Hand rehabilitation device and system based on multi-source data analysis

Abstract

The present application relates to the technical field of limb training equipment, and particularly relates to a hand rehabilitation device and system based on multi-source data analysis. The hand rehabilitation device comprises a wearing glove, a positive and negative pressure air pump module, an electromagnetic valve module, a temperature control module, a human-computer interaction interface and a memory storage module. According to multi-source hand sensing data, an actual intention motion polar coordinate layout of a current hand of a target user is constructed, a standardization dislocation analysis and a multi-dimensional lag compensation actual intention motion polar coordinate layout are performed, and then a virtual execution posture rotation independent angle alternation of the hand rehabilitation device when a stroke driving compensation amount is applied is solved by reverse kinematics, joint flexion and extension control parameters required by the hand rehabilitation device are obtained, and air pressure control cumulative calculation is performed according to the joint flexion and extension control parameters, so that the hand rehabilitation device is intelligently trained and controlled. The present application provides accurate control feedback for the hand rehabilitation device by analyzing multi-source data, effectively improves training efficiency and accuracy, and reduces control error risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of limb training equipment, and in particular to a hand rehabilitation device and system based on multi-source data analysis. BACKGROUND

[0002] With the development of neural rehabilitation medicine and intelligent rehabilitation equipment, hand rehabilitation devices are widely used in the fields of post-stroke hand dysfunction and post-traumatic hand rehabilitation. Traditional hand rehabilitation devices mostly use a single sensor (such as an electromyography sensor, a force sensor, or an angle sensor) for signal acquisition, and drive the mechanical execution part for rehabilitation training through a pre-set program. However, the intended actions of patients are complex and diverse, and individual differences are significant, so a single data source often cannot accurately reflect the true movement intention and recovery state of the patient.

[0003] With the advancement of wearable technology, sensor fusion, and artificial intelligence technology, multi-source data fusion has become a key path to improve the intelligent level of rehabilitation training. By fusing multi-source information such as electromyography (EMG), electroencephalogram, hand posture, joint angle, or physiological parameters, the movement ability of the patient can be more comprehensively evaluated, and rehabilitation strategies can be dynamically adjusted to achieve personalized and refined control. However, existing hand rehabilitation devices still have the problem of relying only on a single type of data, which makes it difficult to deal with the ambiguity and interference in movement intention recognition, and easily leads to deviation of training actions from the patient's intention, making it difficult for the hand rehabilitation device to make reasonable and precise control decisions based on the current hand condition of the target user. Secondly, most existing devices lack real-time perception and analysis of user movement feedback (such as force output, muscle response, or coordinated action), making it difficult to achieve accurate human-machine collaboration and closed-loop control, resulting in training errors such as incoordination, which greatly reduces the training effectiveness and efficiency of the hand. SUMMARY

[0004] The present application overcomes the shortcomings of the prior art and provides a hand rehabilitation device and system based on multi-source data analysis.

[0005] To achieve the above purpose, the technical solution adopted by the present application is as follows:

[0006] The present application provides a hand rehabilitation device based on multi-source data analysis in the first aspect, which comprises:

[0007] Wearable glove: the wearable glove is used to completely wrap the finger joints and palm joints of the target user;

[0008] Positive and negative pressure air pump module: the positive and negative pressure air pump module is used to control the generation of positive pressure gas and negative pressure gas, to provide air pressure power for the auxiliary flexion and extension training activities of the finger joints, and to adjust the training intensity and force;

[0009] Solenoid valve module: the solenoid valve module is connected with the positive and negative pressure air pump module, is responsible for controlling the on-off and flow of compressed air, and simulates the natural gripping action of the hand;

[0010] Temperature control module: the temperature control module is used for adjusting the hand temperature according to the training condition of the target user, and provides a heating effect for hand rehabilitation training;

[0011] Man-machine interaction interface: the man-machine interaction interface is used for providing the target user with the switching of the working mode and the individualized setting of the time, intensity and frequency of the hand training;

[0012] Memory storage module: the memory storage module is responsible for memorizing and saving the last hand training scheme, so that the effect of direct application next time is realized without repeated debugging.

[0013] The second aspect of the application provides a control method of a hand rehabilitation device based on multi-source data analysis, which is applied to the hand rehabilitation device based on multi-source data analysis and includes the following steps:

[0014] S102: acquiring multi-source hand sensing data of the hand rehabilitation device and acquiring a human intention motion system, performing polar coordinate radial fusion on the multi-source hand sensing data according to the human intention motion system, and obtaining an actual intention motion polar coordinate layout of the multi-source hand sensing data;

[0015] S104: according to the actual intention motion polar coordinate layout, performing multi-dimensional lag compensation according to the ideal working condition parameter set standardization dislocation analysis of the human intention motion when the hand rehabilitation device executes the selected working mode, to obtain a stroke driving compensation amount of the hand rehabilitation device intention motion reaching the selected working mode;

[0016] S106: constructing a chain joint motion structure of the target user, synchronously injecting a virtual end effector into a hand rehabilitation device model, and solving the independent angle alternation of each virtual end effector posture rotation when the hand rehabilitation device applies the stroke driving compensation amount according to the chain joint motion structure by using a reverse kinematics algorithm, to obtain joint flexion and extension control parameters of the hand rehabilitation device executing the stroke driving compensation amount;

[0017] S108: based on the rotation posture scalar field derived in the simulation process and the preset training control strategy of the hand rehabilitation device, a quaternion mapping field is built, the joint flexion and extension control parameters are mapped to the quaternion mapping field and executed, and air pressure control cumulative calculation is performed, so as to intelligently train and control the hand rehabilitation device.

[0018] More specifically, the S102 specifically includes the following steps:

[0019] A multi-source sensor array of a hand rehabilitation device is acquired, and when a target user wears the hand rehabilitation device, target user data in a preset time period is collected through the multi-source sensor array to acquire multi-source hand sensing data.

[0020] Each group of multi-source hand sensing data is centrally preprocessed, a covariance algorithm is introduced to calculate self-correlation dependence of each group of centrally preprocessed multi-source hand sensing data, and coupled variance between each data variable in each group of multi-source hand sensing data is output.

[0021] A human anatomy knowledge graph is acquired through big data, the multi-source hand sensing data is identified by using the human anatomy knowledge graph, a human intent motion system conforming to the target user's stage characteristics is output, and different human intent motion indexes and logical specification trend indexes of each human intent motion index are extracted through the human intent motion system.

[0022] The principal component source coordinate angle is allocated in the form of Cartesian coordinates according to different human intent motion indexes to form a human intent motion polar coordinate architecture, a radial joint projection vector constraint is established according to the logical specification trend index, and the coupled variance is taken as an autonomous projection criterion.

[0023] Each group of multi-source hand sensing data is radially decomposed one by one in the covariance algorithm according to the autonomous projection criterion based on the radial joint projection vector constraint, and cross calculation of the collaborative correlation is performed, at this time, the radial covariance scale of each group of multi-source hand sensing data and the cross projection covariance matrix are output.

[0024] If the weighted total sum value of the cross projection covariance matrix exceeds the maximum weighted total sum threshold value, each group of multi-source hand sensing data is projected onto the membership echo radial covariance scale inside the human intent motion polar coordinate architecture to fit the coordinate radius, and an actual intent motion polar coordinate layout of the multi-source hand sensing data is generated.

[0025] More specifically, the S104 specifically includes the following steps:

[0026] The selected working mode of the target user for the hand rehabilitation device and the conceptual design drawing of the hand rehabilitation device are acquired, and the ideal working condition parameter set of the human intent motion of the hand rehabilitation device when executing the selected working mode is acquired through the conceptual design drawing;

[0027] The ideal intent motion polar coordinate layout of the multi-source hand sensing data under the condition of the selected working mode is constructed according to the ideal working condition parameter set, a hash misplacement area function of a non-overlapping area between the actual intent motion polar coordinate layout and the ideal intent motion polar coordinate layout is calculated by introducing a hash misplacement algorithm, and the steady-state misplacement degree of the current hand rehabilitation device for the target user to run the selected working mode is determined according to the hash misplacement area function.

[0028] If the steady-state misalignment is greater than the preset steady-state misalignment, then the intentional motion state space model of the current hand rehabilitation device is constructed based on the coupling correlation of multi-source hand sensing data on different human intentional motion indicators. The open-loop operation transfer function of the current hand rehabilitation device is determined by calculating the zeros and poles to perform decoupling analysis between different input and output channels of the intentional motion state space model.

[0029] Obtain the desired diagonal control system of the selected working mode, determine the compensation cutoff frequency on the Bode plot based on the phase margin of the desired diagonal control system, and establish a gain compensation matrix for different multi-source hand sensing data based on the steady-state misalignment.

[0030] By globally steady-state tuning the gain compensation matrix, the open-loop gain of the motion state space model is adjusted so that the center frequency of the hysteresis network of the open-loop operating transfer function continuously approaches the compensation cutoff frequency, thus generating a multi-dimensional hysteresis compensation structure.

[0031] By deploying a multidimensional hysteresis compensation structure into the actual diagonal control system of the current hand rehabilitation device, the stroke drive compensation amount is obtained when the target user intends to move to the selected point working mode when using the current hand rehabilitation device.

[0032] More specifically, S106 includes the following steps:

[0033] By using a multi-sensor array to mimic the hand features of the target user, the actual joint line contour of the target user is obtained. The joint line contour with the greatest similarity to the actual joint line contour is extracted from the hand feature database preset by the hand rehabilitation device and defined as the joint reference blueprint.

[0034] Obtain standard product drawings of hand rehabilitation equipment, and reconstruct a simulation model of the hand rehabilitation equipment according to the standard product drawings using SolidWorks model simulation software. In the simulation model, inject a virtual end effector into each hand joint.

[0035] Based on the ideal working condition parameter set, the virtual end effector is preset to meet the selected working mode of the termination posture rotation matrix. According to the joint reference blueprint, the joint degrees of freedom, skeletal connection relationship and starting motion angle are defined from the root to the virtual end effector to fit the target user's hand line shape, and the chain joint motion structure is output.

[0036] Acquire personalized settings parameters for the target user when using the selected working mode, and simultaneously acquire the movable points of the current hand rehabilitation device and the real-time joint angles of the target user when maintaining the previous default posture state while capturing multi-source hand sensor data;

[0037] The inverse kinematics algorithm is introduced, and the stroke driving compensation amount is applied to the motion premise constraint of the chain joint motion structure and the personalized setting parameter in the inverse kinematics algorithm from the real-time joint angle of the given origin, so as to deduce the actual position of each virtual end effector, obtain the end posture rotation matrix, calculate the deviation of the end posture rotation matrix and the termination posture rotation matrix, and obtain the matrix error degree of each movable point.

[0038] The matrix error degree is analyzed, the bone binding weight is distributed and weighted transformation rendering is generated, the active trade-off vertex grid is generated, the independent angle alternation of the virtual end effector is solved by using the active trade-off vertex grid, and the joint flexion control parameter of the stroke driving compensation amount of the hand rehabilitation device is obtained.

[0039] More specifically, the matrix error degree is analyzed, the bone binding weight is distributed and weighted transformation rendering is generated, the active trade-off vertex grid is generated, the independent angle alternation of the virtual end effector is solved by using the active trade-off vertex grid, and the joint flexion control parameter of the stroke driving compensation amount of the hand rehabilitation device is obtained. Specifically, the following steps are included:

[0040] If the matrix error degree is less than the preset matrix error degree, the movable point is calibrated as a low-amplitude response active point;

[0041] If the matrix error degree is greater than the preset matrix error degree, the movable point is calibrated as a high-amplitude response active point;

[0042] Based on the matrix error degree, the bone binding weights of different low-amplitude response active points and high-amplitude response active points are distributed, and the active trade-off vertex grid is generated by using the termination posture rotation matrix to weight and transform the rendering of the movable point according to the bone binding weight;

[0043] In the process of weighting and transformation, the Jacobian matrix between the change of joint angle and the change of virtual end effector posture is established by linear approximation, and the joint angle of the active trade-off vertex grid is solved by damped least squares according to the Jacobian matrix, to obtain the independent angle alternation chain of each virtual end effector;

[0044] According to the independent angle alternation chain, the joint angle corresponding to each finger joint of the current hand rehabilitation device is continuously adjusted until the matrix error degrees of the low-amplitude response active points and the high-amplitude response active points are eliminated, and the joint flexion control parameter of the stroke driving compensation amount of the hand rehabilitation device is obtained.

[0045] More specifically, the S108 specifically includes the following steps:

[0046] A preset training control strategy of the hand rehabilitation device is acquired, a plurality of rotation change vectors planned by the hand rehabilitation device for different joint flexion and extension processes are extracted through the preset training control strategy, and a positive and negative air pressure pump is used to realize air pressure control magnitude of each rotation change vector decision output;

[0047] A simulation model of the hand rehabilitation device is used to import the preset training control strategy for simulation, at this time, a rotation posture scalar field of the hand rehabilitation device in the process of executing the stroke driving compensation amount to reach the selected working mode is simulated from the real-time joint angle through the SolidWorks model simulation software, and a quaternion mapping field is built based on the architecture layout of the rotation posture scalar field;

[0048] Each rotation change vector is designated as a prefix label of the quaternion mapping field, the air pressure control magnitude corresponding to each rotation change vector is set as a suffix label, a control query item of the prefix-suffix mapping binding is formed, and each rotation change vector is assigned a quaternion string;

[0049] According to the joint flexion and extension angle, the angular velocity, the angle switching step and the joint flexion and extension increment recorded in the joint flexion and extension control parameter, rotation differential conversion is performed to obtain the necessary rotation change vector of the stroke driving compensation amount of the hand rehabilitation device;

[0050] An index mapping formula is introduced to convert the necessary rotation change vector into an incremental quaternion, obtain the necessary rotation incremental quaternion, project the necessary rotation incremental quaternion to the quaternion mapping field, and make the necessary rotation incremental quaternion traverse each control query item corresponding to the quaternion string one by one for approximate comparison;

[0051] If the approximate coincidence degree of the necessary rotation incremental quaternion and the quaternion string is greater than the preset approximate coincidence degree, a cumulative superposition hat mark is added on the quaternion string; otherwise, the quaternion string is ignored, and a series of candidate superposition quaternions are obtained;

[0052] The air pressure control magnitude of the control query item corresponding to the series of candidate superposition quaternions is rotated and accumulated by using quaternion multiplication to update the posture quaternion of the hand rehabilitation device, finally output the air pressure control accumulation parameter, and the hand rehabilitation device is intelligently trained and controlled based on the air pressure control accumulation parameter.

[0053] The third aspect of the present application provides a hand rehabilitation system based on multi-source data analysis, which comprises a memory and a processor, and the memory stores a hand rehabilitation method program based on multi-source data analysis.

[0054] The present application solves the technical defects in the background art, and has the beneficial technical effects of

[0055] The present application can fuse multi-source sensing information such as electromyographic signals, motion postures, joint angles, and force changes, and can realize real-time perception of the patient's hand movement intention, state, and functional changes. Through positive and negative pressure driving, the training intensity, frequency, and time of the intelligent training scheme can be adjusted, and personalized and phased hand rehabilitation training can be realized. Compared with traditional equipment, the present application has stronger adaptability and feedback control capability, can effectively improve the training efficiency and accuracy, reduce the control risk of manual misoperation, and realize quantitative evaluation and visual tracking of rehabilitation effect. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings of other embodiments can be obtained by those skilled in the art without creative labor.

[0057] Figure 1 A device architecture diagram of a hand rehabilitation device based on multi-source data analysis is shown;

[0058] Figure 2 A first method flowchart of the control method of the present device is shown;

[0059] Figure 3 A second method flowchart of the control method of the present device is shown;

[0060] Figure 4 A system framework diagram of a hand rehabilitation system based on multi-source data analysis is shown. DETAILED DESCRIPTION

[0061] In order to more clearly illustrate the above-mentioned purposes, features and advantages of the present application, the present application will be further described in detail below in combination with the drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0062] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can also be implemented in other ways different from those described herein, therefore, the scope of protection of the present application is not limited by the specific embodiments disclosed below.

[0063] The first aspect of the present application provides a hand rehabilitation device based on multi-source data analysis, as shown in Figure 1 The hand rehabilitation device comprises:

[0064] Wearing gloves 11: the wearing gloves are used for completely wrapping the finger joints and palm joints of the target user;

[0065] Positive and negative pressure air pump module 12: the positive and negative pressure air pump module is used for controlling the generation of positive pressure gas and negative pressure gas, providing air pressure power for the auxiliary flexion and extension training activities of the finger joints, and adjusting the training intensity and strength;

[0066] Solenoid valve module 13: the solenoid valve module is connected with the positive and negative pressure air pump module, and is responsible for controlling the on-off and flow of compressed air, and simulating the natural gripping action of the hand;

[0067] Temperature control module 14: the temperature control module is used for adjusting the hand temperature according to the training condition of the target user, and providing heating effect for hand rehabilitation training;

[0068] Human-computer interaction interface 15: the human-computer interaction interface is used for providing the target user with the switching of working modes and the individualized setting of the time, strength and frequency of hand training;

[0069] Memory storage module 16: the memory storage module is responsible for memorizing and saving the last hand training scheme, realizing the effect of direct application next time without repeated debugging.

[0070] The second aspect of the present application provides a control method of a hand rehabilitation device based on multi-source data analysis, applied to the hand rehabilitation device based on multi-source data analysis, as shown in the figure, including the following steps: Figure 2

[0071] S102: acquiring multi-source hand sensing data of the hand rehabilitation device and acquiring human intention motion system, performing polar coordinate radial fusion of logical specification trend on the multi-source hand sensing data according to the human intention motion system, and obtaining actual intention motion polar coordinate layout of the multi-source hand sensing data;

[0072] S104: according to the actual intention motion polar coordinate layout, the ideal working condition parameter set standardization misplacement analysis of human intention motion when the hand rehabilitation device executes the selected working mode is carried out, multi-dimensional lag compensation is carried out, and the stroke driving compensation amount of the current hand rehabilitation device intention motion reaching the selected working mode is obtained;

[0073] S106: constructing the chain joint motion structure of the target user, synchronously injecting a virtual end effector into the hand rehabilitation device model, and solving the independent angle alternation of each virtual end effector posture rotation when the hand rehabilitation device applies the stroke driving compensation amount according to the chain joint motion structure through the inverse kinematics algorithm, and obtaining the joint flexion and extension control parameters of the hand rehabilitation device executing the stroke driving compensation amount;

[0074] ​S108: Based on the rotation attitude scalar field derived in the simulation process and the preset training control strategy of the hand rehabilitation device, a quaternion mapping field is established, the joint flexion and extension control parameters are mapped to the quaternion mapping field, and air pressure control cumulative calculation is performed, so as to intelligently train and control the hand rehabilitation device.

[0075] More specifically, the S102, as shown, specifically includes the following steps: Figure 3 The S102, as shown, specifically includes the following steps:

[0076] S202: Obtain a multi-source sensor array of a hand rehabilitation device, and when a target user wears the hand rehabilitation device, collect target user data in a preset time period through the multi-source sensor array to obtain multi-source hand sensing data;

[0077] S204: Centralize and preprocess each group of multi-source hand sensing data, introduce a covariance algorithm to calculate self-correlation dependence of each group of centralized and preprocessed multi-source hand sensing data, and output coupling variance between each data variable in each group of multi-source hand sensing data;

[0078] S206: Obtain an anthropometric knowledge graph through big data, identify the multi-source hand sensing data by using the anthropometric knowledge graph, output an anthropometric motion system conforming to the stage characteristics of the target user, and extract different anthropometric motion indexes and logical specification trend indexes of each anthropometric motion index through the anthropometric motion system;

[0079] S208: Assign principal component source coordinate angles in the form of Cartesian coordinates according to different anthropometric motion indexes, form an anthropometric motion polar coordinate architecture, take the coupling variance as an autonomous projection criterion, establish a radial joint projection vector constraint according to the logical specification trend index;

[0080] S210: Based on the radial joint projection vector constraint, each group of multi-source hand sensing data is radially decomposed one by one in the covariance algorithm according to the autonomous projection criterion, and cross calculation of the synergistic correlation is performed, at this time, the radial covariance scale of each group of multi-source hand sensing data and the cross projection covariance matrix are output;

[0081] S212: If the weighted sum value of the cross projection covariance matrix exceeds the maximum weighted sum threshold value, each group of multi-source hand sensing data is projected to the membership corresponding radial covariance scale to fit the coordinate radius inside the anthropometric motion polar coordinate architecture, and an actual anthropometric motion polar coordinate layout of the multi-source hand sensing data is generated.

[0082] It should be noted that the device will drive the hand training according to the intention and movement of the target user, but the multi-source data of the hand is various, and it is difficult to unify the real-time feedback of the current hand from the intention and movement two aspects, resulting in errors in the subsequent air pressure propulsion control. In view of this, the method first acquires multi-source hand sensing data through a multi-source sensor array of a hand rehabilitation device, wherein the multi-source hand sensing data includes electromyographic signal data, electroencephalographic signal data, galvanic skin response data, joint angle data, finger and palm posture data, grip data, finger force data, and target user basic information. Subsequently, the multi-source hand sensing data is centralized preprocessed, which can eliminate the offset of the data, avoid the correlation calculation of the mean interference, make the covariance matrix correctly reflect the real relationship between the data variables, and significantly improve the effectiveness and consistency of the multi-source data in the subsequent polar coordinate fusion. The covariance calculates the autocorrelation of each group of centralized preprocessed multi-source hand sensing data, and outputs the coupling variance of each data. The coupling variance quantifies the typical dependency relationship between each data variable in each group of multi-source hand sensing data, and provides a basic matrix for solving the projection direction of the multi-source data according to the size of the coupling variance, which describes the expression structure between different data sources. Since the multi-source data contains different intention and movement trends, for example, the electroencephalographic signal data, joint angle data and finger force data reflect the feedback of the target user's desire to stretch the fingers rather than flex, so a multi-source and standardized intention and movement reference must be established to feedback the current hand condition, that is, different human intention and movement indexes (diversity) are set, wherein the human intention and movement indexes include grasping intention, pinching intention, opening intention, fist intention, standing finger intention, wrist flexion movement, single finger movement, pronation gesture movement and supination gesture movement. Each index has a logical norm index (standardization) that conforms to human anatomy, that is, a logical norm trend index.

[0083] It should be noted that the method constructs a human intention motion polar coordinate framework, which uses Cartesian coordinates to define multiple equal angles starting from the center position, and each equal angle corresponds to a polar coordinate field for representing a human intention motion indicator, thereby enabling polar coordinate display of real-time intention motion conditions of multi-source data in all directions, providing a multi-dimensional visual fusion base for multi-source data feedback. The logical specification trend index constrains the projection reference of each polar coordinate angle to meet the logical specification of the human intention motion indicator, avoids arbitrary scaling of the projection vector of multi-source data, and provides a normalized projection result; and the polar coordinate projection of multi-source data always maintains the typical correlation between them, thereby ensuring the real-time and restoration of multi-source data, so that the maintenance coupling variance is set as a projection criterion that is highly autonomous, independent and spontaneous, and then the multi-source hand sensing data is radially decomposed one by one according to the radial joint projection vector constraint and in compliance with the autonomous projection criterion, to obtain the corresponding radial covariance scale and cross-projection covariance matrix. The radial covariance scale measures the projection vector that jointly shares feedback of the current hand condition between each group of multi-source hand sensing data, that is, it maximizes the total correlation between the projection results of each multi-source data, reflecting the common structure between multi-source data; and the cross-projection covariance matrix represents the total correlation of different data variables between multi-source hand sensing data. If the weighted total sum of the cross-projection covariance matrix exceeds the maximum weighted total sum threshold, it means that the total correlation between all multi-source hand sensing data at a certain human intention motion indicator level reaches the maximum effect on the radial covariance scale, that is, it can cooperatively feedback the current hand condition of the target user from the indicator level, so that each group of multi-source hand sensing data is projected on the radial covariance scale, so that the multi-source hand sensing data is fitted to the polar coordinate radius amplitude representation, forming a polar coordinate summary view that expresses the real-time conditions of the current hand on the intention and motion two levels in terms of angle and radius. Through the method, multi-source data can be analyzed in multiple dimensions, diversely and standardized, and then the intention and motion of the real-time hand of the target user are fed back in real time, providing analysis basis for subsequent accurate control of the device.

[0084] More specifically, the S104 specifically includes the following steps:

[0085] Obtaining the selected working mode of the target user for the hand rehabilitation device and the conceptual design drawing of the hand rehabilitation device, and obtaining the ideal working condition parameter set of human intention motion of the hand rehabilitation device executing the selected working mode through the conceptual design drawing;

[0086] According to the ideal working condition parameter set, a multi-source hand sensing data trend is constructed to an ideal intended motion polar coordinate layout under a selected working mode condition, a hash dislocation algorithm is introduced to calculate a hash dislocation area function of a non-overlapping area between an actual intended motion polar coordinate layout and the ideal intended motion polar coordinate layout, and a steady-state dislocation degree of the target user in running the selected working mode by the current hand rehabilitation device is determined according to the hash dislocation area function.

[0087] If the steady-state dislocation degree is greater than a preset steady-state dislocation degree, an intended motion state space model of the current hand rehabilitation device is constructed based on coupling correlations of the multi-source hand sensing data on different human body intended motion indicators, and a decoupling analysis between different input channels and output channels of the intended motion state space model is performed through calculation of zero points and pole points, and an open-loop running transfer function of the current hand rehabilitation device is determined.

[0088] An expected diagonal control system of the selected working mode is obtained, a compensation cutoff frequency is decided on a Bode diagram according to a phase margin of the expected diagonal control system, and a gain compensation matrix for different multi-source hand sensing data is established based on the steady-state dislocation degree.

[0089] The open-loop gain of the intended motion state space model is globally adjusted by the gain compensation matrix, so that the center frequency of the lag network of the open-loop running transfer function continuously approaches the compensation cutoff frequency, and a multi-dimensional lag compensation structure is generated.

[0090] The multi-dimensional lag compensation structure is deployed in an actual diagonal control system of the current hand rehabilitation device, and a stroke driving compensation amount of the intended motion reaching the selected working mode when the target user uses the current hand rehabilitation device is obtained.

[0091] It should be noted that since the target user can pre-select the working mode, each working mode is preset with a corresponding training template, and the movement between the real-time conditions of the current hand and the training template will inevitably go through a control stroke, but the existing control method usually roughly estimates the movement angle deviation between the real-time hand condition and the training template to obtain the required driving stroke amount of the finger joint, which makes the subsequent output control compensation of the positive and negative pressure air pump not accurate enough, making it difficult for the hand training to meet the standard control requirements of the selected working mode. In view of this, the method first obtains the selected working mode of the target user and the ideal working condition parameter set (training template) under the mode, wherein the working mode of the device mainly includes automatic mode, single finger cycle mode, anti-interference mode and mirror mode. According to the ideal working condition parameter set, a multi-source hand sensing data trend is constructed to the ideal intended motion polar coordinate layout under the condition of the selected working mode. The ideal intended motion polar coordinate layout can reflect the degree of meeting the selected working mode of the current real-time hand condition, that is, from the hash misplacement area function of the non-overlapping area between the two, which globally represents a steady-state misplacement degree of the current hand rehabilitation device running the selected working mode; if the steady-state misplacement degree is greater than the preset steady-state misplacement degree, it means that the phase margin of the current hand rehabilitation device is insufficient, making it difficult for the current hand condition to complete the selected working mode, and a certain amount of air pressure compensation control needs to be applied by the positive and negative pressure air pump and the electromagnetic valve. At this time, the method constructs the intended motion state space model of the current hand rehabilitation device based on the coupling correlation of multi-source hand sensing data on different human intended motion indicators, and then determines the dynamic relationship between the input and output of the controlled object, including coupling structure, stability or response, etc. Subsequently, the zero point and the pole are calculated to decouple the intended motion state space model, analyze the directionality and coupling strength of multi-source data from different input channels and output channels of the current hand rehabilitation device, further lock the control bit plane, link and amplitude that need to be compensated, improve the steady-state performance, and effectively improve the compensation accuracy of the stroke driving.

[0092] It should be noted that the desired diagonal control system prompts the current hand rehabilitation device to achieve the approximate diagonalization architecture of the desired decoupling effect (the best decoupling of the directionality and coupling strength of the multi-source hand sensing data) after the compensation of the selected working mode control. According to the compensation cutoff frequency decided by the expected diagonal control system as the center point of the compensation structure, a degree target can be defined for the compensation range. Then, the steady-state misalignment degree is used to establish the open-loop gain of the global steady-state tuning intention motion state space model of the gain compensation matrix for different multi-source hand sensing data, so that the center frequency of the lag network of the open-loop running transfer function continuously approaches the compensation cutoff frequency, thereby accurately controlling the influence frequency band of the compensation structure and improving the pertinence and directivity of the air pressure propulsion compensation to meet the requirements of the selected working mode. The steady-state compensation accuracy and stability of the current hand rehabilitation device can be further improved to ensure the accuracy of the auxiliary operation control of the current hand rehabilitation device for the target user's intended motion to meet the requirements of the selected working mode.

[0093] More specifically, the S106 specifically includes the following steps:

[0094] The hand features of the target user are imitated by the multi-sensor array to obtain the actual joint linear profile of the target user. The joint linear profile with the maximum similarity to the actual joint linear profile is extracted from the hand feature database preset by the hand rehabilitation device, which is defined as the joint reference blueprint.

[0095] The standard product drawing of the hand rehabilitation device is obtained, and the simulation model of the hand rehabilitation device is reconstructed according to the standard product drawing by using the SolidWorks model simulation software. A virtual end effector is injected into each hand joint in the simulation model;

[0096] Based on the ideal working condition parameter set, the termination posture rotation matrix of the virtual end effector meeting the selected working mode is preset. According to the joint reference blueprint, the joint degrees of freedom, skeletal connection relationship and starting motion angle that fit the target user's hand linear profile are defined from the root to the virtual end effector, and the chain joint motion structure is output.

[0097] The personalized setting parameters of the target user using the selected working mode are obtained, and the movable points of the current hand rehabilitation device and the real-time joint angles of the target user maintaining the last default posture state when the multi-source hand sensing data is captured are obtained synchronously.

[0098] The inverse kinematics algorithm is introduced. From the real-time joint angle, the travel driving compensation amount is applied in the inverse kinematics algorithm in accordance with the motion premise constraints of the chain joint motion structure and the personalized setting parameters, so as to deduce the actual positions of each virtual end effector and obtain the end posture rotation matrix. The deviation between the end posture rotation matrix and the termination posture rotation matrix is calculated to obtain the matrix error degree of each movable point.

[0099] The analysis matrix error degree positioning assigns bone binding weights and weighted transform rendering, generates an active trade-off vertex grid, uses the active trade-off vertex grid to solve the independent angle alternation of the virtual end effector, and obtains the joint flexion and extension control parameters of the hand rehabilitation device execution stroke driving compensation.

[0100] It should be noted that the training template of the current hand rehabilitation device execution stroke driving compensation to meet the selected working mode is actually the motion process of the joint timing joint rotation of different fingers, but the existing control method is difficult to accurately deduce the flexion and extension rotation degree of the finger joint according to the known stroke driving compensation. The positive and negative pressure air pumps on the finger joints of the device wearing gloves are controlled by corresponding nodes, so that the positive and negative pressure air pumps corresponding to different finger joints cannot output accurate and reliable propulsion control air pressure, which may cause the uncoordination and excessive movement of the finger training. To this end, the method constructs a standardized hand rehabilitation device simulation model, which is mainly used for simulation of stroke driving compensation. It is worth mentioning that the method virtually injects an end effector (virtual end effector) at the end of the finger joint in the simulation model. The virtual end effector can be used to capture the position and attitude during the current hand rehabilitation device simulation stroke driving compensation process, thereby providing a reverse calculation basis for subsequent joint flexion and extension angle or displacement parameters. Compared with traditional artificial experience intervention, the accuracy of joint rotation flexion and extension deduction can be significantly improved, and the control reliability can be ensured. The joint linear profile (joint reference blueprint) conforming to the target user's hand linear type (maximum similarity) is acquired synchronously as a joint chain topology and degree of freedom limited sketch template, thereby constructing a joint degree of freedom, skeleton connection relationship and starting motion angle that fit the target user's hand linear type from the root to the virtual end effector, forming a chain joint motion structure, thereby clearly adjusting the joint position node and its physical limit, forming the basis of the reverse deduction structure clue. Since the selected point working mode can be customized with different personalized setting parameters such as intensity, frequency and time by the target user, the hand control is extremely flexible, which may cause simulation simulation to exist drift phenomenon, so it is necessary to add the personalized setting parameters selected by the target user as the directional constraint condition of the motion premise in the simulation simulation process, thereby effectively improving the traceability reliability of the joint driving error of the hand rehabilitation device.

[0101] It should be noted that the method introduces a reverse kinematics algorithm to apply a stroke driving compensation amount to the current hand rehabilitation device according to the movement premise constraints of the chain joint movement structure and the personalized setting parameters from the real-time joint angle of the given origin, thereby performing reverse deduction on the stroke rotation landing points of different active joints (movable points) of the current hand rehabilitation device when the stroke driving compensation amount is executed. At this time, the matrix error between the end pose rotation matrix and the termination pose rotation matrix records the matrix error of each movable point, which quantifies the deviation of the current rotation pose from the target rotation pose when the selected working mode is reached. This drives the reverse movement process of different joints, which is the key collaborative scheduling feedback information. Through this method, the joint flexion and extension landing points, angles, and displacements of the current hand rehabilitation device executing the stroke driving compensation amount in time sequence can be calculated in a simulation manner, thereby providing a complete flexion clue chain for the positive and negative pressure air pumps on different joint nodes of the hand rehabilitation device to execute the stroke driving compensation amount to reach the selected working mode, ensuring the coordination and reasonable specification of hand training, and improving the accuracy of positive and negative pressure air pump air pressure propulsion control.

[0102] More specifically, the analysis matrix error positioning assigns a skeletal binding weight and weighted transformation rendering to generate an active trade-off vertex grid, uses the active trade-off vertex grid to solve the independent angle alternation of the virtual end effector, and obtains the joint flexion control parameters of the hand rehabilitation device executing the stroke driving compensation amount, specifically including the following steps:

[0103] If the matrix error degree is less than the preset matrix error degree, the movable point is marked as a low-amplitude response active point; if the matrix error degree is greater than the preset matrix error degree, the movable point is marked as a high-amplitude response active point;

[0104] Based on the matrix error degree, the skeletal binding weights of different low-amplitude response active points and high-amplitude response active points are assigned, and the termination pose rotation matrix is used to perform weighted transformation rendering on the movable points according to the skeletal binding weights, to generate an active trade-off vertex grid;

[0105] In the weighted transformation process, the Jacobian matrix between the joint angle change and the virtual end effector pose change is established by synchronous linear approximation, and the joint angles of the active trade-off vertex grid are solved by damped least squares according to the Jacobian matrix, to obtain an independent angle alternation chain of each virtual end effector;

[0106] According to the independent angle alternation chain, the joint angles corresponding to each finger joint of the current hand rehabilitation device are continuously adjusted until the matrix error degrees of the low-amplitude response active points and the high-amplitude response active points are eliminated, to obtain the joint flexion control parameters of the hand rehabilitation device executing the stroke driving compensation amount.

[0107] It should be noted that the method utilizes matrix error degree to analyze the posture rotation scheduling of different joint nodes of the current hand rehabilitation device. If the matrix error degree is less than the preset matrix error degree, it indicates that the necessary scheduling proportion of the joint participating in the posture rotation is low, which is a joint node that does not need frequent rotation activity, and the movable point is calibrated as a low-amplitude response activity point. On the contrary, it indicates that the necessary scheduling proportion of the joint participating in the posture rotation is high, and if the rotation definition and holding are missing, it may cause the posture requirement of the selected working mode to be unable to be completed, so it is calibrated as a high-amplitude response activity point. The matrix error degree responds to the influence degree of the positive and negative air pressure pumps at the low-amplitude response activity point and the high-amplitude response activity point on the joint skeleton posture rotation collaborative scheduling corresponding to the virtual end effector, so the method assigns a skeletal binding weight to the movable point at the place and the joint skeleton corresponding to the virtual end effector according to the matrix error degree, and responds to the skeletal transformation through the weight, and finally renders a vertex grid that presents the stroke driving compensation control of the current hand rehabilitation device, that is, the activity trade-off vertex grid, realizes the visualization of reverse motion deduction, improves the expressiveness, restoration and realism of joint flexion and extension rotation, and makes the operation of joint flexion control parameters more accurate. In addition, the method establishes a Jacobian matrix between the joint angle change and the virtual end effector posture change through linear approximation. The Jacobian matrix provides detailed information about the trend of the movable joint moving towards the target direction, thereby accurately capturing the skeletal motion flexibility in the simulation process and improving the linear alternation accuracy of the independent angle opening of the joint during the time sequence change of different virtual end effectors.

[0108] More specifically, the S108 specifically includes the following steps:

[0109] A preset training control strategy of the hand rehabilitation device is obtained, and a plurality of rotation change vectors planned by the hand rehabilitation device for different joint flexion processes and air pressure control levels of the positive and negative air pressure pumps for realizing decision output of each rotation change vector are extracted through the preset training control strategy;

[0110] The simulation model of the hand rehabilitation device is used to import the preset training control strategy for simulation. At this time, the rotation posture scalar field of the hand rehabilitation device during the process of simulating the execution of the stroke driving compensation amount from the real-time joint angle to achieve the selected working mode is exported through the SolidWorks model simulation software, and a quaternion mapping field is built based on the architecture layout of the rotation posture scalar field;

[0111] Each rotation change vector is designated as a prefix label of the quaternion mapping field, and the air pressure control level corresponding to each rotation change vector is set as a suffix label, forming a control query item of prefix-suffix mapping binding, and each rotation change vector is assigned a quaternion string;

[0112] The joint flexion and extension angle, angular velocity, angle switching step and joint flexion and extension increment recorded according to the joint flexion and extension control parameter are subjected to rotational differential conversion to obtain a necessary rotational change vector for stroke driving compensation of the hand rehabilitation device;

[0113] The necessary rotational change vector is converted into an incremental quaternion by introducing an exponential mapping formula to obtain a necessary rotational incremental quaternion. The necessary rotational incremental quaternion is projected into a quaternion mapping field to enable the necessary rotational incremental quaternion to sequentially traverse each control query item corresponding quaternion string for approximate comparison;

[0114] If the necessary rotational incremental quaternion and the quaternion string approximately match to a degree greater than a preset approximate matching degree, a cumulative superposition hat mark is added to the quaternion string; otherwise, the quaternion string is ignored to obtain a series of candidate superposition quaternions;

[0115] The air pressure control magnitude of the control query item corresponding to the series of candidate superposition quaternions is subjected to rotational accumulation by using quaternion multiplication to update the attitude quaternion of the hand rehabilitation device, and finally output the air pressure control cumulative parameter to intelligently control the hand rehabilitation device based on the air pressure control cumulative parameter.

[0116] It should be noted that the preset control strategy of the device usually decides a air pressure control magnitude according to the rotation vector of the glove joint to provide reasonable driving power control its execution selected work mode training movement, but the traditional control method can only rely on artificial control experience fixed matching control amount, resulting in more refined rotation vector difficult to determine the air pressure control magnitude, thereby easily making the positive and negative pressure air pump appear larger air pressure control error, causing training amplitude unreasonable and unable to meet the selected work mode standard, greatly increasing the training risk accident. In view of this, the method builds a quaternion mapping field through the architecture layout of the rotation posture scalar field derived in the simulation process of the hand rehabilitation device, which provides a reliable index query platform for subsequent joint flexion control parameters. In the quaternion mapping field, the control query items with mutual echo relationship of the preset training control strategy are recorded, for example: prefix x, y, z (rotation change vector) -suffix 1.5MPa (air pressure control magnitude) [0.133, 0.87, 1.62, 0.5225 (quaternion string)]. Then the necessary rotation change vector converted after the joint flexion control parameter is constructed into an incremental quaternion form, and the necessary rotation incremental quaternion is obtained, and it is mapped to the constructed quaternion mapping field to perform traversal approximate query comparison for each control query item. If the approximate coincidence degree of the necessary rotation incremental quaternion and the quaternion string is greater than the preset approximate coincidence degree, it means that the rotation change vector corresponding to the quaternion string in the control query item has a higher matching precision with the necessary rotation incremental quaternion, so the air pressure control magnitude of the corresponding suffix can be used for the air pressure propulsion control of the joint flexion control parameter. Therefore, the method uses a cumulative superposition cap marker to mark the control query item, which can improve the index accuracy of quaternion rotation accumulation, avoid accumulation omission, error or missing, and ensure the reliability and stability of the air pressure control of the positive and negative pressure air pump. Finally, the marked candidate superposition quaternion is accumulated to form a control amount that meets the joint flexion control parameter and conforms to the change characteristics of the joint flexion with time sequence promotion, which can significantly reduce the risk and probability of unreasonable air pressure control training accidents, and improve the coordination and smoothness of the transition control of the hand rehabilitation device.

[0117] The third aspect of the present application provides a hand rehabilitation system based on multi-source data analysis, as shown in Figure 4 The hand rehabilitation system includes a memory 41 and a processor 42, the memory 41 stores a hand rehabilitation method program based on multi-source data analysis, when the hand rehabilitation method program is executed by the processor 42, the hand rehabilitation method steps are realized.

[0118] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A control method for a hand rehabilitation device based on multi-source data analysis, characterized in that, Includes the following steps: S102: Acquire multi-source hand sensor data from hand rehabilitation equipment and acquire human intention movement system. Follow the human intention movement system to perform polar coordinate radial fusion of multi-source hand sensor data according to logical standardization trend to obtain the actual intention movement polar coordinate layout of multi-source hand sensor data. S104: Based on the standardized misalignment analysis of the ideal working condition parameter set of the human body's intended movement when the hand rehabilitation device executes the selected working mode, the actual intended movement polar coordinate layout is analyzed, and multi-dimensional lag compensation is performed to obtain the stroke drive compensation amount for the current hand rehabilitation device's intended movement to reach the selected working mode. S106: Construct the chain joint motion structure of the target user, simultaneously inject virtual end effectors into the hand rehabilitation device model, and solve the independent angle changes of the posture rotation of each virtual end effector when the hand rehabilitation device applies the stroke drive compensation amount according to the chain joint motion structure through the inverse kinematics solution algorithm, so as to obtain the joint flexion and extension control parameters of the hand rehabilitation device when it executes the stroke drive compensation amount. S108: Based on the rotational attitude scalar field derived during the simulation and the preset training control strategy of the hand rehabilitation device, a quaternion mapping domain is constructed. The joint flexion and extension control parameters are mapped to the quaternion mapping domain and the air pressure control cumulative calculation is performed to intelligently train and control the hand rehabilitation device.

2. The control method for a hand rehabilitation device based on multi-source data analysis according to claim 1, characterized in that, The hand rehabilitation device includes: Wearing gloves: The wearable gloves are designed to completely cover the finger joints and palm joints of the target user; Positive and negative pressure air pump module: The positive and negative pressure air pump module is used to control the generation of positive and negative pressure gas to provide air pressure power for assisted flexion and extension training activities of finger joints and to adjust the training intensity and force; Solenoid valve module: The solenoid valve module is connected to the positive and negative pressure air pump module and is responsible for controlling the on / off state and flow rate of compressed air, simulating the natural grasping action of the hand; Temperature control module: The temperature control module is used to adjust the hand temperature according to the target user's training status, providing a heating effect for hand rehabilitation training; Human-computer interaction interface: The human-computer interaction interface is used to allow target users to switch working modes and personalize the time, intensity and frequency of hand training; Memory storage module: The memory storage module is responsible for remembering and saving the previous hand training plan, so that it can be directly applied next time without repeated debugging.

3. The control method for a hand rehabilitation device based on multi-source data analysis according to claim 1, characterized in that, S102 specifically includes the following steps: The multi-source sensor array of the hand rehabilitation device is used to collect data on the target user within a preset time period when the target user wears the hand rehabilitation device, thereby obtaining multi-source hand sensing data. Centralized preprocessing of multi-source hand sensing data for each group is performed, and a covariance algorithm is introduced to calculate the autocorrelation dependency of the centralized preprocessed multi-source hand sensing data for each group, outputting the coupling variance between each data variable in each group of multi-source hand sensing data. By acquiring human body knowledge graphs through big data, identifying multi-source hand sensor data using human body knowledge graphs, outputting a human intention movement system that conforms to the stage characteristics of the target user, and extracting different human intention movement indicators and logical normative trend indices of each human intention movement indicator through the human intention movement system. According to different human intention movement indicators, the principal component source coordinate angles are assigned in Cartesian coordinate form to form a human intention movement polar coordinate framework. The coupling variance is used as the autonomous projection criterion, and a radial joint projection vector constraint is established based on the logical norm trend index. Based on the radial joint projection vector constraint, the covariance algorithm radially decomposes each group of multi-source hand sensing data according to the autonomous projection criterion and performs cross-calculation of co-correlation. At this time, the radial covariance scale and cross-projection covariance matrix of each group of multi-source hand sensing data are output. If the weighted sum of the cross-projection covariance matrix exceeds the maximum weighted sum threshold, then within the human intention motion polar coordinate architecture, each set of multi-source hand sensing data is projected onto the corresponding radial covariance scale to fit the coordinate radius, generating the actual intention motion polar coordinate layout of the multi-source hand sensing data.

4. The control method for a hand rehabilitation device based on multi-source data analysis according to claim 1, characterized in that, S104 specifically includes the following steps: Obtain the target user's selected working mode for the hand rehabilitation device and the conceptual design drawings of the hand rehabilitation device. Obtain the ideal set of working condition parameters for the human body's intended movement when the hand rehabilitation device executes the selected working mode through the conceptual design drawings. Based on the ideal working condition parameter set, a multi-source hand sensing data is constructed to tend to the ideal intention motion polar coordinate layout under the selected working mode condition. A hash misalignment algorithm is introduced to calculate the hash misalignment area function of the non-overlapping area between the actual intention motion polar coordinate layout and the ideal intention motion polar coordinate layout. Based on the hash misalignment area function, the steady-state misalignment degree of the current hand rehabilitation device to enable the target user to run the selected working mode is determined. If the steady-state misalignment is greater than the preset steady-state misalignment, then the intentional motion state space model of the current hand rehabilitation device is constructed based on the coupling correlation of multi-source hand sensing data on different human intentional motion indicators. The open-loop operation transfer function of the current hand rehabilitation device is determined by calculating the zeros and poles to perform decoupling analysis between different input and output channels of the intentional motion state space model. Obtain the desired diagonal control system of the selected working mode, determine the compensation cutoff frequency on the Bode plot based on the phase margin of the desired diagonal control system, and establish a gain compensation matrix for different multi-source hand sensing data based on the steady-state misalignment. By globally steady-state tuning the gain compensation matrix, the open-loop gain of the motion state space model is adjusted so that the center frequency of the hysteresis network of the open-loop operating transfer function continuously approaches the compensation cutoff frequency, thus generating a multi-dimensional hysteresis compensation structure. By deploying a multidimensional hysteresis compensation structure into the actual diagonal control system of the current hand rehabilitation device, the stroke drive compensation amount is obtained when the target user intends to move to the selected point working mode when using the current hand rehabilitation device.

5. The control method for a hand rehabilitation device based on multi-source data analysis according to claim 1, characterized in that, S106 specifically includes the following steps: By using a multi-sensor array to mimic the hand features of the target user, the actual joint line contour of the target user is obtained. The joint line contour with the greatest similarity to the actual joint line contour is extracted from the hand feature database preset by the hand rehabilitation device and defined as the joint reference blueprint. Obtain standard product drawings of hand rehabilitation equipment, and reconstruct a simulation model of the hand rehabilitation equipment according to the standard product drawings using SolidWorks model simulation software. In the simulation model, inject a virtual end effector into each hand joint. Based on the ideal working condition parameter set, the virtual end effector is preset to meet the selected working mode of the termination posture rotation matrix. According to the joint reference blueprint, the joint degrees of freedom, skeletal connection relationship and starting motion angle are defined from the root to the virtual end effector to fit the target user's hand line shape, and the chain joint motion structure is output. Acquire personalized settings parameters for the target user when using the selected working mode, and simultaneously acquire the movable points of the current hand rehabilitation device and the real-time joint angles of the target user when maintaining the previous default posture state while capturing multi-source hand sensor data; Introducing an inverse kinematics algorithm, starting from a real-time joint perspective with a predetermined origin, the inverse kinematics algorithm applies stroke drive compensation in accordance with the motion premise constraints of the chain joint motion structure and personalized setting parameters. This allows for the deduction of the actual position of each virtual end effector, obtaining the end attitude rotation matrix. The deviation between the end attitude rotation matrix and the termination attitude rotation matrix is ​​then calculated to obtain the matrix error degree of each movable point. The analysis matrix error degree is used to locate and assign bone binding weights and perform weighted transformation rendering to generate an activity balance vertex grid. The activity balance vertex grid is used to solve the independent angle changes of the virtual end effector to obtain the joint flexion and extension control parameters of the stroke drive compensation of the hand rehabilitation device.

6. The control method for a hand rehabilitation device based on multi-source data analysis according to claim 5, characterized in that, The analysis matrix error degree is used to locate and assign bone binding weights and perform weighted transformation rendering to generate an activity-weighted vertex grid. The activity-weighted vertex grid is then used to solve for the independent angle changes of the virtual end effector, obtaining the joint flexion and extension control parameters of the hand rehabilitation device's stroke drive compensation. The specific steps include: If the matrix error is less than the preset matrix error, the active point is calibrated as a low-amplitude response active point; if the matrix error is greater than the preset matrix error, the active point is calibrated as a high-amplitude response active point. Based on the matrix error degree, different low-amplitude response activity points and high-amplitude response activity points are assigned bone binding weights. According to the bone binding weights, the termination pose rotation matrix is ​​used to perform weighted transformation rendering on the movable points to generate an activity balance vertex grid. During the weighted transformation, a Jacobian matrix is ​​established between the joint angle change and the virtual end effector posture change using a synchronous linear approximation. Based on the Jacobian matrix, damped least squares solution is performed on the joint angle of the active tradeoff vertex grid to obtain the independent angle iteration chain of each virtual end effector. Based on the independent angle iteration chain, the joint angle corresponding to each finger joint on the current hand rehabilitation device is continuously adjusted and changed until the matrix error degree of low amplitude response activity point and high amplitude response activity point is eliminated, so as to obtain the joint flexion and extension control parameters of the hand rehabilitation device to perform stroke drive compensation.

7. The control method for a hand rehabilitation device based on multi-source data analysis according to claim 1, characterized in that, S108 specifically includes the following steps: The preset training control strategies of the hand rehabilitation equipment are obtained. Through the preset training control strategies, the various rotational change vectors planned by the hand rehabilitation equipment for different joint flexion and extension processes and the air pressure control level of the decision output of each rotational change vector by the positive and negative air pressure pumps are extracted. The simulation model of the hand rehabilitation device is used to import the preset training control strategy for simulation. At this time, the rotational attitude scalar field of the hand rehabilitation device is exported from the SolidWorks model simulation software to simulate the stroke drive compensation amount of the hand rehabilitation device from the real-time joint angle to the selected working mode. The quaternion mapping domain is built based on the architecture layout of the rotational attitude scalar field. Each rotation change vector is designated as a prefix label in the quaternion mapping domain, and the corresponding air pressure control level is set as a suffix label, forming a control query item bound by prefix-suffix mapping. Each rotation change vector is pre-assigned a quaternion string. Based on the joint flexion and extension control parameters recorded in the joint flexion and extension control parameters, the joint flexion and extension angle, angular velocity, angle switching step size, and joint flexion and extension increment, a rotational differential transformation is performed to obtain the necessary rotational change vector for the stroke drive compensation amount of the hand rehabilitation device. The necessary rotation change vector is converted into an incremental quaternion by introducing an exponential mapping formula, and the necessary rotation incremental quaternion is obtained. The necessary rotation incremental quaternion is then projected onto the quaternion mapping domain, so that the necessary rotation incremental quaternion traverses the quaternion string corresponding to each control query item for approximate comparison. If the approximate fit between the necessary rotation increment quaternion and the quaternion string is greater than the preset approximate fit, then add a cumulative stacking cap marker on the quaternion string; otherwise, ignore the quaternion string and obtain a series of candidate stacked quaternions. Quaternion multiplication is used to rotate and accumulate the air pressure control magnitude of a series of candidate superimposed quaternions corresponding to control query items in order to update the posture quaternion of the hand rehabilitation device. Finally, the accumulated air pressure control parameters are output, and intelligent training control of the hand rehabilitation device is performed based on the accumulated air pressure control parameters.

8. A hand rehabilitation system based on multi-source data analysis, characterized in that, The hand rehabilitation system includes a memory and a processor. The memory stores a control method program for a hand rehabilitation device based on multi-source data analysis. When the control method program is executed by the processor, it implements the control method steps as described in any one of claims 1-7.

Images