microbend-loss plastic optical fibers are then laid out in two mutually perpendicular directions. Simultaneously, cardiopulmonary microvibration signals and body pressure distribution-micromotion signals are collected.

[0023] S2. The Bragg fiber reflected light and the plastic fiber transmitted light are introduced into a narrow-linewidth tunable laser light source. A temperature-compensated photodetector and a unified time-base synchronous sampling circuit perform amplitude-phase demodulation and temperature drift compensation to obtain multi-channel digital raw data.

[0024] S3. Perform frequency-domain energy spectrum analysis on the cardiopulmonary microvibration signal and high-order tensor decomposition on the body pressure distribution-micromotion signal to extract cardiopulmonary eigenvectors and body pressure-motion eigenvectors, respectively. These eigenvectors are then concatenated within a unified time domain and input into a transformer-long short-term memory hybrid network model to generate fused features representing the fatigue-stress state of the human body.

[0025] S4. Based on a reinforcement learning algorithm, the fusion features are inferred online according to preset evaluation indicators, and fatigue level, stress risk index, and posture stability are output in real time. The weight parameters are adaptively updated to improve the accuracy and robustness of the judgment.

[0026] S5. Drive the vibration-heating integrated actuator array in the seat cushion to implement rhythmic vibration, zoned heating, or sitting posture prompt intervention based on the fatigue level and stress risk index. Complete global aggregation and issue updates through the federated averaging-knowledge distillation hybrid strategy to achieve end-cloud closed-loop performance iteration.

[0027] The present invention provides a non-sensing fatigue pressure monitoring seat cushion system and method based on multi-modal fiber optic sensing. It has the following beneficial effects:

[0028] This non-sensing fatigue and pressure monitoring seat cushion system and method based on multimodal fiber sensing achieves synchronous, non-sensing, and high-resolution acquisition of cardiopulmonary micro-vibrations, body pressure distribution, and micro-body movements by constructing a two-dimensional multimodal sensing grid formed by interweaving Bragg gratings and micro-bend loss plastic optical fibers inside the seat cushion. This completely eliminates the long-term dependence of patches, electrodes, or handheld sensors on the skin, significantly improving comfort and user compliance in sedentary scenarios. Combined with a narrow-linewidth laser light source and a temperature-compensated photoelectric demodulation circuit, it can effectively suppress signal drift caused by temperature drift and material aging, thereby ensuring measurement consistency and long-term stability across environments and individuals. By introducing wavelet energy spectrum analysis, high-order tensor decomposition, and transformer-long short-term memory hybrid networks at the edge, multi-source features can be deeply integrated in a unified time domain, significantly improving the accuracy of determining fatigue level, pressure risk, and posture stability, and achieving high-fidelity monitoring requirements covering multiple physiological dimensions with a single device.

[0029] The present invention adopts reinforcement learning-driven real-time evaluation and vibration-heating integrated actuator array linkage, combined with the cloud-based model management platform of federated averaging and knowledge distillation, to build a closed-loop health management architecture of collection-identification-intervention-iteration. This architecture can not only dynamically adjust the intervention intensity according to the individual status and keep the intervention error below 5%, but also achieve continuous improvement in the generalization accuracy of cross-user models while ensuring data privacy, while controlling the communication overhead to within 1MB / round, taking into account energy consumption, bandwidth and algorithm performance. Overall, the present invention has the advantages of non-invasive comfort, high-precision identification, adaptive intervention, low power consumption and long battery life. It can be widely used in scenarios such as office, driving, rehabilitation care and smart home, providing accurate and continuous fatigue and stress management solutions for sedentary people. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a flowchart for implementing the invention. DETAILED DESCRIPTION

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0032] like Figure 1 As shown, an embodiment of the present invention provides a non-sensing fatigue pressure monitoring seat cushion system based on multimodal fiber optic sensing, including a fiber optic multimodal sensing layer embedded in the foam elastic body of the seat cushion, and a two-dimensional sensing grid composed of staggered Bragg grating array optical fibers and micro-bend loss plastic optical fibers, which is used to collect cardiopulmonary micro-vibration signals and body pressure distribution-micro-body motion signals.

[0033] The Bragg grating array optical fibers are arranged in a serpentine staggered pattern with a grid spacing of 15mm-25mm. The microbend loss plastic optical fiber is located between adjacent Bragg grating optical fibers and is arranged in a 0° / 90° weave pattern.

[0034] The optical signal conditioning and synchronous acquisition unit is coupled to the optical path of the optical fiber multimodal sensing layer and is used to perform amplitude-phase demodulation, temperature drift compensation and digital output of multi-channel raw data on the reflected / transmitted light of the optical fiber. The optical signal conditioning and synchronous acquisition unit includes a narrow linewidth adjustable laser light source, a temperature compensated photodetector and a unified time base synchronous sampling circuit.

[0035] The optical signal conditioning and synchronous acquisition unit uses a bilateral reference achromatic dispersion modulation-demodulation algorithm to compensate for temperature drift. Its strain calculation model is:

[0036]

[0037] Among them, ε c (i, j, t) is the strain at position (i, j) after compensation, Δλ B (i, j, t) is the real-time Bragg wavelength shift, λ B0 (i, j) is the initial wavelength of the installation, T(t) is the real-time temperature, T0 is the calibration temperature, and κ is the thermal sensitivity coefficient, so that the temperature-strain cross-sensitivity error is ≤1με.

[0038] The multimodal feature fusion processing unit is connected to the optical signal conditioning and synchronous acquisition unit. It is used to fuse the cardiopulmonary micro-vibration signals and the body pressure distribution-micro-body motion signals in a unified time domain through the edge computing processor and the transformer-long short-term memory hybrid network model and extract fatigue-stress characterization features.

[0039] The multimodal feature fusion processing unit performs wavelet packet energy spectrum decomposition on the cardiopulmonary micro-vibration signal at the input end, performs third-order tensor high-order singular value decomposition on the body pressure distribution-micro-body motion signal, and concatenates the obtained 128-dimensional cardiopulmonary features with the 256-dimensional body pressure-body motion features and inputs them into the transformer-long short-term memory hybrid network model.

[0040] The fatigue-stress assessment unit is connected to the multimodal feature fusion processing unit and uses a reinforcement learning algorithm to output fatigue level, stress risk index and posture stability in real time according to preset evaluation indicators.

[0041] The immediate reward function of the reinforcement learning algorithm is defined as:

[0042] R t =α(1-|F t -F * |)-βE t +γP t

[0043] Among them, F t is the current fatigue factor, F * is the target safety threshold, E t is the energy consumption of the actuator, P t is the attitude stability, α, β, and γ are positive weight coefficients, which are obtained through global search of genetic algorithm to minimize fatigue error and energy consumption and maximize attitude stability simultaneously.

[0044] The intervention execution and cloud-based collaborative learning unit includes a vibration-heating integrated actuator array arranged corresponding to the seat cushion and a cloud-based model management platform. The vibration-heating integrated actuator array implements rhythmic vibration, zoned heating or sitting posture prompt intervention based on fatigue level and stress risk index. The cloud-based model management platform receives anonymized model gradients and updates global parameters through the federated learning mechanism, and then transmits them down to the fatigue-stress assessment unit to achieve end-cloud closed-loop performance iteration.

[0045] The integrated vibration and heating actuator array constructs a drive map based on the seat pressure matrix and fatigue heat map, automatically adjusting local amplitude and temperature rise to achieve a surface temperature rise of ≤6°C and an amplitude of ≤1.2mm. Through closed-loop feedback, the intervention error is maintained at ≤5%. The cloud-based model management platform utilizes a hybrid strategy of federated averaging and knowledge distillation, globally aggregating client model weights every time period and employing a temperature-adaptive distillation coefficient. This ensures cross-user generalization accuracy is improved by ≥5%, while keeping communication overhead within 1MB per round.

[0046] A non-sensing fatigue stress monitoring method based on multimodal optical fiber sensing, comprising:

[0047] S1. Fiber Bragg grating arrays are embedded in the seat cushion foam at a preset grid spacing in a serpentine pattern. Microbend-loss plastic optical fibers are then weaved in two perpendicular directions to simultaneously collect cardiopulmonary microvibration signals and body pressure distribution-microbody motion signals.

[0048] S2. The Bragg fiber reflected light and the plastic fiber transmitted light are introduced into a narrow linewidth tunable laser light source respectively. After amplitude-phase demodulation and temperature drift compensation are performed by a temperature-compensated photoelectric detector and a unified time-base synchronous sampling circuit, multi-channel digital raw data is obtained.

[0049] S3. Perform frequency domain energy spectrum analysis on the cardiopulmonary micro-vibration signals and high-order tensor decomposition on the body pressure distribution-micro-body motion signals. Extract the cardiopulmonary eigenvectors and body pressure-body motion eigenvectors respectively. Concatenate the two types of eigenvectors in the unified time domain and input them into the transformer-long short-term memory hybrid network model to generate fused features that represent the fatigue-stress state of the human body.

[0050] S4. Based on the reinforcement learning algorithm, the fusion features are inferred online according to the preset evaluation indicators, and the fatigue level, pressure risk index and posture stability are output in real time. The weight parameters are adaptively updated to improve the judgment accuracy and robustness.

[0051] S5. Drive the vibration-heating integrated actuator array in the seat cushion to implement rhythmic vibration, zoned heating, or sitting posture prompt intervention based on fatigue level and stress risk index. Use the federated averaging-knowledge distillation hybrid strategy to complete global aggregation and issue updates, achieving end-to-cloud closed-loop performance iteration.

[0052] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A non-sensing fatigue pressure monitoring seat cushion system based on multi-modal optical fiber sensing, characterized in that: include: The optical fiber multimodal sensing layer is embedded in the foam elastic body of the seat cushion. It is composed of a two-dimensional sensing grid composed of interlaced Bragg grating array optical fibers and micro-bend loss plastic optical fibers. It is used to collect cardiopulmonary micro-vibration signals and body pressure distribution-micro-body motion signals. An optical signal conditioning and synchronous acquisition unit, coupled to the optical path of the optical fiber multimodal sensing layer, is used to perform amplitude-phase demodulation and temperature drift compensation on the optical fiber reflected / transmitted light and digitally output multi-channel raw data. The optical signal conditioning and synchronous acquisition unit includes a narrow linewidth adjustable laser light source, a temperature compensated photodetector, and a unified time-base synchronous sampling circuit; a multimodal feature fusion processing unit, connected to the optical signal conditioning and synchronization acquisition unit, for fusing the cardiopulmonary microvibration signal with the body pressure distribution-micro motion signal in a unified time domain and extracting fatigue-stress characterization features through an edge computing processor and a transformer-long short-term memory hybrid network model; a fatigue-stress assessment unit, connected to the multimodal feature fusion processing unit, which uses a reinforcement learning algorithm to output fatigue level, stress risk index, and posture stability in real time according to preset evaluation indicators; The intervention execution and cloud-based collaborative learning unit includes a vibration-heating integrated actuator array arranged corresponding to the seat cushion and a cloud-based model management platform. The vibration-heating integrated actuator array implements rhythmic vibration, zoned heating or sitting posture prompt intervention according to the fatigue level and stress risk index. The cloud-based model management platform receives anonymized model gradients and updates global parameters through a federated learning mechanism, and then transmits them downstream to the fatigue-stress assessment unit to achieve end-cloud closed-loop performance iteration.

2. The seat cushion system for monitoring fatigue and pressure based on multimodal optical fiber sensing according to claim 1, characterized in that: The Bragg grating array optical fibers are arranged in a serpentine staggered manner with a grid spacing of 15mm-25mm, and the microbend loss plastic optical fibers are located between adjacent Bragg grating optical fibers and arranged in a 0° / 90° weave pattern.

3. The seat cushion system for monitoring fatigue and pressure based on multimodal optical fiber sensing according to claim 1, characterized in that: The optical signal conditioning and synchronous acquisition unit uses a bilateral reference achromatic dispersion modulation-demodulation algorithm to compensate for temperature drift, and its strain calculation model is: Among them, ε c (i, j, t) is the strain at position (i, j) after compensation, Δλ B (i, j, t) is the real-time Bragg wavelength shift, λ B0 (i, j) is the initial wavelength of the installation, T(t) is the real-time temperature, T0 is the calibration temperature, and κ is the thermal sensitivity coefficient.

4. The seat cushion system for monitoring fatigue and pressure based on multimodal optical fiber sensing according to claim 1, characterized in that: The multimodal feature fusion processing unit performs wavelet packet energy spectrum decomposition on the cardiopulmonary micro-vibration signal at the input end, performs third-order tensor high-order singular value decomposition on the body pressure distribution-micro-body motion signal, and splices the obtained 128-dimensional cardiopulmonary features with the 256-dimensional body pressure-body motion features and inputs them into the transformer-long short-term memory hybrid network model.

5. The seat cushion system for monitoring fatigue and pressure based on multimodal optical fiber sensing according to claim 1 is characterized by: The instantaneous reward function of the reinforcement learning algorithm is defined as: R t =α(1-|F t -F * |)-βE t +γP t Among them, F t is the current fatigue factor, F * is the target safety threshold, E t is the energy consumption of the actuator, P t is the attitude stability, α, β, and γ are positive weight coefficients, which are obtained through global search of genetic algorithm.

6. The seat cushion system for monitoring fatigue and pressure based on multimodal optical fiber sensing according to claim 1, characterized in that: The vibration-heating integrated actuator array constructs a drive map based on the seat cushion pressure matrix and fatigue hotspot map, automatically adjusts the local amplitude and temperature rise, achieves a surface temperature rise of ≤6°C and an amplitude of ≤1.2mm, and maintains an intervention error of ≤5% through closed-loop feedback.

7. The seat cushion system for monitoring fatigue and pressure based on multimodal optical fiber sensing according to claim 1 is characterized by: The cloud-based model management platform is based on a hybrid strategy of federated averaging and knowledge distillation. It globally aggregates the client model weights once per time period and adopts a temperature-adaptive distillation coefficient.

8. A non-sensing fatigue stress monitoring method based on multimodal optical fiber sensing, characterized in that: include: S1. A Bragg grating array of optical fibers is embedded in the seat cushion foam at a preset grid spacing in a serpentine pattern. Microbend-loss plastic optical fibers are then laid out in two mutually perpendicular directions. Simultaneously, cardiopulmonary microvibration signals and body pressure distribution-micromotion signals are collected. S2. The Bragg fiber reflected light and the plastic fiber transmitted light are introduced into a narrow-linewidth tunable laser light source. A temperature-compensated photodetector and a unified time-base synchronous sampling circuit perform amplitude-phase demodulation and temperature drift compensation to obtain multi-channel digital raw data. S3. Perform frequency-domain energy spectrum analysis on the cardiopulmonary microvibration signal and high-order tensor decomposition on the body pressure distribution-micromotion signal to extract cardiopulmonary eigenvectors and body pressure-motion eigenvectors, respectively. These eigenvectors are then concatenated within a unified time domain and input into a transformer-long short-term memory hybrid network model to generate fused features representing the fatigue-stress state of the human body. S4. Based on a reinforcement learning algorithm, the fusion features are inferred online according to preset evaluation indicators, and fatigue level, stress risk index, and posture stability are output in real time. The weight parameters are adaptively updated to improve the accuracy and robustness of the judgment. S5. Drive the vibration-heating integrated actuator array in the seat cushion to implement rhythmic vibration, zoned heating, or sitting posture prompt intervention based on the fatigue level and stress risk index. Complete global aggregation and issue updates through the federated averaging-knowledge distillation hybrid strategy to achieve end-cloud closed-loop performance iteration.

Images