Intelligent control method and system for lower limb hip and knee rehabilitation training equipment

Abstract

The present invention relates to the technical field of medical rehabilitation equipment, and in particular to an intelligent control method and system for lower limb hip and knee rehabilitation training equipment. Identify and obtain the fatigue-prone key points of the target training subject performing the current walking posture, construct a multidimensional electromyographic fatigue thermal model based on the electromyographic fatigue sensor data of the lower limb hip and knee rehabilitation training equipment, and use the multidimensional electromyographic fatigue thermal model to perform thermal analysis on the fatigue-prone key points to control the rate and frequency of the lower limb hip and knee rehabilitation training equipment; calculate the lower limb hip and knee timing state variables of the lower limb hip and knee rehabilitation training equipment transferred from the adjacent training electromyogram to the current training electromyogram through the reflex contraction replacement of the worst solution simple vertex, and control the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training equipment according to the change in the lower limb hip and knee timing state. The present invention can intelligently control the hip and knee joint training effect of the lower limb hip and knee rehabilitation training equipment on the user.

Description

Technical Field

[0001] The present invention relates to the technical field of medical rehabilitation equipment, and in particular to an intelligent control method and system for lower limb hip and knee rehabilitation training equipment. Background Art

[0002] Lower limb hip and knee rehabilitation training equipment is primarily used to assist patients with hip and knee rehabilitation training, particularly those undergoing hip replacement or knee surgery, or those experiencing lower limb dysfunction due to sports injuries, aging, and other factors. Existing rehabilitation equipment mostly relies on manual adjustment or simple electronic control systems to assist with movement. While these devices can meet certain treatment needs, they still have significant deficiencies in precise control, personalized training, and intelligent feedback. With the continuous development of intelligent technology, a growing number of researchers are committed to improving the efficiency and effectiveness of rehabilitation training by introducing intelligent control systems.

[0003] At present, most lower limb hip and knee rehabilitation training equipment still relies on traditional mechanical design and manual adjustment, lacking real-time matching with the patient's physiological state and motor ability. The control systems of these devices are usually relatively simple, and it is difficult to automatically control reasonable training parameters according to the patient's rehabilitation progress and muscle fatigue level. On the other hand, although some devices have introduced simple electronic control systems, they often lack intelligence, data feedback and personalized adjustment functions, and cannot provide customized rehabilitation control solutions according to the needs of different patients. In addition, the existing technology lacks the function of evaluating and analyzing the patient's hip and knee joint injuries, which makes it difficult to quantify the effect of rehabilitation training. Patients can often only rely on subjective feelings to make adjustments, making it difficult to achieve accurate and effective rehabilitation training. Summary of the Invention

[0004] The present invention overcomes the deficiencies of the prior art and provides an intelligent control method and system for lower limb hip and knee rehabilitation training equipment.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A first aspect of the present invention provides an intelligent control method for a lower limb hip and knee rehabilitation training device, comprising the following steps:

[0007] S102: Obtain static data of the target training subject's lower limb hip and knee joints and force distribution data of the thigh and calf dynamic training, and perform Gaussian distribution calculation on the target training subject's gait using the static data of the lower limb hip and knee joints and the force distribution data to obtain the target training subject's current walking posture;

[0008] S104: Based on the target training subject's lower limb hip and knee injury rating radar chart, a preset rehabilitation training strategy of the lower limb hip and knee rehabilitation training device is subjected to variational inference learning to identify the activity range and rehabilitation training amplitude, generating a rehabilitation training amplitude pattern recognition model. The rehabilitation training amplitude pattern recognition model is used to identify and obtain the current rehabilitation training amplitude of the current walking posture to control the lower limb hip and knee rehabilitation training device;

[0009] S106: Identify and obtain fatigue-prone key points of the target training subject during the current walking posture, construct a multi-dimensional electromyographic fatigue thermal model based on electromyographic fatigue sensor data of the lower limb hip and knee rehabilitation training device, and use the multi-dimensional electromyographic fatigue thermal model to perform thermal analysis on the fatigue-prone key points to control the rate and frequency of the lower limb hip and knee rehabilitation training device;

[0010] S108: calculating the lower limb hip and knee time sequence state variables of the lower limb hip and knee rehabilitation training device transferred from the adjacent training electromyogram to the current training electromyogram by reflex contraction replacement of the worst solution simple vertex, and controlling the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training device according to the change in the lower limb hip and knee time sequence state;

[0011] The step S104 specifically includes the following steps:

[0012] Obtaining a preset rehabilitation training strategy for a lower limb hip and knee rehabilitation training device, obtaining preset activity intervals for the lower limb hip and knee rehabilitation training device for different injury ratings based on the preset rehabilitation training strategy, and obtaining rehabilitation training amplitudes corresponding to each preset activity interval;

[0013] Obtain the lower limb hip and knee case reports and injury rating criteria for the target training subjects, overwrite and add descriptions of the rating indicators to each injury description in the lower limb hip and knee case reports using the description conditions of the injury rating criteria, and generate a radar chart of the lower limb hip and knee injury ratings for the target training subjects;

[0014] A variational autoencoder is constructed based on the preset activity intervals of the lower limb hip and knee rehabilitation training device for different injury ratings, and a variational autodecoder is constructed based on the rehabilitation training amplitude corresponding to each preset activity interval;

[0015] Constructing a latent recognition space for lower limb hip and knee walking postures, inputting the lower limb hip and knee injury rating radar chart into a variational autoencoder and mapping it to the latent recognition space to obtain an implicit activity interval coding sequence, and performing variational inference learning on the latent variable distribution of the implicit activity interval coding sequence in a variational autodecoder;

[0016] Through variational inference learning, the reconstruction loss difference of the rehabilitation training range value planning under the constraints of the lower limb hip and knee injury rating radar chart is obtained, and the allowable reconstruction loss threshold is preset based on the current walking posture of the target training subject;

[0017] If the reconstruction loss difference is greater than the allowable reconstruction loss threshold, a minimization process is performed on the reconstruction loss difference to train the rehabilitation training amplitude corresponding to the activity interval until the reconstruction loss difference is less than the allowable reconstruction loss threshold, thereby obtaining a rehabilitation training amplitude pattern recognition model of the lower limb hip and knee rehabilitation training equipment under the constraints of the lower limb hip and knee injury rating radar chart;

[0018] The current walking posture is identified by a rehabilitation training amplitude pattern recognition model to obtain a current rehabilitation training amplitude suitable for the target training object, and the power joints of the lower limb hip and knee rehabilitation training equipment are intelligently controlled according to the current rehabilitation training amplitude.

[0019] More specifically, the step S102 includes the following steps:

[0020] monitoring static data of the lower limb hip and knee joints of the target training subject through an inertial measurement unit sensor on a lower limb hip and knee rehabilitation training device, and defining an initial static gait of the target training subject based on the static data of the lower limb hip and knee joints;

[0021] acquiring force sensor monitoring data of a lower limb hip and knee rehabilitation training device, extracting force distribution data of a target training subject's thigh and calf during dynamic training on the lower limb hip and knee rehabilitation training device through the force sensor monitoring data, and presetting a gait covariance function of the target training subject's dynamic training starting from an initial stationary gait based on the force distribution data;

[0022] Acquire a human motion knowledge graph based on a big data network, identify the force distribution data through the human motion knowledge graph, and output the gait force characteristics of the target training subject and the standard force movement law of dynamic walking starting from an initial static gait;

[0023] Constructing a Gaussian process regression model for gait transfer, performing covariance calculation on a gait covariance function in the Gaussian process regression model based on the gait force characteristics to generate a covariance matrix, and determining a Gaussian gait prior distribution describing the target training subject's walking training from an initial stationary gait based on the covariance matrix;

[0024] Based on the standard force movement law, a probability likelihood function for dynamic training starting from an initial static gait is constructed, the logarithm of the probability likelihood function is calculated and the partial derivative is solved to obtain the gradient of the probability likelihood function, the gradient is used to infer the Gaussian gait prior distribution to obtain the Gaussian gait posterior distribution, and the current walking posture of the target training object is determined based on the description of the Gaussian gait posterior distribution.

[0025] More specifically, the method of obtaining a lower limb hip and knee case report of a target training subject and injury rating criteria for the lower limb hip and knee, overwriting and adding descriptions of the injury description information in the lower limb hip and knee case report with rating indicators using the description conditions of the injury rating criteria, and generating a radar chart of the lower limb hip and knee injury rating of the target training subject specifically includes the following steps:

[0026] Obtaining a lower limb hip and knee case report of a target training subject, and extracting a plurality of descriptive information about the target training subject's lower limb hip and knee injuries from the lower limb hip and knee case report;

[0027] Obtain injury rating criteria for the lower limb hip and knee based on a big data network, construct a set of descriptive conditions for each different injury rating based on the injury rating criteria, and define each descriptive condition as a conditional predicate;

[0028] A gain trade-off algorithm is introduced to calculate the information gain between each injury description and each conditional predicate. Based on the information gain, the coverage of the injury rating for the injury description is determined. The rating index of each injury rating in the injury rating criteria is obtained, and a radar feature matrix for lower limb hip and knee injury descriptions is established based on the rating index of each injury rating.

[0029] If the coverage is greater than the preset coverage, the injury rating corresponding to the conditional predicate is not added and is ignored; if the coverage is greater than the preset coverage, the injury rating corresponding to the conditional predicate is added to describe the rating feature of the injury description information, and the described rating feature is fitted into the radar feature matrix;

[0030] Repeat the above steps of adding the injury ratings corresponding to the conditional predicates to the radar feature matrix one by one to describe the injury description information until the rating features of all injury description information are traversed and the description fitting is completed, and finally a radar chart of the lower limb hip and knee injury ratings of the target training subject is generated.

[0031] More specifically, the step S106 includes the following steps:

[0032] Obtaining electromyographic fatigue sensor data and the arrangement array of electromyographic sensors of lower limb hip and knee rehabilitation training equipment, obtaining a standard human body structure diagram of the lower limb hip and knee joints based on a big data network, constructing a multidimensional thermal density framework of the lower limb hip and knee joints based on the standard human body structure diagram, and presetting a density kernel function and bandwidth smoothing parameters based on the electromyographic fatigue sensor data;

[0033] Calculating a density contribution function of each myoelectric fatigue sensing data based on a density kernel function and a bandwidth smoothing parameter to obtain a plurality of density contribution functions, and thermally summing the plurality of density contribution functions in a multidimensional thermal density framework according to the arrangement array to obtain a multidimensional myoelectric fatigue thermal model;

[0034] The current walking posture of the target training subject is identified through the human motion knowledge graph to obtain the muscle movement distribution of the target training subject's lower limb hip and knee when walking. Based on the muscle movement distribution, the muscle fatigue area is delineated in the standard human body structure diagram and marked as the fatigue key point;

[0035] Construct the myoelectric fatigue thermal color gamut table of the target training object's fatigue-prone key points in a non-fatigued state, and obtain the myoelectric fatigue thermal chromaticity of the fatigue-prone key points through a multi-dimensional myoelectric fatigue thermal model;

[0036] If the electromyographic fatigue thermal color gamut table can query the electromyographic fatigue thermal chromaticity, the training rate and frequency of the lower limb hip and knee rehabilitation training equipment are controlled incrementally; if the electromyographic fatigue thermal color gamut table cannot query the electromyographic fatigue thermal chromaticity, the training rate and frequency of the lower limb hip and knee rehabilitation training equipment are controlled incrementally.

[0037] More specifically, the step S108 includes the following steps:

[0038] Acquire a real-time electromyographic fatigue signal of a lower limb hip and knee rehabilitation training device at a current training time sequence, introduce a wavelet transform algorithm to perform wavelet basis transform on the electromyographic fatigue signal and reconstruct the electromyographic state display, and obtain a training electromyogram of the lower limb hip and knee of the target training subject at the current training time sequence, which is defined as the current training electromyogram;

[0039] Acquire a training electromyogram of a lower limb hip and knee rehabilitation training device located on an adjacent training time sequence relative to the current training time sequence, which is defined as an adjacent training electromyogram;

[0040] Constructing a multidimensional space according to the time series change from the adjacent time series to the current time series, taking the current training electromyogram as the objective function, planning a simplex transferred from the adjacent training electromyogram to the current training electromyogram based on the multidimensional space, and calculating the objective function value of each simplex vertex on the simplex;

[0041] Only the simple vertex corresponding to the maximum objective function value is extracted and marked as the worst solution simple vertex. The reflection point positioning formula is introduced to calculate the reflection point of the worst solution simple vertex to obtain the objective function value of the reflection point.

[0042] If the objective function value of the reflection point is less than the objective function value of the worst solution simple vertex, then the reflection point is accepted and replaced with the worst solution simple vertex; if the objective function value of the reflection point is greater than the objective function value of the worst solution simple vertex, then a contraction operation is performed on the reflection point to generate a contraction point, and the contraction point is accepted to replace the worst solution simple vertex;

[0043] Based on the preset iteration frequency of the current training electromyogram, the above steps of calculating and analyzing the objective functions of the reflection points and the contraction points and replacing the simple vertex of the worst solution are repeated until the replacement iteration of the simple vertex of the worst solution reaches the iteration frequency, and the lower limb hip and knee temporal state variables of the target training object transferred from the adjacent training electromyogram to the current training electromyogram are obtained;

[0044] The lower limb hip and knee timing state variables are input into the lower limb hip and knee rehabilitation training device to control and adjust the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training device.

[0045] The second aspect of the present invention provides an intelligent control system for lower limb hip and knee rehabilitation training equipment. The intelligent control system for lower limb hip and knee rehabilitation training equipment includes a memory and a processor. The memory stores an intelligent control method program for lower limb hip and knee rehabilitation training equipment. When the intelligent control method program is executed by the processor, any step of the intelligent control method is implemented.

[0046] The present invention solves the technical defects existing in the background technology, and the beneficial technical effects of the present invention are:

[0047] Obtain the static data of the lower limb hip and knee joints of the target training subject and the force distribution data of the thigh and calf dynamic training, use the static data of the lower limb hip and knee joints and the force distribution data to calculate the Gaussian distribution of the target training subject's gait, and obtain the current walking posture of the target training subject; based on the lower limb hip and knee injury rating radar chart of the target training subject, perform variational inference learning on the recognition of activity interval to rehabilitation training amplitude of the preset rehabilitation training strategy of the lower limb hip and knee rehabilitation training equipment, generate a rehabilitation training amplitude pattern recognition model, and obtain the current rehabilitation training amplitude control of the current walking posture through the rehabilitation training amplitude pattern recognition model. Lower limb hip and knee rehabilitation training equipment; identifying and obtaining fatigue-prone key points of a target training subject when performing a current walking posture, constructing a multidimensional electromyographic fatigue thermal model based on the electromyographic fatigue sensor data of the lower limb hip and knee rehabilitation training equipment, and using the multidimensional electromyographic fatigue thermal model to perform thermal analysis on the fatigue-prone key points to control the rate and frequency of the lower limb hip and knee rehabilitation training equipment; calculating the lower limb hip and knee time series state variables of the lower limb hip and knee rehabilitation training equipment transferred from the adjacent training electromyogram to the current training electromyogram through the reflex contraction replacement of the worst solution simple vertex, and controlling the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training equipment according to the change in the lower limb hip and knee time series state. The present invention can accurately and intelligently control the dynamic joint training amplitude, training rate and frequency, as well as electromyographic detection-based adjustment, when the target training subject uses the lower limb hip and knee rehabilitation training equipment, replacing the phenomenon that traditional manual rehabilitation training is difficult to achieve the expected results. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, without paying any creative work, they can also obtain drawings of other embodiments based on these drawings.

[0049] Figure 1 A first method flow chart of an intelligent control method for lower limb hip and knee rehabilitation training equipment is shown;

[0050] Figure 2 A second method flow chart of an intelligent control method for lower limb hip and knee rehabilitation training equipment is shown;

[0051] Figure 3 The present invention shows a system framework diagram of an intelligent control system for lower limb hip and knee rehabilitation training equipment. DETAILED DESCRIPTION

[0052] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0053] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0054] The first aspect of the present invention provides an intelligent control method for lower limb hip and knee rehabilitation training equipment, such as Figure 1 As shown, the following steps are included:

[0055] S102: Obtain static data of the target training subject's lower limb hip and knee joints and force distribution data of the thigh and calf dynamic training, and perform Gaussian distribution calculation on the target training subject's gait using the static data of the lower limb hip and knee joints and the force distribution data to obtain the target training subject's current walking posture;

[0056] S104: Based on the target training subject's lower limb hip and knee injury rating radar chart, a preset rehabilitation training strategy of the lower limb hip and knee rehabilitation training device is subjected to variational inference learning to identify the activity range and rehabilitation training amplitude, generating a rehabilitation training amplitude pattern recognition model. The rehabilitation training amplitude pattern recognition model is used to identify and obtain the current rehabilitation training amplitude of the current walking posture to control the lower limb hip and knee rehabilitation training device;

[0057] S106: Identify and obtain fatigue-prone key points of the target training subject during the current walking posture, construct a multi-dimensional electromyographic fatigue thermal model based on electromyographic fatigue sensor data of the lower limb hip and knee rehabilitation training device, and use the multi-dimensional electromyographic fatigue thermal model to perform thermal analysis on the fatigue-prone key points to control the rate and frequency of the lower limb hip and knee rehabilitation training device;

[0058] S108: calculating the lower limb hip and knee time sequence state variables of the lower limb hip and knee rehabilitation training device transferred from the adjacent training electromyogram to the current training electromyogram by reflex contraction replacement of the worst solution simple vertex, and controlling the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training device according to the change in the lower limb hip and knee time sequence state;

[0059] The step S104 specifically includes the following steps:

[0060] Obtaining a preset rehabilitation training strategy for a lower limb hip and knee rehabilitation training device, obtaining preset activity intervals for the lower limb hip and knee rehabilitation training device for different injury ratings based on the preset rehabilitation training strategy, and obtaining rehabilitation training amplitudes corresponding to each preset activity interval;

[0061] Obtain the lower limb hip and knee case reports and injury rating criteria for the target training subjects, overwrite and add descriptions of the rating indicators to each injury description in the lower limb hip and knee case reports using the description conditions of the injury rating criteria, and generate a radar chart of the lower limb hip and knee injury ratings for the target training subjects;

[0062] A variational autoencoder is constructed based on the preset activity intervals of the lower limb hip and knee rehabilitation training device for different injury ratings, and a variational autodecoder is constructed based on the rehabilitation training amplitude corresponding to each preset activity interval;

[0063] Constructing a latent recognition space for lower limb hip and knee walking postures, inputting the lower limb hip and knee injury rating radar chart into a variational autoencoder and mapping it to the latent recognition space to obtain an implicit activity interval coding sequence, and performing variational inference learning on the latent variable distribution of the implicit activity interval coding sequence in a variational autodecoder;

[0064] Through variational inference learning, the reconstruction loss difference of the rehabilitation training range value planning under the constraints of the lower limb hip and knee injury rating radar chart is obtained, and the allowable reconstruction loss threshold is preset based on the current walking posture of the target training subject;

[0065] If the reconstruction loss difference is greater than the allowable reconstruction loss threshold, a minimization process is performed on the reconstruction loss difference to train the rehabilitation training amplitude corresponding to the activity interval until the reconstruction loss difference is less than the allowable reconstruction loss threshold, thereby obtaining a rehabilitation training amplitude pattern recognition model of the lower limb hip and knee rehabilitation training equipment under the constraints of the lower limb hip and knee injury rating radar chart;

[0066] The current walking posture is identified by a rehabilitation training amplitude pattern recognition model to obtain a current rehabilitation training amplitude suitable for the target training object, and the power joints of the lower limb hip and knee rehabilitation training equipment are intelligently controlled according to the current rehabilitation training amplitude.

[0067] It should be noted that, since the degree of hip and knee joint injury may be severe or mild in different patients, this requires the lower limb hip and knee rehabilitation training equipment to have a reasonable range of activity to meet the corresponding injury training, and the lower limb hip and knee rehabilitation training equipment is preset with multiple adjustable rehabilitation training amplitudes or angles for each activity interval, so as to meet the patient's training needs for different strengths and flexibility of the hip and knee joints during the bending and straightening of the legs. However, the control method of the existing rehabilitation training equipment is difficult to accurately control the appropriate rehabilitation training amplitude for the patient according to the patient's walking posture habits and the severity of the injury, which may cause the patient's hip and knee joints to increase more load intensity and resistance during the extension and flexion process, which is easy to aggravate the degree of injury to the patient's hip and knee joints. Therefore, it is necessary for the lower limb hip and knee rehabilitation training equipment to be able to adapt the leg activity interval for the patient in advance according to the degree of hip and knee joint injury and further identify and limit the appropriate bending and straightening amplitude. To this end, this method evaluates lower limb hip and knee case reports according to the lower limb hip and knee injury rating criteria to generate a lower limb hip and knee injury rating radar chart for the target training subjects, thereby more quickly and meticulously clarifying the patient's hip and knee joint injury status; then, by constructing a set of preset activity intervals for different injury ratings and the corresponding rehabilitation training amplitudes for each preset activity interval, the variational autoencoder and variational autodecoder can be used to map the lower limb hip and knee injury rating radar chart to the potential recognition space of the lower limb hip and knee walking posture based on the constraint premise of the preset activity interval, and then the walking range can be quickly determined through variational inference learning of the implicit activity interval coding sequence.

[0068] It should be noted that variational inference learning can obtain multiple rehabilitation training amplitude patterns corresponding to preset activity ranges suitable for patients' hip and knee joint injuries. Since walking posture determines the patient's maximum tolerable rehabilitation training amplitude, the selection and control of these rehabilitation training amplitude patterns need to be determined based on the patient's current walking posture. Therefore, this method uses variational inference learning to obtain the reconstruction loss difference of rehabilitation training amplitude value planning under the constraints of the lower limb hip and knee injury rating radar chart. The reconstruction loss measures the difference between the samples reconstructed by the variational self-decoder and the original input, reflecting the model's ability and accuracy in recognition and generation. Excessive reconstruction loss may cause the recognition model's recognition results to fail to fully meet the input premise. Therefore, it is necessary to minimize the reconstruction loss to ensure that the rehabilitation training amplitude pattern generated by the variational self-decoder is as consistent as possible with the input's current walking posture. This improves the accuracy and reliability of rehabilitation training amplitude recognition based on the patient's injury constraint and meets the current walking posture. Ultimately, a rehabilitation training amplitude pattern recognition model for lower limb hip and knee rehabilitation training equipment is generated under the constraints of the lower limb hip and knee injury rating radar chart. The current rehabilitation training amplitude identified by the rehabilitation training amplitude pattern recognition model for the current walking posture is the patient's optimal rehabilitation training amplitude, which can avoid improper control of the lower limb hip and knee rehabilitation training equipment and realize intelligent control of the rehabilitation training of the power joints of the lower limb hip and knee rehabilitation training equipment.

[0069] More specifically, the step S102 is as follows: Figure 2 As shown, the specific steps include:

[0070] S202: Monitoring static data of the lower limb hip and knee joints of the target training subject using an inertial measurement unit sensor on the lower limb hip and knee rehabilitation training device, and defining an initial static gait of the target training subject based on the static data of the lower limb hip and knee joints;

[0071] S204: Acquiring force sensor monitoring data of a lower limb hip and knee rehabilitation training device, extracting force distribution data of a target training subject's thigh and calf during dynamic training on the lower limb hip and knee rehabilitation training device based on the force sensor monitoring data, and presetting a gait covariance function for the target training subject's dynamic training starting from an initial static gait based on the force distribution data;

[0072] S206: Acquire a human motion knowledge graph based on a big data network, identify the force distribution data using the human motion knowledge graph, and output gait force characteristics of the target training subject and a standard force motion law for dynamic walking starting from an initial static gait;

[0073] S208: constructing a Gaussian process regression model for gait transition, performing covariance calculation on a gait covariance function in the Gaussian process regression model based on the gait force characteristics to generate a covariance matrix, and determining a Gaussian gait prior distribution describing the target training subject's walking training from an initial stationary gait based on the covariance matrix;

[0074] S210: Construct a probability likelihood function for dynamic training starting from an initial static gait based on the standard force movement law, calculate the logarithm of the probability likelihood function and solve the partial derivative to obtain the gradient of the probability likelihood function, use the gradient to infer the Gaussian gait prior distribution to obtain the Gaussian gait posterior distribution, and determine the current walking posture of the target training object based on the description of the Gaussian gait posterior distribution.

[0075] It should be noted that the lower limb hip and knee rehabilitation training equipment mainly uses power joints to achieve power output and the movement of the exoskeleton robot. The change in the angle of the exoskeleton robot's knee joint causes the movement of the exoskeleton thigh and calf, changes the inclination angle of the thigh and calf, and achieves the rehabilitation training effect of the lower limb hip and knee joints. However, due to the differences in the gait posture formed between the thigh and calf of different patients, this means that the maximum range of active inclination angles that different patients can adapt to or withstand is also inconsistent. If the control of the equipment is difficult to adaptively and accurately control the flexion and straightening habits of the hip and knee joints according to the patient's walking gait, it may cause discomfort or even excessive stretching and pain in the patient's hip and knee joints during training, which is not conducive to the rapid recovery of the patient's hip and knee joints. Therefore, it is crucial to analyze and adapt the patient's walking posture for the control of the lower limb hip and knee rehabilitation training equipment. To this end, this method captures the target training subject's calf and hind legs, as contained in static data of the lower limb hip and knee joints, when bound to a lower limb hip and knee rehabilitation training device. This method then replicates the patient's static gait habits, known as the initial static gait. The method then sets the gait covariance function for dynamic training starting from the initial static gait based on the patient's dynamic training force distribution data. The force distribution data reveals the force trends at the start and end of the patient's gait as they walk. This force distribution data can be identified from the human motion knowledge graph, namely, the gait force characteristics and the standard force motion patterns for dynamic walking starting from the initial static gait. Gait force characteristics include gait force amplitude, gait force direction, and gait force. The gait covariance function determines the smoothness and periodic coherence of the random deduction of the walking posture, effectively capturing the underlying patterns of gait force changes and affecting the accuracy of the corresponding walking posture formed by different gait habits of different patients. Therefore, a corresponding gait covariance function must be set based on the force points of each patient's legs to perform walking posture deduction.

[0076] It should be noted that the covariance calculation of the gait covariance function is performed in the Gaussian process regression model based on the gait force characteristics. The generated covariance matrix provides the force distribution deduction and coordination information under the premise of gait force characteristics, so that the Gaussian process can effectively and accurately perform walking inference between different force distribution data points, making the Gaussian gait prior distribution of walking posture more accurate and reliable. Then, based on the standard force motion law, dynamic training is constructed from the initial static gait to obtain a gradient of the probability likelihood function. The gradient is used to infer the Gaussian gait prior distribution to obtain the Gaussian gait posterior distribution of the walking posture. The probability likelihood function is a bridge between the Gaussian process regression model and the force distribution data. It can more accurately explain the probability of the corresponding walking process of the gait following the patient's leg force trend based on the standard force motion law, reducing the control deviation caused by the randomness of the gait. The gradient represents the gait walking trajectory that best matches the patient's leg force trend. Based on the gradient, the Gaussian gait prior distribution is inferred to accurately obtain the dynamic walking posture of the patient applying the corresponding leg force under the initial static gait binding. The lower limb hip and knee rehabilitation training equipment can adaptively control the appropriate curved and straight range of motion for different patients based on this walking posture. This method can realize the deductive description of the walking form and posture of different patients, providing a more accurate and reliable control basis for the lower limb hip and knee rehabilitation training equipment, thereby realizing intelligent adaptive control.

[0077] More specifically, the method of obtaining a lower limb hip and knee case report of a target training subject and injury rating criteria for the lower limb hip and knee, overwriting and adding descriptions of the injury description information in the lower limb hip and knee case report with rating indicators using the description conditions of the injury rating criteria, and generating a radar chart of the lower limb hip and knee injury rating of the target training subject specifically includes the following steps:

[0078] Obtaining a lower limb hip and knee case report of a target training subject, and extracting a plurality of descriptive information about the target training subject's lower limb hip and knee injuries from the lower limb hip and knee case report;

[0079] Obtain injury rating criteria for the lower limb hip and knee based on a big data network, construct a set of descriptive conditions for each different injury rating based on the injury rating criteria, and define each descriptive condition as a conditional predicate;

[0080] A gain trade-off algorithm is introduced to calculate the information gain between each injury description and each conditional predicate. Based on the information gain, the coverage of the injury rating for the injury description is determined. The rating index of each injury rating in the injury rating criteria is obtained, and a radar feature matrix for lower limb hip and knee injury descriptions is established based on the rating index of each injury rating.

[0081] If the coverage is greater than the preset coverage, the injury rating corresponding to the conditional predicate is not added and is ignored; if the coverage is greater than the preset coverage, the injury rating corresponding to the conditional predicate is added to describe the rating feature of the injury description information, and the described rating feature is fitted into the radar feature matrix;

[0082] Repeat the above steps of adding the injury ratings corresponding to the conditional predicates to the radar feature matrix one by one to describe the injury description information until the rating features of all injury description information are traversed and the description fitting is completed, and finally a radar chart of the lower limb hip and knee injury ratings of the target training subject is generated.

[0083] It should be noted that existing control methods for lower limb hip and knee rehabilitation training equipment are not effective in assessing the severity of a patient's hip and knee injuries based on diagnosed case reports. This makes it difficult to identify and obtain the control range appropriate for the patient's rehabilitation training, reducing the device's control diversity and accuracy. Therefore, this method aims to maintain the most reasonable level of injury assessment based on the patient's case report, providing a more powerful control basis for the intelligent rehabilitation training control of subsequent lower limb rehabilitation training equipment. This method covers and adds the descriptive conditions for different injury ratings in the lower limb hip and knee injury rating criteria to each injury description in the case report in the form of conditional predicates, wherein the information gain represents the support of the conditional predicate for the injury description, reflecting the coverage of the injury rating for the injury description information, which is the key to accurate assessment. If the coverage is greater than the preset coverage, it means that the injury rating corresponding to the conditional predicate cannot accurately assess the injury description in a comprehensive manner, and other conditional predicates need to be selected for assessment. Therefore, the injury rating corresponding to the conditional predicate is not added and is ignored. Otherwise, it means that the injury rating corresponding to the conditional predicate can accurately assess the hip and knee joint injury degree corresponding to the injury description in a comprehensive manner, ensuring the credibility of the hip and knee joint injury assessment based on the patient's diagnosed case report. Therefore, the conditional predicate is added to describe the characteristics of the injury information and fit into the radar feature matrix. The comprehensive presentation of radar charts can improve the recognition rate of lower limb hip and knee rehabilitation training equipment for multi-indicator injury assessment, replacing the traditional tedious steps of manual analysis of different indicators one by one, reducing the amount of information data calculations, and optimizing equipment control efficiency and accuracy.

[0084] More specifically, the step S106 includes the following steps:

[0085] Obtaining electromyographic fatigue sensor data and the arrangement array of electromyographic sensors of lower limb hip and knee rehabilitation training equipment, obtaining a standard human body structure diagram of the lower limb hip and knee joints based on a big data network, constructing a multidimensional thermal density framework of the lower limb hip and knee joints based on the standard human body structure diagram, and presetting a density kernel function and bandwidth smoothing parameters based on the electromyographic fatigue sensor data;

[0086] Calculating a density contribution function of each myoelectric fatigue sensing data based on a density kernel function and a bandwidth smoothing parameter to obtain a plurality of density contribution functions, and thermally summing the plurality of density contribution functions in a multidimensional thermal density framework according to the arrangement array to obtain a multidimensional myoelectric fatigue thermal model;

[0087] The current walking posture of the target training subject is identified through the human motion knowledge graph to obtain the muscle movement distribution of the target training subject's lower limb hip and knee when walking. Based on the muscle movement distribution, the muscle fatigue area is delineated in the standard human body structure diagram and marked as the fatigue key point;

[0088] Construct the myoelectric fatigue thermal color gamut table of the target training object's fatigue-prone key points in a non-fatigued state, and obtain the myoelectric fatigue thermal chromaticity of the fatigue-prone key points through a multi-dimensional myoelectric fatigue thermal model;

[0089] If the electromyographic fatigue thermal color gamut table can query the electromyographic fatigue thermal chromaticity, the training rate and frequency of the lower limb hip and knee rehabilitation training equipment are controlled incrementally; if the electromyographic fatigue thermal color gamut table cannot query the electromyographic fatigue thermal chromaticity, the training rate and frequency of the lower limb hip and knee rehabilitation training equipment are controlled incrementally.

[0090] It should be noted that patients with different degrees of hip and knee injuries experience muscle fatigue after a period of training. Therefore, lower limb hip and knee rehabilitation training equipment is usually equipped with electromyographic sensors to detect the degree of rehabilitation training fatigue of the patient's calf muscles. However, existing lower limb hip and knee rehabilitation training equipment can only respond to simple warnings based on sensor detection results to remind patients to adjust training rate and frequency, etc., without corresponding control decisions to intelligently control based on muscle fatigue. This significantly reduces the intelligent level of rehabilitation training, increases the risk of patients' activities aggravating their injuries, and is not conducive to the rapid recovery of patients' hip and knee joints. To address this problem, this method uses a preset density kernel function and bandwidth smoothing parameter based on electromyographic fatigue sensor data. The density contribution of the electromyographic fatigue sensor data is then fitted based on this density kernel function and bandwidth smoothing parameter to construct a multidimensional electromyographic fatigue thermal model of the lower limb hip and knee joints. The thermal chromaticity displayed in this multidimensional electromyographic fatigue thermal model can quickly identify the patient's leg area with muscle fatigue. Then, a thermal query analysis is performed on the key fatigue points that drive muscle movement under the patient's current walking posture. If the EMG fatigue thermal color gamut table can be queried for EMG fatigue thermal chromaticity, it means that the muscle at the fatigue-prone key point is not yet in a fatigue state. Rehabilitation training can still be carried out according to the current device's operating rate and frequency, and even appropriately increased. Therefore, the training rate and frequency of the lower limb hip and knee rehabilitation training device are controlled incrementally. If the EMG fatigue thermal color gamut table does not query for EMG fatigue thermal chromaticity, it means that the muscle at the fatigue-prone key point on the leg is in a fatigue state after long-term training and is no longer suitable for rehabilitation training at the current device's operating rate and frequency. Therefore, the training rate and frequency of the lower limb hip and knee rehabilitation training device need to be controlled incrementally to reduce the intensity of rehabilitation training. This method can analyze whether the muscle is overly fatigued based on the expression of the thermal model fitted by EMG fatigue sensor data, thereby accurately controlling the training intensity of the lower limb hip and knee rehabilitation training device.

[0091] More specifically, the step S108 includes the following steps:

[0092] Acquire a real-time electromyographic fatigue signal of a lower limb hip and knee rehabilitation training device at a current training time sequence, introduce a wavelet transform algorithm to perform wavelet basis transform on the electromyographic fatigue signal and reconstruct the electromyographic state display, and obtain a training electromyogram of the lower limb hip and knee of the target training subject at the current training time sequence, which is defined as the current training electromyogram;

[0093] Acquire a training electromyogram of a lower limb hip and knee rehabilitation training device located on an adjacent training time sequence relative to the current training time sequence, which is defined as an adjacent training electromyogram;

[0094] Constructing a multidimensional space according to the time series change from the adjacent time series to the current time series, taking the current training electromyogram as the objective function, planning a simplex transferred from the adjacent training electromyogram to the current training electromyogram based on the multidimensional space, and calculating the objective function value of each simplex vertex on the simplex;

[0095] Only the simple vertex corresponding to the maximum objective function value is extracted and marked as the worst solution simple vertex. The reflection point positioning formula is introduced to calculate the reflection point of the worst solution simple vertex to obtain the objective function value of the reflection point.

[0096] If the objective function value of the reflection point is less than the objective function value of the worst solution simple vertex, then the reflection point is accepted and replaced with the worst solution simple vertex; if the objective function value of the reflection point is greater than the objective function value of the worst solution simple vertex, then a contraction operation is performed on the reflection point to generate a contraction point, and the contraction point is accepted to replace the worst solution simple vertex;

[0097] Based on the preset iteration frequency of the current training electromyogram, the above steps of calculating and analyzing the objective functions of the reflection points and the contraction points and replacing the simple vertex of the worst solution are repeated until the replacement iteration of the simple vertex of the worst solution reaches the iteration frequency, and the muscle fatigue time series state variables of the lower limb hip and knee of the target training subject transferred from the adjacent training electromyogram to the current training electromyogram are obtained;

[0098] The lower limb hip and knee timing state variables are input into the lower limb hip and knee rehabilitation training device to control and adjust the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training device.

[0099] It should be noted that myoelectric sensors can monitor the degree of muscle fatigue in the patient's thighs and calves during hip and knee joint rehabilitation training in real time. Therefore, the myoelectric fatigue sensor data at different training time sequences vary. Lower limb hip and knee rehabilitation training equipment should be controlled in real time according to the patient's muscle fatigue status during the rehabilitation training process, so that the equipment's rehabilitation training intensity regulation is more real-time and accurate. However, existing lower limb hip and knee rehabilitation training equipment does not have this function and still relies on manual judgment of muscle conditions for manual adjustment and control, which is time-consuming and labor-intensive. It is easy for the equipment to reduce the accuracy of the patient's hip and knee joint rehabilitation training, making it difficult to guarantee the effectiveness of rehabilitation training. Therefore, this method first obtains two adjacent training EMGs from the training time series: the current training EMG and the adjacent training EMG. The EMG state changes from the adjacent training EMG to the current training EMG are mapped as a simplex in the multidimensional space based on the time series. Since training EMGs are mostly presented in two dimensions, this simplex can be a triangle. This simplex reveals the state exploration range of the EMG signal when the adjacent training EMG changes to the current training EMG, quantifying the specific magnitude of the state transition. The objective function value of each simplex vertex on the simplex is then calculated. The objective function value determines the quality of the simplex vertex, providing reliable information on fatigue changes during the EMG state transition of each vertex in the current simplex.

[0100] It should be noted that the simple vertex corresponding to the maximum objective function value is the worst solution of the electromyographic state change, which is a position to be optimized that is difficult to reflect the temporal change of muscle fatigue. The reflection calculation can explore a better vertex about the worst solution simple vertex, but not every reflection point explored can replace the worst solution simple vertex. If it is replaced blindly, it may cause errors in the expression of muscle fatigue changes from the adjacent training electromyogram state change to the current training electromyogram. Therefore, this method determines whether the objective function value of the reflection point is less than the objective function value of the worst solution simple vertex. If it is less than, it means that the objective function value of the reflection point is better than that of the worst point. The objective function value is better, making the described muscle fatigue state change more linear and accurate, so the reflection point is accepted and replaced with the worst solution simple vertex; otherwise, it means that the objective function value of the reflection point is still not better than the worst solution simple vertex, and the objective function of the reflection point can be contracted to further express the muscle fatigue state change demand of the worst solution simple vertex. Finally, through continuous iteration, the muscle fatigue time series state variable of the target training object's lower limb hip and knee transferred from the adjacent training electromyogram to the current training electromyogram can be obtained, and the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training equipment can be controlled and adjusted according to the muscle fatigue time series state variable. Through this method, the state update change of muscle fatigue of the training electromyograms on two adjacent training time series can be analyzed and calculated, so that the lower limb hip and knee rehabilitation training equipment can maintain the real-time nature of the electromyographic fatigue monitoring data for the patient's rehabilitation training amplitude control, thereby reasonably adjusting and controlling the optimal rehabilitation training amplitude.

[0101] The second aspect of the present invention provides an intelligent control system for lower limb hip and knee rehabilitation training equipment, such as Figure 3 As shown, the intelligent control system of the lower limb hip and knee rehabilitation training equipment includes a memory 31 and a processor 32. The memory 31 stores an intelligent control method program for the lower limb hip and knee rehabilitation training equipment. When the intelligent control method program is executed by the processor 32, any step of the intelligent control method is implemented.

[0102] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. An intelligent control system for lower limb hip and knee rehabilitation training equipment, characterized in that: The intelligent control system of a lower limb hip and knee rehabilitation training device includes a memory and a processor, wherein the memory stores an intelligent control method program for the lower limb hip and knee rehabilitation training device, and when the intelligent control method program is executed by the processor, an intelligent control method for the lower limb hip and knee rehabilitation training device is implemented; The intelligent control method of the lower limb hip and knee rehabilitation training device comprises the following steps: S102: Obtain static data of the target training subject's lower limb hip and knee joints and force distribution data of the thigh and calf dynamic training, and perform Gaussian distribution calculation on the target training subject's gait using the static data of the lower limb hip and knee joints and the force distribution data to obtain the target training subject's current walking posture; S104: Based on the target training subject's lower limb hip and knee injury rating radar chart, a preset rehabilitation training strategy of the lower limb hip and knee rehabilitation training device is subjected to variational inference learning to identify the activity range and rehabilitation training amplitude, generating a rehabilitation training amplitude pattern recognition model. The rehabilitation training amplitude pattern recognition model is used to identify and obtain the current rehabilitation training amplitude of the current walking posture to control the lower limb hip and knee rehabilitation training device; S106: Identify and obtain fatigue-prone key points of the target training subject during the current walking posture, construct a multi-dimensional electromyographic fatigue thermal model based on electromyographic fatigue sensor data of the lower limb hip and knee rehabilitation training device, and use the multi-dimensional electromyographic fatigue thermal model to perform thermal analysis on the fatigue-prone key points to control the rate and frequency of the lower limb hip and knee rehabilitation training device; S108: calculating the lower limb hip and knee time sequence state variables of the lower limb hip and knee rehabilitation training device transferred from the adjacent training electromyogram to the current training electromyogram by reflex contraction replacement of the worst solution simple vertex, and controlling the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training device according to the change in the lower limb hip and knee time sequence state; The step S104 specifically includes the following steps: Obtaining a preset rehabilitation training strategy for a lower limb hip and knee rehabilitation training device, obtaining preset activity intervals for the lower limb hip and knee rehabilitation training device for different injury ratings based on the preset rehabilitation training strategy, and obtaining rehabilitation training amplitudes corresponding to each preset activity interval; Obtain the lower limb hip and knee case reports and injury rating criteria for the target training subjects, overwrite and add descriptions of the rating indicators to each injury description in the lower limb hip and knee case reports using the description conditions of the injury rating criteria, and generate a radar chart of the lower limb hip and knee injury ratings for the target training subjects; A variational autoencoder is constructed based on the preset activity intervals of the lower limb hip and knee rehabilitation training device for different injury ratings, and a variational autodecoder is constructed based on the rehabilitation training amplitude corresponding to each preset activity interval; Constructing a latent recognition space for lower limb hip and knee walking postures, inputting the lower limb hip and knee injury rating radar chart into a variational autoencoder and mapping it to the latent recognition space to obtain an implicit activity interval coding sequence, and performing variational inference learning on the latent variable distribution of the implicit activity interval coding sequence in a variational autodecoder; Through variational inference learning, the reconstruction loss difference of the rehabilitation training range value planning under the constraints of the lower limb hip and knee injury rating radar chart is obtained, and the allowable reconstruction loss threshold is preset based on the current walking posture of the target training subject; If the reconstruction loss difference is greater than the allowable reconstruction loss threshold, a minimization process is performed on the reconstruction loss difference to train the rehabilitation training amplitude corresponding to the activity interval until the reconstruction loss difference is less than the allowable reconstruction loss threshold, thereby obtaining a rehabilitation training amplitude pattern recognition model of the lower limb hip and knee rehabilitation training equipment under the constraints of the lower limb hip and knee injury rating radar chart; The current walking posture is identified by a rehabilitation training amplitude pattern recognition model to obtain a current rehabilitation training amplitude suitable for the target training object, and the power joints of the lower limb hip and knee rehabilitation training equipment are intelligently controlled according to the current rehabilitation training amplitude.

2. The intelligent control system of the lower limb hip and knee rehabilitation training equipment according to claim 1, characterized in that: The step S102 specifically includes the following steps: monitoring static data of the lower limb hip and knee joints of the target training subject through an inertial measurement unit sensor on a lower limb hip and knee rehabilitation training device, and defining an initial static gait of the target training subject based on the static data of the lower limb hip and knee joints; acquiring force sensor monitoring data of a lower limb hip and knee rehabilitation training device, extracting force distribution data of a target training subject's thigh and calf during dynamic training on the lower limb hip and knee rehabilitation training device through the force sensor monitoring data, and presetting a gait covariance function of the target training subject's dynamic training starting from an initial stationary gait based on the force distribution data; Acquire a human motion knowledge graph based on a big data network, identify the force distribution data through the human motion knowledge graph, and output the gait force characteristics of the target training subject and the standard force movement law of dynamic walking starting from an initial static gait; Constructing a Gaussian process regression model for gait transfer, performing covariance calculation on a gait covariance function in the Gaussian process regression model based on the gait force characteristics to generate a covariance matrix, and determining a Gaussian gait prior distribution describing the target training subject's walking training from an initial stationary gait based on the covariance matrix; Based on the standard force movement law, a probability likelihood function for dynamic training starting from an initial static gait is constructed, the logarithm of the probability likelihood function is calculated and the partial derivative is solved to obtain the gradient of the probability likelihood function, the gradient is used to infer the Gaussian gait prior distribution to obtain the Gaussian gait posterior distribution, and the current walking posture of the target training object is determined based on the description of the Gaussian gait posterior distribution.

3. The intelligent control system of the lower limb hip and knee rehabilitation training equipment according to claim 1, characterized in that: The method of obtaining a lower limb hip and knee case report of a target training subject and a criterion for rating the injury of the lower limb hip and knee, overwriting and adding descriptions of rating indicators to each injury description information in the lower limb hip and knee case report using the description conditions of the injury rating criterion, and generating a radar chart for the rating of the lower limb hip and knee injury of the target training subject specifically includes the following steps: Obtaining a lower limb hip and knee case report of a target training subject, and extracting a plurality of descriptive information about the target training subject's lower limb hip and knee injuries from the lower limb hip and knee case report; Obtain injury rating criteria for the lower limb hip and knee based on a big data network, construct a set of descriptive conditions for each different injury rating based on the injury rating criteria, and define each descriptive condition as a conditional predicate; A gain trade-off algorithm is introduced to calculate the information gain between each injury description and each conditional predicate. Based on the information gain, the coverage of the injury rating for the injury description is determined. The rating index of each injury rating in the injury rating criteria is obtained, and a radar feature matrix for lower limb hip and knee injury descriptions is established based on the rating index of each injury rating. If the coverage is greater than the preset coverage, the injury rating corresponding to the conditional predicate is not added and is ignored; if the coverage is greater than the preset coverage, the injury rating corresponding to the conditional predicate is added to describe the rating feature of the injury description information, and the described rating feature is fitted into the radar feature matrix; Repeat the above steps of adding the injury ratings corresponding to the conditional predicates to the radar feature matrix one by one to describe the injury description information until the rating features of all injury description information are traversed and the description fitting is completed, and finally a radar chart of the lower limb hip and knee injury ratings of the target training subject is generated.

4. The intelligent control system of the lower limb hip and knee rehabilitation training equipment according to claim 1, characterized in that: The step S106 specifically includes the following steps: Obtaining electromyographic fatigue sensor data and the arrangement array of electromyographic sensors of lower limb hip and knee rehabilitation training equipment, obtaining a standard human body structure diagram of the lower limb hip and knee joints based on a big data network, constructing a multidimensional thermal density framework of the lower limb hip and knee joints based on the standard human body structure diagram, and presetting a density kernel function and bandwidth smoothing parameters based on the electromyographic fatigue sensor data; Calculating a density contribution function of each myoelectric fatigue sensing data based on a density kernel function and a bandwidth smoothing parameter to obtain a plurality of density contribution functions, and thermally summing the plurality of density contribution functions in a multidimensional thermal density framework according to the arrangement array to obtain a multidimensional myoelectric fatigue thermal model; The current walking posture of the target training subject is identified through the human motion knowledge graph to obtain the muscle movement distribution of the target training subject's lower limb hip and knee when walking. Based on the muscle movement distribution, the muscle fatigue area is delineated in the standard human body structure diagram and marked as the fatigue key point; Construct the myoelectric fatigue thermal color gamut table of the target training object's fatigue-prone key points in a non-fatigued state, and obtain the myoelectric fatigue thermal chromaticity of the fatigue-prone key points through a multi-dimensional myoelectric fatigue thermal model; If the electromyographic fatigue thermal color gamut table can query the electromyographic fatigue thermal chromaticity, the training rate and frequency of the lower limb hip and knee rehabilitation training equipment are controlled incrementally; if the electromyographic fatigue thermal color gamut table cannot query the electromyographic fatigue thermal chromaticity, the training rate and frequency of the lower limb hip and knee rehabilitation training equipment are controlled incrementally.

5. The intelligent control system of the lower limb hip and knee rehabilitation training equipment according to claim 1, characterized in that: The step S108 specifically includes the following steps: Acquire a real-time electromyographic fatigue signal of a lower limb hip and knee rehabilitation training device at a current training time sequence, introduce a wavelet transform algorithm to perform wavelet basis transform on the electromyographic fatigue signal and reconstruct the electromyographic state display, and obtain a training electromyogram of the lower limb hip and knee of the target training subject at the current training time sequence, which is defined as the current training electromyogram; Acquire a training electromyogram of a lower limb hip and knee rehabilitation training device located on an adjacent training time sequence relative to the current training time sequence, which is defined as an adjacent training electromyogram; Constructing a multidimensional space according to the time series change from the adjacent time series to the current time series, taking the current training electromyogram as the objective function, planning a simplex transferred from the adjacent training electromyogram to the current training electromyogram based on the multidimensional space, and calculating the objective function value of each simplex vertex on the simplex; Only the simple vertex corresponding to the maximum objective function value is extracted and marked as the worst solution simple vertex. The reflection point positioning formula is introduced to calculate the reflection point of the worst solution simple vertex to obtain the objective function value of the reflection point. If the objective function value of the reflection point is less than the objective function value of the worst solution simple vertex, then the reflection point is accepted and replaced with the worst solution simple vertex; if the objective function value of the reflection point is greater than the objective function value of the worst solution simple vertex, then a contraction operation is performed on the reflection point to generate a contraction point, and the contraction point is accepted to replace the worst solution simple vertex; Based on the preset iteration frequency of the current training electromyogram, the above steps of calculating and analyzing the objective functions of the reflection points and the contraction points and replacing the simple vertex of the worst solution are repeated until the replacement iteration of the simple vertex of the worst solution reaches the iteration frequency, and the lower limb hip and knee temporal state variables of the target training object transferred from the adjacent training electromyogram to the current training electromyogram are obtained; The lower limb hip and knee timing state variables are input into the lower limb hip and knee rehabilitation training device to control and adjust the current rehabilitation training amplitude of the lower limb hip and knee rehabilitation training device.

Images