Intent-Aware Rehabilitation Robot Data Acquisition Method

Abstract

This invention relates to a data acquisition method for rehabilitation robots based on intent perception, belonging to the field of rehabilitation robot technology. This invention performs pre-simulation and analysis of rehabilitation exercise comfort based on the rehabilitation robot's motion parameters and a patient-specific biomechanical intent model to obtain the patient's rehabilitation exercise comfort. Finally, based on the patient's rehabilitation exercise comfort, the motion parameters of the rehabilitation robot are analyzed and configured, and the control of the rehabilitation robot is optimized according to the final motion parameters. This invention processes the noise data collected by sensors using a Kal's filtering algorithm, thereby improving the accuracy of user intent recognition. Furthermore, it can perform motion simulation analysis on the processed real-time muscle strength signal characteristics of the patient using virtual reality technology and digital twin technology, which can optimize the control of the rehabilitation robot, ensuring user comfort during rehabilitation and ensuring that the patient's exercise volume is within the predetermined range.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation robot data technology, and in particular to a method for acquiring rehabilitation robot data based on intent perception. Background Technology

[0002] Rehabilitation training plays a vital role in the early rehabilitation of stroke patients with hemiplegia and in patients with severe injuries. Previous methods primarily focused on gait training and standing balance training. While conventional rehabilitation training can improve limb motor function and balance to some extent by training gait and balance, the effects are not significant and lack necessary feedback exercises, easily leading to poor gait development. Rehabilitation robots, intelligent lower limb feedback rehabilitation training systems, combine movement patterns and visual feedback to trigger lower limb motor function, enabling patients to regain walking ability, and have high application value. However, rehabilitation robots often rely on sensors for data collection, and the data collected by sensors often contains noise, hindering the perception of the user's intentions and preventing better coordination and assistance between the robot and the user. Secondly, current technologies do not consider user comfort during rehabilitation exercises, resulting in a poor user experience. Summary of the Invention

[0003] This invention overcomes the shortcomings of the prior art and provides a data acquisition method for rehabilitation robots based on intent perception.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of this invention provides a data acquisition method for a rehabilitation robot based on intent perception, characterized by comprising the following steps:

[0006] By installing a flexible piezoresistive sensor at the patient's joint, the patient's real-time muscle force signal characteristic data is collected through the flexible piezoresistive sensor, and the real-time muscle force signal characteristic data of the patient is processed by Kalman filtering.

[0007] The real-time muscle strength signal feature data of the processed patient is obtained, and a patient-specific biomechanical intent model is constructed based on the real-time muscle strength signal feature data of the processed patient.

[0008] Initialize the motion parameters of the rehabilitation robot, and perform rehabilitation exercise comfort pre-simulation and analysis based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model to obtain the patient's rehabilitation exercise comfort.

[0009] The motion parameters of the rehabilitation robot are analyzed and configured based on the patient's rehabilitation exercise comfort, and the final motion parameters of the rehabilitation robot are obtained. The control of the rehabilitation robot is then optimized according to the final motion parameters of the rehabilitation robot.

[0010] Furthermore, in this method, a flexible piezoresistive sensor is installed at the patient's joint to collect real-time muscle force signal characteristic data. This real-time muscle force signal characteristic data is then processed using a Kalman filter, specifically:

[0011] By setting flexible piezoresistive sensors on the patient's joints, a flexible piezoresistive sensor sequence is formed, and the patient's real-time muscle force signal characteristic data is collected through the flexible piezoresistive sensors to determine the effective range of muscle force signal characteristic data.

[0012] The system state vector is set based on the effective range of muscle strength signal feature data using the Kalman filter method, the real-time muscle strength signal feature data of the patient is set as the observation vector, and the system model is set based on the system state vector.

[0013] An observation model is set up based on the observation vector, the state estimate and error covariance matrix at the current moment are predicted based on the system model and the observation model, and the Kalman gain is calculated based on the state estimate and error covariance matrix at the current moment.

[0014] Based on the Kalman gain update state estimate and covariance matrix, adaptive filtering is performed using the weights updated by the Kalman filter to complete the denoising process of the patient's real-time muscle strength signal feature data.

[0015] Furthermore, in this method, the processed real-time muscle strength signal feature data of the patient is obtained, and a patient-specific biomechanical intent model is constructed based on the processed real-time muscle strength signal feature data, specifically as follows:

[0016] The real-time muscle strength signal feature data of the patient after processing is obtained, and a biological joint model of the patient is constructed using three-dimensional modeling software. The real-time muscle strength signal feature data of the patient after processing is then pre-simulated in the biological joint model of the patient.

[0017] Based on the processed real-time muscle force signal characteristic data of the patient, the finite element analysis method is used to pre-determine the force distribution of the tissue, obtain a schematic diagram of the force distribution of the tissue, and construct a patient-specific biomechanical intent model based on the schematic diagram of the force distribution of the tissue.

[0018] Furthermore, in this method, the motion parameters of the rehabilitation robot are initialized, and based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model, a rehabilitation exercise comfort simulation and analysis are performed to obtain the patient's rehabilitation exercise comfort. Specifically:

[0019] Set exercise time thresholds, obtain the tissue stress threshold range of patients in each rehabilitation state through big data, and set comfort evaluation index characteristics based on the tissue stress threshold range and exercise time threshold of patients in each rehabilitation state.

[0020] The motion parameters of the rehabilitation robot, including its speed, torque, and angle, are initialized and configured based on the patient-specific biomechanical intent model.

[0021] Motion pre-simulation is performed based on motion parameters of the rehabilitation robot, including its speed, torque, and angle.

[0022] By performing exercise rehearsals within the exercise time threshold, and evaluating the patient's specific biomechanical intention model based on the comfort evaluation index characteristics, the patient's rehabilitation exercise comfort is obtained.

[0023] Furthermore, in this method, a motion time threshold is set, specifically as follows:

[0024] The optimal exercise time of patients in each rehabilitation stage is obtained by big data, and a knowledge graph is constructed. A graph neural network is introduced, and the optimal exercise time of patients in each rehabilitation stage is input into the graph neural network for processing.

[0025] The rehabilitation status and optimal exercise time are used as graph nodes in the graph neural network. A directed descriptive relationship is constructed, and the graph nodes are connected based on the directed descriptive relationship to construct a topology graph. The topology graph is then input into the knowledge graph for node representation.

[0026] The current patient's rehabilitation status information is obtained, and the current patient's rehabilitation status information is input into the knowledge graph for analysis. The optimal exercise time for the patient in the current rehabilitation process is obtained, and an exercise time threshold is set based on the optimal exercise time for the patient in the current rehabilitation process.

[0027] Furthermore, in this method, the motion parameters of the rehabilitation robot are analyzed and configured based on the patient's rehabilitation exercise comfort to obtain the final motion parameters of the rehabilitation robot, specifically as follows:

[0028] A particle swarm optimization algorithm is introduced, and the number of iterations is set based on the particle swarm optimization algorithm to determine whether the patient's rehabilitation exercise comfort is not lower than the preset rehabilitation exercise comfort evaluation index.

[0029] When the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index, the patient will exercise according to the current rehabilitation robot's motion parameters.

[0030] When the patient's rehabilitation exercise comfort level is lower than the preset rehabilitation exercise comfort evaluation index, the motion parameters of the rehabilitation robot are reset based on the number of iterations until the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index.

[0031] A second aspect of the present invention provides a data acquisition system for a rehabilitation robot based on intent perception, including a memory and a processor. The memory includes a program for a data acquisition method for a rehabilitation robot based on intent perception. When the program for a data acquisition method for a rehabilitation robot based on intent perception is executed by the processor, it implements the steps of the data acquisition method for a rehabilitation robot based on intent perception as described in any one of the present invention.

[0032] A third aspect of the present invention provides a computer-readable storage medium including a data acquisition method program for an intention-aware rehabilitation robot, wherein when the intention-aware rehabilitation robot data acquisition method program is executed by a processor, it implements the steps of the intention-aware rehabilitation robot data acquisition method described in any one of the present invention.

[0033] This invention addresses the shortcomings of the prior art and has the following beneficial effects:

[0034] This invention utilizes flexible piezoresistive sensors placed at the patient's joints to collect real-time muscle force signal characteristic data. This data is then processed using Kalman filtering to obtain processed real-time muscle force signal characteristic data. Based on this processed data, a patient-specific biomechanical intention model is constructed to initialize the motion parameters of a rehabilitation robot. Based on these parameters and the patient-specific biomechanical intention model, a pre-simulation and analysis of rehabilitation exercise comfort is performed to obtain the patient's rehabilitation exercise comfort level. Finally, based on this comfort level, the robot's motion parameters are analyzed and configured to obtain the final motion parameters. The robot's control is then optimized according to these final parameters. This invention improves the accuracy of user intention recognition by processing the noise data collected by the sensors using the Kalman filtering algorithm. Furthermore, it optimizes the control of the rehabilitation robot by using virtual reality and digital twin technologies to perform motion simulation analysis on the processed real-time muscle force signal characteristics, ensuring user comfort during rehabilitation and keeping the patient's exercise intensity within the predetermined range. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0036] Figure 1 A flowchart illustrating the overall process of data acquisition for rehabilitation robots based on intent awareness is provided.

[0037] Figure 2 A system block diagram of an intent-aware rehabilitation robot data acquisition system is shown. Detailed Implementation

[0038] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0040] like Figure 1 As shown, the first aspect of the present invention provides a data acquisition method for a rehabilitation robot based on intent perception, characterized by comprising the following steps:

[0041] S102: By setting a flexible piezoresistive sensor at the patient's joint, the patient's real-time muscle force signal characteristic data is collected through the flexible piezoresistive sensor, and the patient's real-time muscle force signal characteristic data is processed by Kalman filtering.

[0042] S104: Obtain the processed real-time muscle strength signal feature data of the patient, and construct a patient-specific biomechanical intent model based on the processed real-time muscle strength signal feature data of the patient;

[0043] S106: Initialize the motion parameters of the rehabilitation robot, perform rehabilitation exercise comfort pre-simulation and analysis based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model, and obtain the patient's rehabilitation exercise comfort.

[0044] S108: Analyze and configure the motion parameters of the rehabilitation robot based on the patient's rehabilitation exercise comfort, obtain the final motion parameters of the rehabilitation robot, and optimize the control of the rehabilitation robot according to the final motion parameters of the rehabilitation robot.

[0045] It should be noted that this invention processes the noise data collected by the sensor using the Karl Filter algorithm, thereby improving the accuracy of recognizing the user's intention. Secondly, it can perform motion simulation analysis on the processed real-time muscle strength signal characteristics of the patient using virtual reality technology and digital twin technology, which can optimize the control of the rehabilitation robot, ensure the user's exercise comfort during the rehabilitation process, and ensure that the patient's exercise volume is within the predetermined range.

[0046] Furthermore, in this method, a flexible piezoresistive sensor is installed at the patient's joint to collect real-time muscle force signal characteristic data. This data is then processed using Kalman filtering, specifically:

[0047] By placing flexible piezoresistive sensors on the patient's joints, a sequence of flexible piezoresistive sensors is formed, and real-time muscle force signal characteristic data of the patient is collected through the flexible piezoresistive sensors to determine the effective range of muscle force signal characteristic data.

[0048] The system state vector is set based on the effective range of muscle force signal feature data using the Kalman filter method, the real-time muscle force signal feature data of the patient is set as the observation vector, and the system model is set based on the system state vector.

[0049] The observation model is set based on the observation vector, the state estimate and error covariance matrix at the current moment are predicted based on the system model and the observation model, and the Kalman gain is calculated based on the state estimate and error covariance matrix at the current moment.

[0050] Based on the Kalman gain update state estimation and covariance matrix, adaptive filtering is performed by combining the weights updated by Kalman filtering to complete the denoising process of the patient's real-time muscle strength signal feature data.

[0051] It should be noted that the Kalman filter method can denoise the patient's real-time muscle strength signal feature data, thereby improving the accuracy of recognizing the patient's movement intention.

[0052] Furthermore, in this method, real-time muscle strength signal feature data of the processed patient is obtained, and a patient-specific biomechanical intent model is constructed based on the processed real-time muscle strength signal feature data, specifically as follows:

[0053] The real-time muscle strength signal feature data of the processed patient is obtained, and a biological joint model of the patient is constructed using 3D modeling software (including virtual reality technology, digital twin technology, etc.). The real-time muscle strength signal feature data of the processed patient is then pre-simulated in the biological joint model of the patient.

[0054] Based on the processed real-time muscle force signal characteristic data of the patient, the finite element analysis method is used to pre-determine the force distribution of the tissue, obtain a schematic diagram of the force distribution of the tissue, and construct a patient-specific biomechanical intent model based on the schematic diagram of the force distribution of the tissue.

[0055] It should be noted that the finite element analysis method is used to simulate the force distribution of the patient's joint tissues, thereby constructing a patient-specific biomechanical intent model. The patient-specific biomechanical intent model can display a schematic diagram of the force distribution of various tissues in the patient, thus enabling dynamic display.

[0056] Furthermore, in this method, the motion parameters of the rehabilitation robot are initialized, and based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model, a rehabilitation exercise comfort simulation and analysis are performed to obtain the patient's rehabilitation exercise comfort. Specifically:

[0057] Set exercise time thresholds, obtain the range of tissue stress thresholds of patients in each rehabilitation state through big data, and set comfort evaluation index characteristics based on the range of tissue stress thresholds and exercise time thresholds of patients in each rehabilitation state.

[0058] The motion parameters of the rehabilitation robot, including motion speed, motion torque, and motion angle, are initialized and configured based on the patient's specific biomechanical intent model.

[0059] Motion pre-simulation is performed based on motion parameters of the rehabilitation robot, including its speed, torque, and angle.

[0060] By rehearsing the exercise within a time threshold, the patient's specific biomechanical intention model is evaluated for comfort based on the characteristics of comfort evaluation indicators, thereby obtaining the patient's rehabilitation exercise comfort.

[0061] It should be noted that the rehabilitation state includes the initial rehabilitation state, the semi-recovery rehabilitation state, the highly recovered rehabilitation state, and the fully recovered rehabilitation state. This method is used to obtain the patient's rehabilitation exercise comfort, so as to better determine the impact of the current rehabilitation robot's motion parameters, such as motion speed, motion torque information, and motion angle information, on the user's exercise comfort.

[0062] Furthermore, in this method, a motion time threshold is set, specifically as follows:

[0063] By using big data to obtain the optimal exercise time of patients in each rehabilitation stage, and constructing a knowledge graph, a graph neural network is introduced to process the optimal exercise time of patients in each rehabilitation stage.

[0064] The rehabilitation status and optimal exercise time are used as graph nodes in a graph neural network. Directed descriptive relations are constructed, and graph nodes are connected based on the directed descriptive relations to construct a topology graph. The topology graph is then input into a knowledge graph for node representation.

[0065] Obtain the current patient's rehabilitation status information, input the current patient's rehabilitation status information into the knowledge graph for analysis, obtain the optimal exercise time for the patient in the current rehabilitation process, and set the exercise time threshold based on the optimal exercise time for the patient in the current rehabilitation process.

[0066] It should be noted that patients in different recovery stages have different optimal exercise times. This method can further optimize the exercise time threshold and improve the rationality of comfort evaluation.

[0067] Furthermore, in this method, the motion parameters of the rehabilitation robot are analyzed and configured based on the patient's rehabilitation exercise comfort to obtain the final motion parameters of the rehabilitation robot, specifically:

[0068] Particle swarm optimization (PSO) algorithm is introduced, and the number of iterations is set based on the PSO algorithm to determine whether the patient's rehabilitation exercise comfort is not lower than the preset rehabilitation exercise comfort evaluation index.

[0069] When the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index, the exercise is performed according to the current rehabilitation robot's motion parameters;

[0070] When the patient's rehabilitation exercise comfort level is lower than the preset rehabilitation exercise comfort evaluation index, the motion parameters of the rehabilitation robot are reset based on the number of iterations until the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index.

[0071] It's important to note that Particle Swarm Optimization (PSO), also known as the microparticle swarm algorithm, originates from a simulation of a simplified social model. The algorithm is based on the foraging process of birds. A core mechanism involves each bird foraging independently and remembering the closest position to food. By communicating with other birds, they obtain the known best position for the entire flock and guide the flock to continue searching in that direction. Two key settings are: the historical best position of each particle (pBest vector) and the historical best position of the entire group (gBest vector). Here, pBest vector is a set of vectors containing the historical best position of each particle, and gBest vector is the vector with the highest fitness value among the pBest vectors, i.e., the global optimum. Note: The algorithm generally uses the objective function to be optimized as the fitness function, evaluates the fitness value, and then updates the pBest and gBest vectors. This method can optimize the motion parameters of rehabilitation robots, ensuring the comfort of users' rehabilitation exercises and avoiding overexertion.

[0072] In addition, this method also includes:

[0073] A Bayesian network is introduced to obtain characteristic data of changes in patients' rehabilitation exercise comfort within a preset time period, and the characteristic data of changes in patients' rehabilitation exercise comfort within the preset time period is input into the Bayesian network;

[0074] The rehabilitation exercise comfort feature data of each time point in the patient's rehabilitation exercise comfort change feature data within the preset time period is used as the observation vector in the Bayesian network, and the transition probability value of the observation vector to another observation vector is calculated.

[0075] Determine whether the transition probability value is greater than a preset transition probability value. If the transition probability value is not greater than the preset transition probability value, maintain the current observation vector unchanged. If the transition probability value is greater than the preset transition probability value, update it to another observation vector.

[0076] Obtain the updated observation vector, and update the patient's rehabilitation exercise comfort change feature data within a preset time period based on the updated observation vector.

[0077] It should be noted that updating the patient's rehabilitation exercise comfort status through Bayesian networks can promptly update the user's comfort level, allowing the rehabilitation robot to be controlled more closely to the user's exercise intentions.

[0078] like Figure 2As shown, a second aspect of the present invention provides a data acquisition system 4 for a rehabilitation robot based on intent perception, including a memory 41 and a processor 42. The memory 41 includes a data acquisition method program for a rehabilitation robot based on intent perception. When the data acquisition method program for a rehabilitation robot based on intent perception is executed by the processor 42, the following steps are implemented:

[0079] By installing a flexible piezoresistive sensor at the patient's joint, the patient's real-time muscle force signal characteristic data is collected through the flexible piezoresistive sensor, and the real-time muscle force signal characteristic data is processed by the Kalman filter method;

[0080] The real-time muscle strength signal feature data of the processed patients were acquired, and a patient-specific biomechanical intent model was constructed based on the real-time muscle strength signal feature data of the processed patients.

[0081] Initialize the motion parameters of the rehabilitation robot, and perform rehabilitation exercise comfort pre-simulation and analysis based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model to obtain the patient's rehabilitation exercise comfort.

[0082] The motion parameters of the rehabilitation robot are analyzed and configured based on the patient's rehabilitation exercise comfort, and the final motion parameters of the rehabilitation robot are obtained. The control of the rehabilitation robot is then optimized according to the final motion parameters.

[0083] It should be noted that this invention processes the noise data collected by the sensor using the Karl Filter algorithm, thereby improving the accuracy of recognizing the user's intention. Secondly, it can perform motion simulation analysis on the processed real-time muscle strength signal characteristics of the patient using virtual reality technology and digital twin technology, which can optimize the control of the rehabilitation robot, ensure the user's exercise comfort during the rehabilitation process, and ensure that the patient's exercise volume is within the predetermined range.

[0084] Furthermore, in this system, a flexible piezoresistive sensor is installed at the patient's joint to collect real-time muscle force signal characteristic data. This data is then processed using Kalman filtering, specifically:

[0085] By placing flexible piezoresistive sensors on the patient's joints, a sequence of flexible piezoresistive sensors is formed, and real-time muscle force signal characteristic data of the patient is collected through the flexible piezoresistive sensors to determine the effective range of muscle force signal characteristic data.

[0086] The system state vector is set based on the effective range of muscle force signal feature data using the Kalman filter method, the real-time muscle force signal feature data of the patient is set as the observation vector, and the system model is set based on the system state vector.

[0087] The observation model is set based on the observation vector, the state estimate and error covariance matrix at the current moment are predicted based on the system model and the observation model, and the Kalman gain is calculated based on the state estimate and error covariance matrix at the current moment.

[0088] Based on the Kalman gain update state estimation and covariance matrix, adaptive filtering is performed by combining the weights updated by Kalman filtering to complete the denoising process of the patient's real-time muscle strength signal feature data.

[0089] It should be noted that the Kalman filter method can denoise the patient's real-time muscle strength signal feature data, thereby improving the accuracy of recognizing the patient's movement intention.

[0090] Furthermore, in this system, processed real-time muscle strength signal feature data of the patient is acquired, and a patient-specific biomechanical intent model is constructed based on the processed real-time muscle strength signal feature data, specifically as follows:

[0091] The real-time muscle strength signal feature data of the processed patient is obtained, and a biological joint model of the patient is constructed using 3D modeling software (including virtual reality technology, digital twin technology, etc.). The real-time muscle strength signal feature data of the processed patient is then pre-simulated in the biological joint model of the patient.

[0092] Based on the processed real-time muscle force signal characteristic data of the patient, the finite element analysis method is used to pre-determine the force distribution of the tissue, obtain a schematic diagram of the force distribution of the tissue, and construct a patient-specific biomechanical intent model based on the schematic diagram of the force distribution of the tissue.

[0093] It should be noted that the finite element analysis method is used to simulate the force distribution of the patient's joint tissues, thereby constructing a patient-specific biomechanical intent model. The patient-specific biomechanical intent model can display a schematic diagram of the force distribution of various tissues in the patient, thus enabling dynamic display.

[0094] Furthermore, in this system, the motion parameters of the rehabilitation robot are initialized, and based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model, a rehabilitation exercise comfort simulation and analysis are performed to obtain the patient's rehabilitation exercise comfort level. Specifically:

[0095] Set exercise time thresholds, obtain the range of tissue stress thresholds of patients in each rehabilitation state through big data, and set comfort evaluation index characteristics based on the range of tissue stress thresholds and exercise time thresholds of patients in each rehabilitation state.

[0096] The motion parameters of the rehabilitation robot, including its speed, torque, and angle, are initialized based on the patient's specific biomechanical intent model.

[0097] Motion pre-simulation is performed based on motion parameters of the rehabilitation robot, including its speed, torque, and angle.

[0098] By rehearsing the exercise within a time threshold, the patient's specific biomechanical intention model is evaluated for comfort based on the characteristics of comfort evaluation indicators, thereby obtaining the patient's rehabilitation exercise comfort.

[0099] It should be noted that the rehabilitation state includes initial rehabilitation state, semi-recovery rehabilitation state, high recovery rehabilitation state, and full recovery rehabilitation state. This system is used to obtain the patient's rehabilitation exercise comfort, so as to better determine the impact of the current rehabilitation robot's motion parameters, such as motion speed, motion torque information, and motion angle information, on the user's exercise comfort.

[0100] Furthermore, in this system, a motion time threshold is set, specifically as follows:

[0101] By using big data to obtain the optimal exercise time of patients in each rehabilitation stage, and constructing a knowledge graph, a graph neural network is introduced to process the optimal exercise time of patients in each rehabilitation stage.

[0102] The rehabilitation status and optimal exercise time are used as graph nodes in a graph neural network. Directed descriptive relations are constructed, and graph nodes are connected based on the directed descriptive relations to construct a topology graph. The topology graph is then input into a knowledge graph for node representation.

[0103] Obtain the current patient's rehabilitation status information, input the current patient's rehabilitation status information into the knowledge graph for analysis, obtain the optimal exercise time for the patient in the current rehabilitation process, and set the exercise time threshold based on the optimal exercise time for the patient in the current rehabilitation process.

[0104] It should be noted that patients in different rehabilitation stages have different optimal exercise times. This system can further optimize exercise time thresholds and improve the rationality of comfort evaluation.

[0105] Furthermore, in this system, the motion parameters of the rehabilitation robot are analyzed and configured based on the patient's rehabilitation exercise comfort to obtain the final motion parameters of the rehabilitation robot, specifically:

[0106] Particle swarm optimization (PSO) algorithm is introduced, and the number of iterations is set based on the PSO algorithm to determine whether the patient's rehabilitation exercise comfort is not lower than the preset rehabilitation exercise comfort evaluation index.

[0107] When the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index, the exercise is performed according to the current rehabilitation robot's motion parameters;

[0108] When the patient's rehabilitation exercise comfort level is lower than the preset rehabilitation exercise comfort evaluation index, the motion parameters of the rehabilitation robot are reset based on the number of iterations until the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index.

[0109] It's important to note that Particle Swarm Optimization (PSO), also known as the microparticle swarm algorithm, originates from a simulation of a simplified social model. The PSO algorithm is based on the foraging process of birds. A core mechanism is that each bird forages independently and remembers the closest position to food. By communicating with other birds, it obtains the known best position for the entire flock and guides the flock to continue searching in that direction. Two other key settings are: the historical best position of each particle (pBest vector) and the historical best position of the entire group (gBest vector). Here, the pBest vector is a set of vectors containing the historical best position of each particle, and the gBest vector is the vector with the highest fitness value among the pBest vectors, i.e., the global optimum. Note: In the algorithm, the objective function to be optimized is generally used as the fitness function. The fitness value is evaluated, and then the pBest and gBest vectors are updated. This system can optimize the motion parameters of rehabilitation robots, ensuring the comfort of users' rehabilitation exercises and avoiding overexertion.

[0110] In addition, this system also includes:

[0111] A Bayesian network is introduced to obtain characteristic data of changes in patients' rehabilitation exercise comfort within a preset time period, and the characteristic data of changes in patients' rehabilitation exercise comfort within the preset time period is input into the Bayesian network;

[0112] The rehabilitation exercise comfort feature data of each time point in the patient's rehabilitation exercise comfort change feature data within the preset time period is used as the observation vector in the Bayesian network, and the transition probability value of the observation vector to another observation vector is calculated.

[0113] Determine whether the transition probability value is greater than a preset transition probability value. If the transition probability value is not greater than the preset transition probability value, maintain the current observation vector unchanged. If the transition probability value is greater than the preset transition probability value, update it to another observation vector.

[0114] Obtain the updated observation vector, and update the patient's rehabilitation exercise comfort change feature data within a preset time period based on the updated observation vector.

[0115] It should be noted that updating the patient's rehabilitation exercise comfort status through Bayesian networks can promptly update the user's comfort level, allowing the rehabilitation robot to be controlled more closely to the user's exercise intentions.

[0116] A third aspect of the present invention provides a computer-readable storage medium including a program for an intention-aware rehabilitation robot data acquisition method, wherein when the program is executed by a processor, it implements the steps of any one of the intention-aware rehabilitation robot data acquisition methods.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0118] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0119] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0120] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0121] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0122] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A data acquisition method for rehabilitation robots based on intent perception, characterized in that, Includes the following steps: By installing a flexible piezoresistive sensor at the patient's joint, the patient's real-time muscle force signal characteristic data is collected through the flexible piezoresistive sensor, and the real-time muscle force signal characteristic data of the patient is processed by Kalman filtering. The real-time muscle strength signal feature data of the processed patient is obtained, and a patient-specific biomechanical intent model is constructed based on the real-time muscle strength signal feature data of the processed patient. Initialize the motion parameters of the rehabilitation robot, and perform rehabilitation exercise comfort pre-simulation and analysis based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model to obtain the patient's rehabilitation exercise comfort. The motion parameters of the rehabilitation robot are analyzed and configured based on the patient's rehabilitation exercise comfort, and the final motion parameters of the rehabilitation robot are obtained. The control of the rehabilitation robot is then optimized according to the final motion parameters of the rehabilitation robot. The processed real-time muscle strength signal feature data of the patient is acquired, and a patient-specific biomechanical intent model is constructed based on the processed real-time muscle strength signal feature data, specifically as follows: The real-time muscle strength signal feature data of the patient after processing is obtained, and a biological joint model of the patient is constructed using three-dimensional modeling software. The real-time muscle strength signal feature data of the patient after processing is then pre-simulated in the biological joint model of the patient. Based on the processed real-time muscle force signal feature data of the patient, the finite element analysis method is used to pre-determine the force distribution of the tissue, obtain a schematic diagram of the force distribution of the tissue, and construct a patient-specific biomechanical intent model based on the schematic diagram of the force distribution of the tissue. Initialize the motion parameters of the rehabilitation robot, and based on the motion parameters of the rehabilitation robot and the patient-specific biomechanical intention model, perform rehabilitation exercise comfort pre-simulation and analysis to obtain the patient's rehabilitation exercise comfort level. Specifically: Set exercise time thresholds, obtain the tissue stress threshold range of patients in each rehabilitation state through big data, and set comfort evaluation index characteristics based on the tissue stress threshold range and exercise time threshold of patients in each rehabilitation state. The motion parameters of the rehabilitation robot, including its speed, torque, and angle, are initialized and configured based on the patient-specific biomechanical intent model. Motion pre-simulation is performed based on motion parameters of the rehabilitation robot, including its speed, torque, and angle. By performing exercise rehearsals within the exercise time threshold, and evaluating the patient's specific biomechanical intention model based on the comfort evaluation index characteristics, the patient's rehabilitation exercise comfort is obtained. Set the exercise time threshold as follows: The optimal exercise time of patients in each rehabilitation stage is obtained by big data, and a knowledge graph is constructed. A graph neural network is introduced, and the optimal exercise time of patients in each rehabilitation stage is input into the graph neural network for processing. The rehabilitation status and optimal exercise time are used as graph nodes in the graph neural network. A directed descriptive relationship is constructed, and the graph nodes are connected based on the directed descriptive relationship to construct a topology graph. The topology graph is then input into the knowledge graph for node representation. The current patient's rehabilitation status information is obtained, and the current patient's rehabilitation status information is input into the knowledge graph for analysis. The optimal exercise time for the patient in the current rehabilitation process is obtained, and an exercise time threshold is set based on the optimal exercise time for the patient in the current rehabilitation process.

2. The data acquisition method for rehabilitation robots based on intent perception according to claim 1, characterized in that, By installing a flexible piezoresistive sensor at the patient's joint, real-time muscle force signal characteristic data of the patient is collected through the flexible piezoresistive sensor. The real-time muscle force signal characteristic data of the patient is then processed using the Kalman filter method, specifically as follows: By setting flexible piezoresistive sensors on the patient's joints, a flexible piezoresistive sensor sequence is formed, and the patient's real-time muscle force signal characteristic data is collected through the flexible piezoresistive sensors to determine the effective range of muscle force signal characteristic data. The system state vector is set based on the effective range of muscle strength signal feature data using the Kalman filter method, the real-time muscle strength signal feature data of the patient is set as the observation vector, and the system model is set based on the system state vector. An observation model is set up based on the observation vector, the state estimate and error covariance matrix at the current moment are predicted based on the system model and the observation model, and the Kalman gain is calculated based on the state estimate and error covariance matrix at the current moment. Based on the Kalman gain update state estimate and covariance matrix, adaptive filtering is performed using the weights updated by the Kalman filter to complete the denoising process of the patient's real-time muscle strength signal feature data.

3. The data acquisition method for rehabilitation robots based on intent perception according to claim 1, characterized in that, Based on the patient's rehabilitation exercise comfort, the motion parameters of the rehabilitation robot are analyzed and configured to obtain the final motion parameters of the rehabilitation robot, specifically: A particle swarm optimization algorithm is introduced, and the number of iterations is set based on the particle swarm optimization algorithm to determine whether the patient's rehabilitation exercise comfort is not lower than the preset rehabilitation exercise comfort evaluation index. When the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index, the patient will exercise according to the current rehabilitation robot's motion parameters. When the patient's rehabilitation exercise comfort level is lower than the preset rehabilitation exercise comfort evaluation index, the motion parameters of the rehabilitation robot are reset based on the number of iterations until the patient's rehabilitation exercise comfort level is not lower than the preset rehabilitation exercise comfort evaluation index.

4. A data acquisition system for rehabilitation robots based on intent perception, characterized in that, The device includes a memory and a processor. The memory includes a program for an intention-aware rehabilitation robot data acquisition method. When the program for the intention-aware rehabilitation robot data acquisition method is executed by the processor, it implements the steps of the intention-aware rehabilitation robot data acquisition method as described in any one of claims 1-3.

5. A computer-readable storage medium, characterized in that, The method includes a data acquisition method program for a rehabilitation robot based on intent perception. When the data acquisition method program for a rehabilitation robot based on intent perception is executed by a processor, it implements the steps of the data acquisition method for a rehabilitation robot based on intent perception as described in any one of claims 1-3.

Images