A single-channel self-decoupled flexible multi-modal hand rehabilitation sensing system

Abstract

This invention discloses a single-channel self-decoupled flexible multimodal hand rehabilitation sensing system, comprising a flexible glove substrate, a signal acquisition module, a series sensing unit, and a motion classification module. The series sensing unit consists of a pressure sensing unit and a strain sensing unit connected in series, both employing a polydopamine-modified multi-walled carbon nanotube / polyurethane sponge structure. The opposing resistance characteristics are achieved solely by whether the pores are filled with silicone elastomer. The two types of sensing units are respectively positioned at the finger joints and fingertips, sharing a single acquisition channel. Physical self-decoupling of the signal is achieved based on the natural timing of the grasping motion, eliminating the need for complex decoupling algorithms. The preprocessed timing data is transmitted to the motion classification module via the signal acquisition module to identify the grasping quality. This invention utilizes homogeneous sensor fabrication and achieves physical self-decoupling of the signal through a single channel, enabling lightweight and quantitative monitoring and evaluation of hand rehabilitation movements.

Description

Technical Field

[0001] This application relates to the field of wearable technology, and more specifically, to a single-channel self-decoupled flexible multimodal hand rehabilitation sensing system. Background Technology

[0002] Hand dysfunction is a common sequela of stroke, hand injury, arthritis, and spinal cord injury, severely affecting patients' ability to live independently in daily life. Hand rehabilitation training is the core means of restoring patients' hand function. Traditional rehabilitation training relies heavily on one-on-one guidance from medical staff, lacks objective and quantitative assessment methods, and is limited by the hospital setting, making it difficult for patients to train independently at home.

[0003] Existing hand rehabilitation equipment is mainly divided into two categories: rigid exoskeleton rehabilitation gloves: These use motors and linkages to drive finger movements and can provide significant assistive force, but they suffer from drawbacks such as bulky structure, high cost, support only passive training, and poor wearing comfort, making them unsuitable for home settings. Flexible smart rehabilitation gloves: These use flexible sensors to achieve real-time monitoring of finger movements and are lightweight.

[0004] However, the existing solutions have obvious technical limitations, mainly including the following three points: (1) Multimodal sensing relies on multi-channel acquisition: In order to capture the bending posture of the finger and the pressure of the fingertip at the same time, independent strain sensors and pressure sensors need to be arranged on a single finger, and independent acquisition channels need to be set up accordingly. This results in complex system wiring, high hardware cost, and is not conducive to the lightweighting of gloves. (2) Difficulty in signal decoupling: When bending and pressure stimulation are applied simultaneously, the two types of sensing signals are coupled to each other and need to be separated by complex algorithm models. The solution accuracy is limited and the system computing power requirement is increased, making it difficult to achieve real-time low-power monitoring. (3) Inconsistent manufacturing processes: Existing strain sensors and pressure sensors are usually developed based on different material systems and processes, which are complicated and have poor production consistency, further increasing the system cost.

[0005] Furthermore, while existing flexible sensors based on porous conductive sponges can achieve tensile or compression sensing through structural design, they cannot achieve two opposite response characteristics—positive and negative piezoresistive resistance—through simple process control on the same material substrate, making it difficult to support the realization of a single-channel multimodal sensing architecture. Summary of the Invention

[0006] To address the aforementioned issues, this application provides a single-channel self-decoupling flexible multimodal hand rehabilitation sensing system.

[0007] This invention provides a single-channel self-decoupled flexible multimodal hand rehabilitation sensing system, including a flexible glove substrate, a signal acquisition module, at least one serial sensing unit, and a motion classification module; The series sensing unit is disposed in the single-fin area of ​​the flexible glove substrate and is composed of a pressure sensing unit and a strain sensing unit connected in series. The pressure sensing unit includes a first conductive sponge substrate and parallel plate electrodes. The interior of the first conductive sponge substrate is an unfilled three-dimensional porous structure, and the parallel plate electrodes are respectively attached and fixed to the upper and lower end faces of the first conductive sponge substrate. The strain sensing unit includes a second conductive sponge substrate, an organosilicon elastomer, and lateral strip electrodes. The organosilicon elastomer completely fills the internal pores of the second conductive sponge substrate, and the lateral strip electrodes are respectively attached and fixed to the left and right side end faces of the second conductive sponge substrate. Both the first conductive sponge matrix and the second conductive sponge matrix are polydopamine-modified multi-walled carbon nanotube / polyurethane conductive sponge structures. The serial sensing unit is connected to the acquisition channel of the signal acquisition module via a lead wire, and the output terminal of the signal acquisition module is connected to the signal of the motion classification module.

[0008] In one optional embodiment, the pressure sensing unit is attached and fixed to the fingertip area of ​​a single finger region of the flexible glove substrate, and a flexible pressure equalizing pad is provided on the side of the pressure sensing unit facing the skin of the human fingertip.

[0009] In one optional embodiment, the parallel plate electrode is a double-layer conductive cloth holding wire structure, and a conductive carbon paste layer is provided on the bonding surface between the conductive cloth and the first conductive sponge substrate.

[0010] In one alternative embodiment, the strain sensing unit is attached and fixed to the back of the interphalangeal joint in a single finger region of the flexible glove substrate, and is bonded to the glove fiber layer by an organosilicon elastomer.

[0011] In one optional embodiment, the silicone elastomer of the strain sensing unit is filled and molded using a vacuum infiltration process, specifically: The silicone prepolymer and curing agent are mixed in a preset ratio and degassed under vacuum. The second conductive sponge matrix is ​​then immersed in the mixed adhesive solution and kept under negative pressure for a preset time until the adhesive solution fills the internal pores of the second conductive sponge matrix. Remove the soaked second conductive sponge substrate, remove excess adhesive from the surface, and heat to cure, forming an organosilicon elastomer within the pores.

[0012] In one optional embodiment, the method for preparing the polydopamine-modified multi-walled carbon nanotube / polyurethane conductive sponge includes: The polyurethane foam substrate was pretreated with hydrophilic modification. Dopamine hydrochloride was added to Tris-HCl buffer to form a mixed solution with a pH of 8.5, which yielded a dopamine reaction solution. The pretreated polyurethane sponge was placed in the dopamine reaction solution to react and generate a polydopamine modified layer on the surface of the sponge. A polyurethane sponge substrate with a polydopamine-modified layer was immersed in a carboxylated multi-walled carbon nanotube dispersion for reaction. The reaction was then carried out by gradient drying to fix the multi-walled carbon nanotubes and polydopamine through covalent bonds.

[0013] In one optional implementation, the signal acquisition module includes data enhancement after acquiring the time-domain signal, specifically: The time axis of the time-domain signal is linearly scaled while keeping the signal amplitude and waveform unchanged. Only the overall duration of the signal is adjusted to simulate the signal differences at different action rates. Gaussian noise with a preset amplitude difference is superimposed on the time-domain signal after scaling along the time axis to simulate force fluctuations.

[0014] In one optional implementation, the action classification module has a pre-trained classification model built in, which is used to extract temporal features from the time-domain signal transmitted by the signal acquisition module, and to identify the grip quality based on the temporal features. The time-series characteristics include the maximum slope of the bending rise, the maximum value of the second derivative, the minimum value of the second derivative, the peak-to-peak value, the relative peak value, and the bending-squeeze switching time.

[0015] In one alternative implementation, the grip quality includes normal grip, insufficient bending, insufficient force, and excessive force.

[0016] This application has at least the following advantages or beneficial effects: This invention, based on the same conductive sponge precursor, allows for the separate fabrication of negative piezoresistive pressure sensors and positive piezoresistive strain sensors by varying the process of whether or not to fill them with silicone elastomer. The two types of sensing units share the same pre-fabrication process, eliminating the need to develop two independent material systems and production processes. This significantly simplifies the fabrication process, reduces manufacturing costs, and ensures consistent device performance.

[0017] This invention connects two types of sensing units with positive and negative piezoresistive responses in series. Leveraging the natural timing characteristics of hand grasping motions and utilizing the opposite response directions, bending and pressure information correspond to the rising and falling edges of the signal, respectively, achieving natural separation in the time domain and realizing physical signal self-decoupling. This reduces the traditional dual-channel acquisition method to a single channel, significantly decreasing the number of leads and the scale of the acquisition hardware, lowering system complexity and hardware costs. Simultaneously, it avoids the errors and computational demands caused by algorithm decoupling, improving system reliability and real-time performance. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the overall architecture and application principle of a single-channel self-decoupled flexible multimodal hand rehabilitation sensing system according to an embodiment of the present invention. Part (a) is a schematic diagram of a typical application scenario of hand rehabilitation training, part (b) is a schematic diagram of the sensing response characteristics of pressure sensor and strain sensor, part (c) is a structural diagram of flexible glove integrated system, part (d) is a schematic diagram of single-channel signal self-decoupling principle, part (e) is a schematic diagram of hand movement signal monitoring, and part (f) is a schematic diagram of the performance comparison of various machine learning algorithms for movement classification. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] Please refer to Figure 1 This is a schematic diagram illustrating the overall architecture and application principle of a single-channel self-decoupling flexible multimodal hand rehabilitation sensing system proposed in an embodiment of this application. Figure 1 As shown in part (a), the present invention is applicable to various hand rehabilitation scenarios such as sports injuries, arthritis, nerve paralysis and postoperative rehabilitation. It can perform real-time, non-invasive dynamic monitoring of fingertip pressure and joint bending angle during the rehabilitation process, providing quantitative data support for rehabilitation effect evaluation.

[0022] The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system described in this embodiment includes a flexible glove substrate, a signal acquisition module, at least one serial sensing unit, and a motion classification module.

[0023] In this embodiment, the flexible glove base is a knitted elastic glove, which is adapted to the normal adult hand shape. Each finger is equipped with a set of series sensing units, and the entire glove is equipped with a total of five sets of series sensing units.

[0024] Each cascaded sensing unit consists of a pressure sensing unit and a strain sensing unit connected in series. The series connection is then routed to a single acquisition channel of the signal acquisition module via a lead wire; that is, a single finger occupies only one acquisition channel. The output of the signal acquisition module is connected to the motion classification module, transmitting the acquired time-domain signal to the motion classification module for feature extraction and motion quality recognition.

[0025] The pressure sensing unit includes a first conductive sponge substrate and parallel plate electrodes. The interior of the first conductive sponge substrate is an unfilled three-dimensional porous structure, and the parallel plate electrodes are respectively attached and fixed to the upper and lower end faces of the first conductive sponge substrate.

[0026] In this embodiment, the parallel plate electrode is a double-layer conductive cloth holding wire structure. The conductive cloth and the first conductive sponge substrate are bonded with a conductive carbon paste layer. After the conductive carbon paste layer is heated and cured, it forms a stable ohmic contact between the electrode and the conductive sponge.

[0027] The strain sensing unit includes a second conductive sponge substrate, an organosilicon elastomer, and lateral strip electrodes. The organosilicon elastomer completely fills the internal pores of the second conductive sponge substrate, and the lateral strip electrodes are respectively attached and fixed to the left and right side ends of the second conductive sponge substrate.

[0028] Both the first conductive sponge matrix and the second conductive sponge matrix are polydopamine-modified multi-walled carbon nanotube / polyurethane conductive sponge structures. They are prepared using the same universal precursor and achieve negative piezoresistive pressure response and positive piezoresistive strain response respectively, only by the difference in process of whether or not to fill with organosilicon elastomer.

[0029] In this embodiment, a 60ppi polyurethane sponge was used as a three-dimensional skeleton substrate, cut into cubic samples with dimensions of 20mm×10mm×5mm. Polydopamine-modified multi-walled carbon nanotubes / polyurethane conductive sponge were prepared according to the following steps as a universal precursor for two types of sensing units: Step 1: Place the cut sponge sample in anhydrous ethanol and deionized water alternately, and ultrasonically clean for 3 minutes each time, repeating twice to remove surface impurities; then immerse it in a 10% NaOH aqueous solution and treat it at room temperature for 1 hour to modify the surface of the polyurethane sponge with hydrophilicity and improve the adhesion of subsequent coatings; after alkali treatment, clean it again alternately with anhydrous ethanol and deionized water, and dry it in a 50℃ constant temperature drying oven for 1 hour to obtain the pretreated polyurethane sponge substrate.

[0030] Step 2: Add Tris-HCl buffer solution to deionized water to adjust the pH of the system to 8.5. Weigh dopamine hydrochloride and add it to the buffer solution. Stir slowly for 3 minutes until completely dissolved to obtain a dopamine reaction solution. Immerse the pretreated polyurethane sponge completely in the reaction solution and stir magnetically at room temperature for 12 hours. Under weakly alkaline conditions, dopamine undergoes oxidative self-polymerization to form a uniform polydopamine nano-coating on the surface of the sponge skeleton. After the reaction is completed, remove the sponge and place it in anhydrous ethanol. Stir slowly for 1 minute to wash away the loose polydopamine particles physically adsorbed on the surface to obtain a polydopamine-modified polyurethane sponge.

[0031] Step 3: Weigh carboxylated multi-walled carbon nanotubes and disperse them in anhydrous ethanol to prepare a 2 mg / mL carboxylated multi-walled carbon nanotube dispersion. Disperse the dispersion using a probe ultrasonic instrument under ice-water bath conditions for 1 hour to obtain a uniform and stable multi-walled carbon nanotube dispersion. Immerse the polydopamine-modified polyurethane sponge in the multi-walled carbon nanotube dispersion and react with magnetic stirring at room temperature for 2 hours to ensure that the multi-walled carbon nanotube dispersion is fully loaded on the polydopamine coating surface. After removal, let it stand at room temperature for 30 minutes, and then transfer it to a 70℃ constant temperature drying oven to dry for 1 hour. During the gradient drying process, the amino groups on the polydopamine surface and the carboxyl groups on the surface of the multi-walled carbon nanotube dispersion undergo dehydration condensation to form covalent amide bonds, achieving firm anchoring of conductive nanoparticles. Gently tap to remove loose particles on the surface to obtain a polydopamine-modified multi-walled carbon nanotube / polyurethane conductive sponge structure, which serves as a universal precursor for two types of sensing units.

[0032] It should be noted that the above process parameters are preferred values ​​for this embodiment. In practical applications, they can be adjusted within the corresponding value range according to the target sensitivity and mass production process requirements.

[0033] Pressure sensing units and strain sensing units were prepared by varying the process of whether or not silicone elastomers were filled in, based on the same universal conductive sponge precursor.

[0034] Furthermore, in this embodiment, the pressure sensing unit is fabricated based on the aforementioned conductive sponge precursor, and the specific fabrication process is as follows: Cut the conductive cloth into a rectangle that matches the upper surface of the conductive sponge. Place the copper core at the end of the wire in the center of the adhesive side of one conductive cloth, and then take another conductive cloth to align and attach it. The wire is then clamped and fixed by the double-layer conductive cloth. A low-resistance conductive carbon paste is evenly coated on the bonding surface of the two electrodes and attached to the upper and lower end faces of the first conductive sponge substrate, respectively. Light pressure is applied to ensure tight contact.

[0035] The assembled device is placed in a 90℃ constant temperature drying oven for 10 minutes to allow the conductive carbon paste to fully solidify and form a stable ohmic contact, thus obtaining a pressure sensing unit with negative piezoresistive response.

[0036] The prepared pressure sensing unit is attached and fixed to the fingertip area of ​​the single finger region of the flexible glove substrate. A flexible pressure equalizing pad is provided on the side of the pressure sensing unit facing the skin of the human fingertip to ensure uniform transmission of pressure and improve the fit.

[0037] It should be noted that the pressure sensing unit adopts a structure with parallel plate electrodes arranged on the upper and lower end faces. The electrode normal direction is coaxial with the loading direction of the fingertip pressure, which can maximize the acquisition of resistance changes caused by compression deformation, ensure the sensitivity and linearity of pressure response, and reduce the interference of lateral deformation on the detection results.

[0038] The first conductive sponge matrix retains an unfilled three-dimensional porous structure, which can achieve a negative piezoresistive response: when compressed by vertical pressure, the pores inside the sponge collapse, the number of contact points between the conductive networks on the skeleton surface increases, the number of parallel conductive paths increases, and the overall resistance decreases with increasing pressure, exhibiting negative piezoresistive characteristics.

[0039] Furthermore, in this embodiment, the strain sensing unit is fabricated based on the same universal conductive sponge precursor, and the silicone elastomer is filled and molded using a vacuum infiltration process. The specific fabrication and installation process is as follows: The conductive cloth is cut into strips, and conductive carbon paste is coated on one side of the conductive cloth. The strips are then attached to symmetrical positions on the left and right sides of the second conductive sponge substrate to form lateral strip electrodes. The length of the electrodes is parallel to the direction of finger bending and stretching, and the reserved area of ​​the electrodes is used to connect wires. After attachment, the strips are placed in a 90°C drying oven for 10 minutes to ensure stable electrode contact.

[0040] Mix the silicone prepolymer and curing agent thoroughly at a mass ratio of 1:1, and remove air bubbles by vacuum degassing for 5 minutes. Immerse the second conductive sponge matrix with lateral strip electrodes completely in the mixed adhesive solution, place it in a vacuum dryer, slowly evacuate to -0.09 MPa, and maintain for 15 minutes. Under negative pressure, allow the adhesive solution to penetrate and fill all the internal pores of the second conductive sponge matrix.

[0041] After slowly releasing the vacuum, the second conductive sponge substrate is removed, and excess adhesive on the surface is absorbed with filter paper. The sample is then placed in an 80°C forced-air oven and heated for 30 minutes to allow the silicone prepolymer to fully crosslink and form a silicone elastomer within the pores, thus obtaining a strain sensing unit with positive piezoresistive response.

[0042] The prepared strain sensing unit is attached and fixed to the back of the finger joint in a single finger area of ​​the flexible glove substrate, and is bonded to the glove fiber layer by an organosilicon elastomer. The elastomer used for bonding has the same modulus as the organosilicon elastomer filling the pores, which can avoid the strain transmission loss caused by the bonding layer being too hard, and ensure that the finger bending deformation can be completely transmitted to the inside of the sensing unit.

[0043] It should be noted that the internal pores of the strain sensing unit are completely filled with silicone elastomer. Under pressure, the pores cannot close to form new conductive paths, and the vertical pressure has minimal impact on the unit's resistance, naturally providing anti-interference capabilities. Under tension, the near-incompressible silicone elastomer exhibits a Poisson effect, and the lateral contraction leads to an increase in the spacing between conductive networks, increasing the interparticle tunneling resistance. The overall resistance increases with increasing strain, exhibiting positive piezoresistive characteristics.

[0044] Reference Figure 1 (b) Since the two types of sensing units respond in opposite directions and the action process has a clear sequence, the bending and pressure information correspond to the rising and falling edges of the single-channel signal, respectively. They are naturally separated in the time domain. No complex decoupling algorithm is required. The bending posture and pressure intensity information can be extracted from the single-channel signal simply by dividing the time window.

[0045] Based on the aforementioned differentiated resistive switching characteristics, a structure such as Figure 1 (c) The flexible glove integrated system shown in section 1. The pressure sensing unit is arranged at the force-bearing position of the glove fingertip, and the strain sensing unit is arranged at the bending position of the corresponding finger joint. The two are connected in series to the same signal acquisition circuit to form a single-channel multimodal sensing node.

[0046] Understandable, such as Figure 1 As shown in section (d), since the hand's grasping action has a natural temporal sequence, consisting of a bending stage and a squeezing stage, and combining the opposite resistive switching characteristics of the two types of sensing units, physical-level self-decoupling can be achieved in a single-channel signal. The specific process is as follows: Squeezing phase: After the finger bends to contact the object, the fingertip begins to apply pressure. The pressure sensing unit at the fingertip is compressed and exhibits a negative piezoresistive response. The resistance gradually decreases, which dominates the change in the total resistance of the series circuit, corresponding to the falling edge of the single-channel signal.

[0047] Bending phase: When the finger joint bends first, it causes the strain sensing unit at the joint to undergo tensile deformation. The strain unit exhibits a positive piezoresistive response, and the resistance rises rapidly, which dominates the change in the total resistance of the series circuit, corresponding to the rising edge of the single-channel signal.

[0048] Compression + Bending Stage: The finger is continuously bent and simultaneously applies a pressing load to the object. The strain sensing unit remains in a tensile high-resistance state, while the pressure sensing unit's resistance decreases under pressure. The two resistance-change effects are superimposed, and the signal rise rate slows down, forming a smooth transition range.

[0049] Furthermore, such as Figure 1As shown in section (e), for typical hand rehabilitation movements, each sensing unit will output response signals with temporal and amplitude differences, forming a multi-channel signal feature combination that corresponds one-to-one with a specific movement. By extracting features from the multi-channel signals and fusing them with multiple modes, the type of hand movement can be accurately identified, and the range of motion and pressure of the joint can be quantified.

[0050] Furthermore, the signal acquisition module performs data enhancement on the acquired time-domain signal, specifically as follows: The time axis of the time-domain signal is linearly scaled while keeping the signal amplitude and waveform unchanged. Only the overall duration of the signal is adjusted to simulate the signal differences at different action rates.

[0051] In this embodiment, the time axis is scaled to 85%-115% of the original duration, covering the fluctuations in the speed of movement of different users and the changes in the rate of movement at different stages of rehabilitation.

[0052] Gaussian noise with a preset amplitude difference is superimposed on the time-domain signal after scaling along the time axis to simulate force fluctuations.

[0053] In this embodiment, the amplitude of Gaussian noise is 1% of the peak-to-peak value of the signal, which is used to simulate the signal disturbance caused by the minute fluctuations and contact posture differences during the application of force by the fingertip.

[0054] The data augmentation process described above can improve the generalization and recognition capabilities of the classification model in the subsequent action classification module.

[0055] The action classification module incorporates a pre-trained classification model to extract temporal features from the time-domain signal obtained from the signal acquisition module, and identifies grip quality based on these features. In this embodiment, the extracted temporal features include the following six categories: The maximum slope of the upward bend reflects the initiation rate of finger flexion and corresponds to the joint's mobility. The maximum value of the second derivative reflects the acceleration of the bending motion, corresponding to the impact intensity at the moment of contact; The minimum value of the second derivative reflects the deceleration of the squeezing action and corresponds to the impact characteristics of the applied force; Signal amplitude difference: The difference between the highest and lowest points of the signal, reflecting the bending amplitude and the magnitude of the applied force; Relative peak value: The ratio of the peak value during the bending stage to the overall peak value, reflecting the proportional relationship between bending and applied force; Bending-compression switching moment: the point in time when the bending edge changes to the falling edge, reflecting the proportional relationship between bending and applied force.

[0056] In this embodiment, as Figure 1As shown in (f), through comparative analysis of the performance of machine learning models, the XGBoost machine learning model trained with 10×10 fold cross-validation was selected. The grasping quality that can be identified includes four categories: normal grasping, insufficient bending, insufficient force, and excessive force. Testing showed that the model's average recognition accuracy for the four types of grasping quality can reach over 99%, enabling objective quantitative assessment of the quality of rehabilitation movements.

[0057] As a further optimization option, a suitable machine learning model can be selected based on the actual usage. For example, a random forest model can be used, which has fast training speed, strong anti-overfitting ability, and low computational overhead, making it suitable for local deployment in embedded, low-computing-power acquisition terminals. A support vector machine model can also be used, as it has stable generalization performance on small to medium-sized datasets, making it suitable for customized rehabilitation training scenarios with limited sample sizes. Other models such as k-nearest neighbors, decision trees, and logistic regression can also be used to further simplify the computational logic and adapt to micro-wearable devices with more limited computing power.

[0058] In this embodiment, the training dataset is obtained using the following method: The experimenter wore sensor gloves and completed the action cycle according to the rhythm. Each cycle consisted of "natural extension - completion of action - recovery action". The relative rate of change of resistance signal of a single channel was collected in real time using a digital multimeter.

[0059] Three typical hand rehabilitation movements were selected for monitoring: Specific finger flexion: individually move a single finger joint to assess the range of motion and flexion control of the interphalangeal joint; Pinch test card: Use your thumb and forefinger to pinch a thin card to assess your fingertips’ precise control over the pinching force. Grip strength ball: Measure the elastic grip strength of the whole hand, and comprehensively assess the coordination of hand flexion and force application, as well as overall muscle strength.

[0060] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0061] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0062] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0063] The above provides a detailed description of a single-channel self-decoupling flexible multimodal hand rehabilitation sensing system provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A single-channel self-decoupling flexible multimodal hand rehabilitation sensing system, characterized in that, It includes a flexible glove base, a signal acquisition module, at least one serial sensing unit, and a motion classification module; The series sensing unit is disposed in the single-fin area of ​​the flexible glove substrate and is composed of a pressure sensing unit and a strain sensing unit connected in series. The pressure sensing unit includes a first conductive sponge substrate and parallel plate electrodes. The interior of the first conductive sponge substrate is an unfilled three-dimensional porous structure, and the parallel plate electrodes are respectively attached and fixed to the upper and lower end faces of the first conductive sponge substrate. The strain sensing unit includes a second conductive sponge substrate, an organosilicon elastomer, and lateral strip electrodes. The organosilicon elastomer completely fills the internal pores of the second conductive sponge substrate, and the lateral strip electrodes are respectively attached and fixed to the left and right side end faces of the second conductive sponge substrate. Both the first conductive sponge matrix and the second conductive sponge matrix are polydopamine-modified multi-walled carbon nanotube / polyurethane conductive sponge structures. The serial sensing unit is connected to the acquisition channel of the signal acquisition module via a lead wire, and the output terminal of the signal acquisition module is connected to the signal of the motion classification module.

2. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The pressure sensing unit is attached and fixed to the fingertip area of ​​the single finger region of the flexible glove substrate, and a flexible pressure equalizing pad is provided on the side of the pressure sensing unit facing the skin of the human fingertip.

3. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The parallel plate electrode is a double-layer conductive cloth holding wire structure, and a conductive carbon paste layer is provided on the bonding surface between the conductive cloth and the first conductive sponge substrate.

4. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The strain sensing unit is attached and fixed to the back of the interphalangeal joint in a single finger area of ​​the flexible glove substrate, and is bonded to the glove fiber layer by an organosilicon elastomer.

5. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The silicone elastomer of the strain sensing unit is formed by vacuum infiltration process, specifically: The silicone prepolymer and curing agent are mixed in a preset ratio and degassed under vacuum. The second conductive sponge matrix is ​​then immersed in the mixed adhesive solution and kept under negative pressure for a preset time until the adhesive solution fills the internal pores of the second conductive sponge matrix. Remove the soaked second conductive sponge substrate, remove excess adhesive from the surface, and heat to cure, forming an organosilicon elastomer within the pores.

6. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The preparation method of the polydopamine-modified multi-walled carbon nanotube / polyurethane conductive sponge includes: The polyurethane foam substrate was pretreated with hydrophilic modification. Dopamine hydrochloride was added to Tris-HCl buffer to form a mixed solution with a pH of 8.5, which yielded a dopamine reaction solution. The pretreated polyurethane sponge was placed in the dopamine reaction solution to react and generate a polydopamine modified layer on the surface of the sponge. A polyurethane sponge substrate with a polydopamine-modified layer was immersed in a carboxylated multi-walled carbon nanotube dispersion for reaction. The reaction was then carried out by gradient drying to fix the multi-walled carbon nanotubes and polydopamine through covalent bonds.

7. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The signal acquisition module includes data enhancement after acquiring the time-domain signal, specifically: The time axis of the time-domain signal is linearly scaled while keeping the signal amplitude and waveform unchanged. Only the overall duration of the signal is adjusted to simulate the signal differences at different action rates. Gaussian noise with a preset amplitude difference is superimposed on the time-domain signal after scaling along the time axis to simulate force fluctuations.

8. The single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 1, characterized in that, The action classification module has a built-in pre-trained classification model, which is used to extract temporal features from the time-domain signal transmitted by the signal acquisition module, and to identify the grip quality based on the temporal features. The time-series characteristics include the maximum slope of the bending rise, the maximum value of the second derivative, the minimum value of the second derivative, the peak-to-peak value, the relative peak value, and the bending-squeeze switching time.

9. A single-channel self-decoupling flexible multimodal hand rehabilitation sensing system according to claim 8, characterized in that, The grip quality includes normal grip, insufficient bending, insufficient force, and excessive force.

Images