Abnormal gait real-time detection system, training method thereof and rehabilitation exoskeleton equipment

Abstract

The invention belongs to the related technical field of rehabilitation medical assistance, and discloses an abnormal gait real-time detection system, a training method thereof and rehabilitation exoskeleton equipment, and the system comprises an acquisition data acquisition module which constructs a multi-dimensional time sequence based on acquisition data; the pathological data acquisition module is used for acquiring pathological characteristics of a patient; the hybrid standardization processing module is used for performing standardization processing according to the pathological feature guidance sequence; the low-pass filter bank design module is used for guiding the learning of the cut-off frequency of the corresponding low-pass filter according to the pathological characteristics; the self-adaptive filtering module is used for filtering by using a filter to obtain an enhanced sequence; and the recognition module enables the sequence to pass through a deep learning network to obtain a gait recognition result of the corresponding lower limb. According to the method, the pathological features of the patient are adopted to guide feature processing, learnable adaptive filtering is realized, the gait recognition precision can be effectively improved, and rehabilitation exoskeleton equipment can provide precise assistance for the patient.

Description

Technical Field

[0001] The present invention belongs to the technical field related to rehabilitation medical assistance, and more specifically, relates to an abnormal gait real-time detection system, a training method thereof, and a rehabilitation exoskeleton device. Background Art

[0002] Human gait is a periodic cycle, and usually, one complete gait cycle is defined as from the touchdown of one heel to the next touchdown of the same heel. However, for patients with lower limb motor dysfunction (such as foot drop or hemiplegia patients), due to abnormal neuromuscular control, the gait cycle often exhibits significant asymmetry and spatio-temporal characteristic disorders. Such abnormal gait usually accompanies problems such as blurred cycle boundaries, lag or advance of phase events, which greatly affects the accuracy of gait analysis and restricts the application of rehabilitation exoskeleton devices in dynamic adaptive assistance.

[0003] At present, significant progress has been made in the research on gait phase and event detection. For example, the gait phase estimation algorithm based on long short-term memory network proposed by Mustafa Sarshar et al. realizes high-precision gait phase estimation through the data of triaxial inertial sensors at the ankle; the time-delay neural network method developed by Shushtari et al. demonstrates excellent robustness and accuracy; the gait phase estimation method based on bidirectional LSTM proposed by Lee et al. is applicable to the control of robotic prosthetics at different walking speeds; Kang et al. proposed a gait phase estimation method based on convolutional neural network and successfully applied it to the real-time control of robotic hip exoskeletons in multi-mode movements. However, the above research has only achieved significant results in healthy populations, but there are difficulties in identifying dynamic gait abnormalities (such as dragging movements) caused by dysfunctions such as foot drop and insufficient hip internal rotation. Summary of the Invention

[0004] In view of the above defects or improvement requirements of the prior art, the present invention provides an abnormal gait real-time detection system, a training method thereof, and a rehabilitation exoskeleton device, aiming to accurately realize the gait detection of abnormal gait and provide precise assistance for patients with motor disorders.

[0005] To achieve the above object, according to the first aspect of the present invention, there is provided an abnormal gait real-time detection system, which includes: A data acquisition module for acquiring gait-related data of the same lower limb of a patient to construct multi-dimensional features, and the multi-dimensional features at multiple consecutive moments constitute a multi-dimensional time series X e ; A pathological data acquisition module for acquiring the pathological features Y of the patient, where the pathological features are a score set of different evaluation indexes of gait; The hybrid normalization processing module is used to calculate the ratio of the total score of feature Y to the highest total score of each evaluation index as the pathological weight λ, and perform hybrid normalization processing on each dimension of the sequence X e in the sequence X u , and the hybrid normalization processing includes making the feature obey the standard normal distribution to obtain the first processing result, making the feature amplitude ratio normalized to obtain the second processing result, and performing weighted summation on the two processing results. The weight of the first processing result is positively correlated with λ, and the weight of the first processing result is negatively correlated with λ; The low-pass filter bank design module is used to map feature Y through M fully connected layer networks respectively to obtain the dynamic frequency adjustment amounts of M groups of low-pass filters, and fuse the dynamic frequency adjustment amounts of each low-pass filter in the filter bank with its base frequency to obtain the cut-off frequencies of each low-pass filter in the corresponding filter bank, where M>1 is the feature dimension; The adaptive filtering module is used to perform convolution processing on the sequence X u using M groups of low-pass filters. The M groups of low-pass filters correspond one by one to M dimensions. Each group of low-pass filters processes the feature sequence of the corresponding dimension. The feature sequences of each dimension are respectively convolved by N low-pass filters in the corresponding low-pass filter bank to obtain N convolution sequences of this dimension. The N convolution sequences of the same dimension are first passed through the channel attention network and then fused to obtain the enhanced sequence of this dimension, where N≥1; The recognition module is used to make the multi-dimensional time series X z composed of the enhanced sequences based on M dimensions pass through the deep learning network to obtain the gait recognition result corresponding to the lower limb.

[0006] Optionally, the gait-related data includes knee joint angle, ankle joint angle, and three-axis angular velocity and three-axis acceleration collected at specific positions, and the specific positions include the front side of the thigh, the front side of the calf, and the instep.

[0007] Optionally, different evaluation indexes of gait include starting, lifting height, step length, gait symmetry, step continuity, walking path, trunk stability, and step width.

[0008] Optionally, a gating mechanism is used to fuse the dynamic frequency adjustment amount of each low-pass filter in the filter bank with its base frequency to obtain the cut-off frequency of each low-pass filter in the corresponding filter bank, and the fusion formula is: ; In the formula, f n c is the cut-off frequency of the nth low-pass filter of any low-pass filter bank, f n a is the base frequency of the nth low-pass filter, f n ' is the dynamic frequency adjustment amount of the nth low-pass filter, and α nis a trainable scaling factor, f max and f min are respectively the set maximum and minimum frequencies, and sigmoid() is the activation function.

[0009] Optionally, the system further includes: An adaptive adjustment module for mapping the feature Y to the feature h through a fully connected layer network 2 , and making the feature h 2 respectively perform linear mapping through two linear mapping networks to obtain an M-dimensional feature amplitude adjustment amount ρ and an M-dimensional baseline offset correction amount ε, and use the adjustment amount ρ to weight the sequence X f and then fuse it with the sequence X f and the baseline offset correction amount ε to obtain the sequence X Z ; wherein, the sequence X f is directly composed of enhanced sequences of M dimensions, and the i-th component in the adjustment amount ρ is used as the weighting weight of the i-th dimensional enhanced sequence in X f , and when fusing, each eigenvalue of the i-th dimensional enhanced sequence in X f is summed with the i-th component in the correction amount ε.

[0010] Optionally, the gait recognition includes single-task recognition or multi-task recognition; The single-task recognition includes: classification of the stance phase and the swing phase, or recognition of continuous gait phase parameters; The multi-task recognition includes: classification of the stance phase and the swing phase, and recognition of continuous gait phase parameters.

[0011] Optionally, the gait recognition result includes the recognition of continuous gait phase parameters, and the deep learning network outputs the gait phase parameters in the form of a two-dimensional polar coordinate vector.

[0012] According to the second aspect of the present invention, there is provided a training method for the abnormal gait real-time detection system described in the first aspect, which includes: Obtaining a training data set, each group of training data including the pathological characteristics of the patient and the gait-related data of the patient; Adding gait labels to the gait-related data in each group of training data; Inputting the training data into the system; The gait-related data of the same lower limb of the patient is acquired by the acquisition data obtaining module in the system to construct multi-dimensional features, and the multi-dimensional features at multiple consecutive moments form a multi-dimensional time series Xe; the pathological features Y of the patient are acquired by the pathological data obtaining module in the system, and the pathological features are a score set of different evaluation indexes of the gait; the total score value of the feature Y and the ratio of the highest total score value of each evaluation index are calculated by the mixed normalization processing module in the system as the pathological weight λ, and each dimension feature in the sequence Xe is subjected to mixed normalization processing to obtain a sequence Xu. The mixed normalization processing includes making the feature obey the standard normal distribution to obtain a first processing result and making the feature amplitude ratio normalized to obtain a second processing result and weighted summing the two processing results. The weight of the first processing result is positively correlated with λ, and the weight of the first processing result is negatively correlated with λ; the feature Y is respectively mapped by the low-pass filter bank design module in the system through M fully-connected layer networks into the dynamic frequency adjustment amounts of M groups of low-pass filters, and the dynamic frequency adjustment amounts of each low-pass filter in the filter bank are fused with its base frequency to obtain the cut-off frequency of each low-pass filter in the corresponding filter bank, where M>1 is the feature dimension; the sequence Xu is subjected to convolution processing by the adaptive filtering module in the system using M groups of low-pass filters. The M groups of low-pass filters correspond one-to-one with M dimensions, and each group of low-pass filters processes the feature sequence of the corresponding dimension. The feature sequences of each dimension are respectively subjected to convolution calculation by N low-pass filters in the corresponding low-pass filter bank to obtain N convolution sequences of this dimension. The N convolution sequences of the same dimension are first passed through the channel attention network and then fused to obtain the enhanced sequence of this dimension, where N≥1; the multi-dimensional time series Xz composed of the enhanced sequences based on M dimensions is passed through a deep learning network by the recognition module in the system to obtain the gait recognition result of the corresponding lower limb; the loss between the gait recognition result and the gait label is calculated to train the system until the loss between the gait recognition result output by the system and the gait label converges.

[0013] Optionally, the gait recognition result includes the classification results of the stance phase and the swing phase; the gait label includes the classification labels of the stance phase and the swing phase; The process of adding the classification labels of the stance phase and the swing phase to the gait-related data of a certain patient includes: Obtain the sagittal plane angular velocity sequence collected by the patient and calculate its mean value ξ and maximum value ω max ; Generate an adaptive coefficient according to the pathological weight λ of the patient ; β min and β maxThey are set values respectively; the pathological weight λ is the ratio of the total score of the pathological features of the patient to the highest total score of each evaluation index. Calculate the angular velocity detection threshold ; Combine the sagittal plane angular velocity curve and the angular velocity detection threshold ω(λ) to identify the heel strike moment HS and the toe off moment TO. Find the maximum point exceeding the angular velocity detection threshold ω(λ) in the curve as the threshold point. Take the first minimum point before the threshold point in the curve as the moment TO, and the first minimum point after the threshold point as the moment HS. If the current gait-related data is in the time interval from the moment TO to the moment HS, it is marked as the stance phase. If the current gait-related data is in the time interval from the moment HS to the moment TO, it is marked as the swing phase.

[0014] According to the third aspect of the present invention, a rehabilitation exoskeleton device is provided, which includes a multi-node wearable data acquisition system, a driving device and a controller. The controller includes the abnormal gait real-time detection system as described in the first aspect. The multi-node wearable data acquisition system is used to collect the gait-related data of the patient and transmit it to the controller; the controller is used to input the gait-related data and the pathological data of the patient into the abnormal gait real-time detection system, and control the driving device to assist the corresponding lower limb movement after obtaining the current gait recognition result.

[0015] Generally speaking, compared with the prior art through the above technical solutions conceived by the present invention, the present invention mainly has the following beneficial effects.

[0016] 1. In the present invention, the pathological features of the patient are introduced to perform hybrid normalization processing on the features of the multi-dimensional time series. When the total score of the pathological features of the patient is higher, it indicates that the gait regularity of the patient is stronger. At this time, it is more inclined to make the features follow the standard normal distribution to eliminate the distribution shift; when the total score of the pathological features of the patient is lower, it indicates that the gait regularity of the patient is weaker. At this time, it is more inclined to perform feature amplitude ratio normalization to retain the drastic change characteristics of the original amplitude and avoid the Z-score from compressing the information of the outliers. Therefore, performing the above hybrid normalization processing on the collected data helps to improve the gait recognition accuracy and enables the rehabilitation exoskeleton device to provide accurate assistance for the patient.

[0017] 2. In the present invention, the cut-off frequency of the low-pass filter is a learnable parameter, so that adaptive filtering of the features can be achieved. Through adaptive filtering, the filtering range can be dynamically adjusted according to the individual pathological features of the patient, filtering out noise interference while retaining the effective gait information, and more accurately distinguishing the real motion features from the abnormal noise, thereby significantly improving the accuracy of gait phase detection.

[0018] 3. In the present invention, the pathological characteristics of the patient are introduced to guide the design of the low-pass filter, and a learnable mapping relationship between the cut-off frequency and the patient's pathological characteristics is established, so as to realize the spectral characteristic adaptation based on the individual pathological state, which helps to improve the gait recognition accuracy and enables the rehabilitation exoskeleton device to provide precise assistance for the patient.

[0019] 4. In one embodiment, the dynamic frequency adjustment amount of each low-pass filter in the filter bank is fused with its base frequency through a gating mechanism to obtain the cut-off frequency of each low-pass filter in the corresponding filter bank. In this way, the adaptive regulation of the cut-off frequency can be realized according to the patient's pathological characteristics. This mechanism not only retains the physical meaning of the base frequency but also generates personalized frequency band selection in combination with the degree of the patient's gait disorder, effectively suppressing the high-frequency noise and motion interference in signal acquisition, while retaining the time-domain characteristics of the key gait phase events and improving the signal-to-noise ratio of the abnormal gait signal.

[0020] 5. In one embodiment, an adaptive adjustment module is set up to adaptively adjust the enhanced features by introducing the pathological characteristics. In this way, the signal features can be adaptively optimized according to the degree of the individual's gait disorder, so as to more accurately capture the subtle changes of the abnormal gait, and enhance the resolution ability of the model for complex gait patterns.

[0021] 6. In one embodiment, the deep learning network outputs the gait phase parameters in the form of two-dimensional polar coordinate vectors. This mapping method converts the linearly increasing phase parameter θ into a continuously changing periodic signal, eliminating the jump problem at the HS event. In this way, it is beneficial to system training, helps to improve the gait recognition accuracy, and enables the rehabilitation exoskeleton device to provide precise assistance for the patient.

[0022] 7. In one embodiment, the pathological characteristics of the patient are introduced to guide the determination of the angular velocity detection threshold, and the threshold can be adaptively adjusted according to the patient's pathological condition. Thus, the moments of heel strike and toe off can be marked more accurately, so as to accurately divide the stance phase and the swing phase, and then train a more accurate real-time abnormal gait detection system, enabling the rehabilitation exoskeleton device to provide precise assistance for the patient. Description of the Drawings

[0023] Figure 1 is the structural block diagram of the real-time abnormal gait detection system in one embodiment of the present invention.

[0024] Figure 2 is the structural block diagram of the real-time abnormal gait detection system in another embodiment of the present invention.

[0025] Figure 3 is the schematic diagram of the wearable data acquisition system for gait estimation in one embodiment of the present invention.

[0026] Figure 4It is a schematic structural diagram of an adaptive filtering module in an embodiment of the present invention.

[0027] Figure 5 It is a comparison diagram of gait labels in the form of continuous gait phase parameters and two-dimensional polar coordinate vectors in an embodiment of the present invention. Detailed implementation manners

[0028] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0029] Embodiment 1 The present invention provides a real-time abnormal gait detection system. As Figure 1 shown is a structural block diagram of a real-time abnormal gait detection system in an embodiment of the present invention, which includes a collected data acquisition module, a pathological data acquisition module, a hybrid normalization processing module, a low-pass filter bank design module, an adaptive filtering module, and an identification module. Each of the modules will be introduced in detail below.

[0030] The collected data acquisition module is used to acquire gait-related data of the same lower limb of a patient to construct multi-dimensional features, and multi-dimensional features at multiple consecutive moments form a multi-dimensional time series X e .

[0031] The present invention adopts a single-leg detection method. If the left leg needs to be assisted, the left leg data is collected, and the left leg gait is detected based on the left leg data. If the right leg needs to be assisted, the right leg data is collected, and the right leg gait is detected based on the right leg data. If both legs need to be assisted, the data of both legs are collected simultaneously, the left leg gait is detected based on the left leg data, and the right leg gait is detected based on the right leg data.

[0032] Specifically, in order to improve the detection accuracy, information at multiple positions is collected. The gait-related data may include, for example, knee joint angles, ankle joint angles, and three-axis angular velocities and three-axis accelerations collected at specific positions. The specific positions include the front side of the thigh, the front side of the calf, and the instep, etc.

[0033] Specifically, before data processing, the signals at the front side of the thigh, the front side of the calf, and the instep of the same lower limb are collected first. In one embodiment, in order to obtain high-precision gait phase detection data, a multi-node wearable data acquisition system is built, as Figure 3The figure shows a schematic diagram of a wearable data acquisition system for gait estimation in an embodiment of the present invention. Taking the two legs as an example, the system uses six micro inertial measurement units (IMUs) as the core sensors, which are respectively installed on the front side of the thighs, the front side of the calves, and the instep positions of the legs of the subject to achieve a comprehensive capture of the lower limb movement information. Each IMU records its quaternion, three-axis angular velocity, and three-axis acceleration in real time at the same sampling frequency (such as 100 Hz), and synchronously communicates with the controller through a wireless transmission module to transmit the collected data to the controller.

[0034] Among them, based on the quaternion data collected by the 3 IMUs on the same lower limb, through the attitude relationship between adjacent body segments, the quaternion difference method can be used to accurately calculate the angles of the knee joint and ankle joint of the lower limb.

[0035] At this time, based on the gait-related data of the same lower limb, a multi-dimensional feature X(t) containing M dimensions at time t for the lower limb can be obtained. The multi-dimensional features X(t) at multiple consecutive times constitute a multi-dimensional time series, X e ∈R T╳M , where T is the number of time instants, that is, the sequence length. The multi-dimensional feature at each time instant can include the three-axis angular velocity (9 dimensions), three-axis acceleration (9 dimensions) collected by the 3 IMUs at that time instant, as well as the calculated knee joint angle (1 dimension) and ankle joint angle (1 dimension).

[0036] Furthermore, after the signal is collected, preprocessing can be performed first, such as performing simple denoising processing on the IMU data, and then performing Z-Score normalization processing on the acceleration and angular velocity.

[0037] The pathological data acquisition module is used to acquire the patient's pathological feature Y, and the pathological feature is a score set of different evaluation indexes of gait.

[0038] Since the degrees of movement disorders of different patients are different, that is, there are differences between individuals. In the present invention, the patient's pathological feature Y is acquired, and the pathological feature Y can reflect the degrees of movement disorders of different individuals. Introducing the pathological feature Y can improve the accuracy of gait recognition.

[0039] Specifically, the patient's pathological feature Y can be constructed based on the patient's Tinetti gait score form. The different evaluation indexes of gait include starting, lifting height of the foot, step length, gait symmetry, step continuity, walking path, trunk stability, and step width.

[0040] Experienced doctors evaluate the gait ability of patients according to the Tinetti gait score form. The doctors test the patients through 8 gait test items including starting, lifting height of the foot, step length, gait symmetry, step continuity, walking path, trunk stability, and step width, and score each item. The score can be set as (0, 1, 2), where a higher score indicates better gait ability of the patient. Thus, 8 parameters of the gait ability of each patient are obtained as the pathological characteristics of the patient , where y k is the score of the k-th index.

[0041] The mixed normalization processing module is used to calculate the ratio of the total score of the feature Y to the highest total score of the pathological characteristics as the pathological weight λ, and perform mixed normalization processing on each dimension feature in the sequence X e to obtain the sequence X u . The mixed normalization processing includes making the feature obey the standard normal distribution to obtain the first processing result, making the feature amplitude ratio normalized to obtain the second processing result, and weighted summing the two processing results. The weight of the first processing result is positively correlated with λ, and the weight of the first processing result is negatively correlated with λ.

[0042] Specifically, , , is the highest total score of the pathological characteristics, that is, the summation result after each index takes the highest score.

[0043] Specifically, the mixed normalization processing can be written in the following function form: ; In the formula, x is one of the data in any dimension of the sequence X e The mean, variance, and maximum value of all data in the dimension to which x belongs are μ, σ, and max respectively, and c is a preset scale factor.

[0044] Actually, the first term in the above function is the conversion function for making the feature obey the standard normal distribution, and the second term is the conversion function for making the feature amplitude ratio normalized. Based on this mixed normalization processing, when the total score S of the patient's pathological feature Y is higher, it indicates that the gait regularity of the patient is stronger. At this time, it is more inclined to make the feature obey the standard normal distribution to eliminate the distribution shift. When the total score S of the patient's pathological feature Y is lower, it indicates that the gait regularity of the patient is weaker. At this time, it is more inclined to make the feature amplitude ratio normalized to retain the drastic change characteristics of the original amplitude and avoid the Z-score from compressing the information of outliers. Therefore, performing the above mixed normalization processing on the collected data helps to improve the gait recognition accuracy.

[0045] The low-pass filter bank design module is used to map the feature Y through M fully-connected layer networks respectively to obtain the dynamic frequency adjustment amounts of M groups of low-pass filters, and fuse the dynamic frequency adjustment amounts of each low-pass filter in the filter bank with its base frequency to obtain the cut-off frequencies of each low-pass filter in the corresponding filter bank, where M is the feature dimension.

[0046] Specifically, the low-pass filter bank design module includes M parallel fully-connected layer networks. Each fully-connected layer network corresponds to the design of a group of low-pass filters. When the feature Y is input into the fully-connected layer network for feature mapping, the dynamic frequency adjustment amounts of each low-pass filter in the corresponding low-pass filter combination can be obtained.

[0047] In one embodiment, each fully-connected layer network includes two fully-connected layers. In each fully-connected layer network, the feature is first mapped to a high-dimensional latent space through a fully-connected layer to obtain an intermediate representation , and the intermediate representation h 1 is then mapped to the dynamic frequency adjustment amount of the corresponding low-pass filter bank through a fully-connected layer , where N is the number of low-pass filters in the corresponding low-pass filter bank, and f n ' is the dynamic frequency adjustment amount of the nth low-pass filter in the corresponding filter bank.

[0048] Specifically, the process of obtaining the intermediate representation h 1 can be written as the following expression: h 1 = ReLU(W 1 Y + b 1 ); In the formula, W 1 ∈R 32╳8 is a trainable weight matrix, b 1 ∈R 32 is a trainable bias term, and ReLU() represents an activation function.

[0049] Specifically, the process of obtaining the dynamic frequency adjustment amount f' can be written as the following expression: f' = W 2 h 1 + b 2 ; In the formula, W 2 ∈R N╳32 is a trainable weight matrix, b 2 ∈R N is a trainable bias term.

[0050] After obtaining the dynamic frequency adjustment amount of each low-pass filter, the dynamic frequency adjustment amount of the low-pass filter is fused with its set base frequency to obtain the cut-off frequency of the corresponding low-pass filter.

[0051] In one embodiment, the dynamic frequency adjustment amount of the low-pass filter can be directly superimposed on its set base frequency to obtain the cut-off frequency corresponding to the low-pass filter.

[0052] In another embodiment, the dynamic frequency adjustment amount of the low-pass filter and its set base frequency can also be fused through a gating mechanism. Taking the nth low-pass filter of any low-pass filter bank as an example, it uses the gating mechanism to fuse the dynamic frequency adjustment amount with its base frequency to obtain the cut-off frequency of the low-pass filter. The fusion process can be written as the following expression: ; In the formula, f n c is the cut-off frequency of the nth low-pass filter of any low-pass filter bank, f n a is the base frequency of the nth low-pass filter, f n ' is the dynamic frequency adjustment amount of the nth low-pass filter, α n is a trainable scaling factor used to dynamically adjust the sensitivity so that the cut-off frequency can be dynamically adapted to the pathological features. f max and f min are respectively the set maximum and minimum frequencies, that is, the frequency band range is preset. sigmoid() is an activation function that maps the unbounded input to the interval (0, 1). In this way, the cut-off frequency is scaled to the interval (f max , f min ).

[0053] In this embodiment, the frequencies are fused through the gating mechanism. In this way, the adaptive regulation of the cut-off frequency can be realized according to the patient's pathological features. This mechanism not only retains the physical meaning of the base frequency but also generates personalized frequency band selection in combination with the degree of the patient's gait disorder, effectively suppressing the high-frequency noise and motion interference in signal acquisition, while retaining the time-domain features of key gait phase events and improving the signal-to-noise ratio of abnormal gait signals.

[0054] It should be noted that the base frequency of the low-pass filter can be set according to the prior knowledge of the gait cycle and event characteristics, and specifically, it can be set to be the same or different.

[0055] In one embodiment, the base frequencies of the N low-pass filters in each low-pass filter bank can be set to be uniformly distributed in the interval (f max , f min ). In this way, a multi-scale filtering network covering the target frequency domain range can be constructed in the initial design stage, and the dynamic characteristics of different frequency components in the gait signal can be captured through the differential distribution of the base frequencies.

[0056] It is understandable that once the cut-off frequency of the low-pass filter is determined, the time-domain impulse response function of the low-pass filter can be determined, thereby realizing the design of the low-pass filter.

[0057] Taking the nth low-pass filter of an arbitrary low-pass filter bank as an example, the expression of the time-domain impulse response of the low-pass filter is as follows: ; In the formula, f n c is the nth low-pass filter of an arbitrary low-pass filter bank, t is time, and h(t) is the response of this low-pass filter.

[0058] In one embodiment, each group of low-pass filter banks may only contain one low-pass filter.

[0059] In another embodiment, each group of low-pass filter banks may contain multiple low-pass filters, thereby realizing multi-scale adaptive filtering.

[0060] In one embodiment, the above-mentioned original time-domain impulse response can be directly used as the actual response of the low-pass filter.

[0061] In another embodiment, the original time-domain impulse response can also be further processed. To suppress spectral leakage, first use a Hamming window function Q(t)=0.54 - 0.46 cos(2πt / T') with a length of T' to window the original impulse response h(t) to obtain the windowed finite impulse response , and then, to ensure the symmetry of the time-domain response of the low-pass filter, generate a symmetric impulse response h s '(t) through mirror extension, and perform normalization to obtain the normalized response , and use this as the actual response of the low-pass filter.

[0062] In this embodiment, by windowing, symmetrizing, and normalizing the original time-domain impulse response, in this way, the fluctuation of the transition band of the frequency response curve can be smoothed, the time-domain shift of gait key features can be avoided, the balance of the multi-channel signal energy distribution can be improved, and the stability and consistency of feature extraction in the motion interference scenario can be enhanced.

[0063] The adaptive filtering module is used to perform convolution processing on the sequence X u using M groups of low-pass filters. The M groups of low-pass filters correspond one-to-one with M dimensions. Each group of low-pass filters processes the feature sequence of the corresponding dimension. The feature sequences of each dimension are respectively convolved by N low-pass filters in the corresponding low-pass filter bank to obtain N convolution sequences of this dimension. The N convolution sequences of the same dimension are first passed through a channel attention network and then fused to obtain the enhanced sequence of this dimension.

[0064] Such asFigure 4 The following is a schematic structural diagram of an adaptive filtering module in an embodiment of the present invention. The adaptive filtering module processes the feature sequences of each dimension in the multi-dimensional time series X respectively to achieve feature enhancement. u Specifically, the adaptive filtering module has M parallel adaptive filtering sub-modules. M groups of low-pass filters are distributed in the M sub-modules one by one. The M sub-modules correspond to the M dimensions in the multi-dimensional time series X one by one, that is, each sub-module is used to process the feature sequence of the corresponding dimension in the multi-dimensional time series X.

[0065] Specifically, the adaptive filtering module has M parallel adaptive filtering sub-modules. M groups of low-pass filters are distributed in the M sub-modules one by one. The M sub-modules correspond to the M dimensions in the multi-dimensional time series X one by one, that is, each sub-module is used to process the feature sequence of the corresponding dimension in the multi-dimensional time series X. u Specifically, the adaptive filtering module has M parallel adaptive filtering sub-modules. M groups of low-pass filters are distributed in the M sub-modules one by one. The M sub-modules correspond to the M dimensions in the multi-dimensional time series X one by one, that is, each sub-module is used to process the feature sequence of the corresponding dimension in the multi-dimensional time series X. u Specifically, the adaptive filtering module has M parallel adaptive filtering sub-modules. M groups of low-pass filters are distributed in the M sub-modules one by one. The M sub-modules correspond to the M dimensions in the multi-dimensional time series X one by one, that is, each sub-module is used to process the feature sequence of the corresponding dimension in the multi-dimensional time series X.

[0066] The process of feature enhancement of the feature sequence of each dimension through its corresponding sub-module includes: The feature sequence is first convolved through N low-pass filters in the sub-module to obtain N convolution sequences, and the N convolution sequences are X'∈R T╳N ; X' is enhanced through a channel attention network to obtain N intermediate sequences; The feature values at the same moment in the N intermediate sequences are superimposed to fuse the N intermediate sequences into one sequence to obtain an enhanced sequence.

[0067] Among them, the specific working process of the channel attention network is: First, global average pooling is performed on the sequence X' to compress the spatio-temporal dimensions to obtain a channel description vector z = GAP(X'), z∈R N , where GAP() is the global average pooling function; Subsequently, the pooling result z is processed using two fully connected layers to calculate the attention weight distribution of N channels (i.e., N convolution sequences) , W 3 , W 4 are the weight matrices of the two fully connected layers respectively, and GELU() and σ() are the activation functions of the two fully connected layers respectively; Finally, each convolution sequence in X' is first weighted with the attention weight of the corresponding channel and then fused with the convolution sequence to obtain N intermediate sequences.

[0068] After obtaining the N intermediate sequences, feature fusion is performed to obtain the enhanced sequence under the corresponding dimension.

[0069] Specifically, the calculation formula of the enhanced sequence under each dimension finally obtained can be expressed as: ; In the formula, X f,j is the enhanced sequence of the jth dimension, Xi ' is the convolution sequence of the i-th channel in the input of sequence X, and g i is the attention weight corresponding to the i-th channel in the attention weight distribution g.

[0070] Finally, directly combine the enhanced sequences of M dimensions, then the multi-dimensional enhanced time series X can be obtained f ∈R T╳M .

[0071] The recognition module is used to make the multi-dimensional time series X composed of the enhanced sequences of M dimensions z pass through a deep learning network to obtain the gait recognition result corresponding to the lower limb.

[0072] Specifically, the choice of the deep learning network is not limited and can be a transformer, RNN, CNN, or LSTM (Long Short-Term Memory).

[0073] In this embodiment, a bidirectional LSTM can be selected. By simultaneously capturing the forward and backward dependencies of the time series, the bidirectional LSTM can more comprehensively model the dynamic characteristics of the gait.

[0074] In one embodiment, the multi-dimensional enhanced time series X obtained by directly combining the enhanced sequences of M dimensions can be directly f used as the input X of the deep learning network z .

[0075] In another embodiment, as Figure 2 shown, the abnormal gait real-time detection system further includes an adaptive adjustment module; The adaptive adjustment module is used to map the feature Y to the feature h through a fully connected layer network 2 , and make the feature h 2 respectively perform linear mappings through two linear mapping networks to obtain the feature amplitude adjustment amount ρ ∈ R M and the baseline offset correction amount ε ∈ R M . Use the adjustment amount ρ to weight the sequence X f and then fuse it with the sequence X f and the baseline offset correction amount ε to obtain the sequence X Z ; where the sequence X f ∈R T╳M is composed of the enhanced sequences of M dimensions, T is the sequence length, and the i-th component in the adjustment amount ρ is used as the weighting weight of the i-th dimensional enhanced sequence in X f . When fusing, each eigenvalue of the i-th dimensional enhanced sequence in X f is summed with the i-th component in the correction amount ε.

[0076] Specifically, the feature Y is mapped to the feature h through a fully connected network 2 , and the semantic information of the medical assessment scale is mapped to the feature h 2 , providing an information basis for subsequent parameter separation. The corresponding process can be expressed as the following expression: h 2 = ReLU(W 5 Y + b 5 ); In the formula, W 5 and b 5 are the weight matrix and bias of this fully connected layer, and ReLU() is the activation function.

[0077] The feature h 2 is respectively independently linearly mapped through two linear mapping networks to obtain the feature amplitude adjustment amount ρ and the baseline offset correction amount ε. The calculation formula can be expressed as: ρ = W 6 h 2 + b 6 ; ε = W 7 h 2 + b 7 ; In the formula, W 6 and W 7 are respectively the weight matrices of the two linear mapping networks, and b 6 is b 7 respectively the biases of the two linear mapping networks.

[0078] The sequence X f is weighted by the adjustment amount ρ and then fused with the sequence X f and the baseline offset correction amount ε to obtain the sequence X Z . The calculation formula can be expressed as: X Z = X f + (ρ ⊙ X f + Ω); In the formula, ⊙ represents element-wise multiplication, and Ω is a matrix with the same dimension as X f formed by replicating ε T times.

[0079] In this embodiment, pathological features are introduced to adaptively adjust the enhanced features. In this way, the signal features can be adaptively optimized according to the degree of gait disorder of an individual, so as to more accurately capture the subtle changes of abnormal gaits and enhance the model's ability to distinguish complex gait patterns.

[0080] Among them, the gait recognition task can be selected according to requirements.

[0081] In one embodiment, gait recognition includes single-task recognition, and the single-task recognition includes: classification of the stance phase and the swing phase, or recognition of continuous gait phase parameters.

[0082] In another embodiment, gait recognition includes multi-task recognition, and the multi-task recognition includes: classification of the stance phase and the swing phase, and recognition of continuous gait phase parameters. At this time, after feature extraction and sequence modeling are completed, the network performs the classification task of the stance phase and the swing phase (output through the classification head) and the regression task of continuous gait phase parameters (output through the regression head) through two independent fully connected layers respectively. The output of each task is based on the last time step of the sequence to ensure that it can integrate the most representative gait features in the whole sequence.

[0083] Since the current mainstream gait detection methods mostly focus on single-task detection. For example, only detect continuous gait phases or only detect whether the current gait is in the stance phase or the swing phase. However, in practical applications, multi-information cooperation is required: continuous phases are used to calculate the assistance intensity, while discrete events (stance phase or swing phase) determine the trigger timing. Since the existing methods focus on single-task detection, it is difficult to meet the demand for precise assistance of lower limb exoskeletons in complex rehabilitation scenarios. In this embodiment, dual-task recognition is adopted, which can meet the demand for precise assistance of lower limb exoskeletons in complex rehabilitation scenarios.

[0084] In one embodiment, when the gait recognition result includes the recognition of continuous gait phase parameters, the deep learning network outputs the gait phase parameters in the form of a two-dimensional polar coordinate vector.

[0085] Specifically, the gait phase parameter θ is defined as follows: within the gait cycle from the start of a heel strike (HS) event to the next heel strike, the gait phase parameter θ linearly increases from 0 to 1, which is used to represent the relative position of the current moment in the gait cycle. However, as Figure 5 shown, this definition will cause the gait phase parameter θ to instantaneously jump from 1 to 0 at the HS event, forming a discontinuity, and this discontinuity will interfere with the training process of the system, affecting the convergence and stability of the system.

[0086] To solve this problem, in this embodiment, a polar coordinate-based gait phase representation method is adopted to transform the scalar-form gait phase parameter θ into a two-dimensional polar coordinate vector form (x 1 , x 2 ), and the expression is: ; After the transformation, the components x 1 and x 2 of the obtained two-dimensional polar coordinate vector (x 1 , x 2are all within the range of (-1, 1) and satisfy x 1 2 + x 2 2 = 1. This mapping method converts the linearly increasing phase parameter θ into a continuously varying periodic signal, eliminating the jump problem at the HS event. Thus, it is beneficial to system training and improves the accuracy of the system.

[0087] Embodiment 2 The present invention also provides a training method for the abnormal gait real-time detection system described in Embodiment 1, including: Obtaining a training data set, where each group of training data includes the pathological characteristics of the patient and the gait-related data of the patient; Adding gait labels to the gait-related data in each group of training data; Input the training data into the system; obtain gait-related data of the same lower limb of the patient through the data acquisition module in the system to construct multi-dimensional features, and the multi-dimensional features at multiple consecutive moments constitute a multi-dimensional time series Xe; obtain the patient's pathological features Y through the pathological data acquisition module in the system, and the pathological features are the score sets of different evaluation indexes of gait; calculate the ratio of the total score of feature Y to the highest total score of each evaluation index as the pathological weight λ through the hybrid normalization processing module in the system, and perform hybrid normalization processing on each dimension feature in sequence Xe to obtain sequence Xu. The hybrid normalization processing includes making the feature obey the standard normal distribution to obtain the first processing result, making the feature amplitude ratio normalized to obtain the second processing result, and weighted summing the two processing results. The weight of the first processing result is positively correlated with λ, and the weight of the first processing result is negatively correlated with λ; make feature Y pass through M fully connected layer networks in the low-pass filter bank design module in the system to be mapped into the dynamic frequency adjustment amounts of M groups of low-pass filters, and fuse the dynamic frequency adjustment amounts of each low-pass filter in the filter bank with its base frequency to obtain the cut-off frequencies of each low-pass filter in the corresponding filter bank, where M>1 is the feature dimension; perform convolution processing on sequence Xu using M groups of low-pass filters through the adaptive filtering module in the system. The M groups of low-pass filters correspond one-to-one with M dimensions, and each group of low-pass filters processes the feature sequence of the corresponding dimension. The feature sequences of each dimension are respectively subjected to convolution calculations by N low-pass filters in the corresponding low-pass filter bank to obtain N convolution sequences of this dimension. The N convolution sequences of the same dimension are first passed through the channel attention network and then fused to obtain the enhanced sequence of this dimension, where N≥1; make the multi-dimensional time series Xz composed of the enhanced sequences based on M dimensions pass through the deep learning network through the recognition module in the system to obtain the gait recognition result of the corresponding lower limb; calculate the loss between the gait recognition result and the gait label to train the system until the loss between the gait recognition result output by the system and the gait label converges.

[0088] When the gait recognition is the classification of the stance phase and the swing phase, the gait label is the classification label of the stance phase and the swing phase; When the gait recognition is the recognition of continuous gait phase parameters, the gait label is the continuous gait phase parameter label; When the gait recognition includes both the classification of the stance phase and the swing phase and the recognition of continuous gait phase parameters, the gait label includes both the classification label of the stance phase and the swing phase and the continuous gait phase parameter label.

[0089] In one embodiment, when the system outputs gait phase parameters in the form of two-dimensional polar coordinate vectors, the continuous gait phase parameter labels used for training are also in the form of two-dimensional polar coordinate vector labels.

[0090] In one embodiment, when gait recognition includes the classification of the stance phase and the swing phase, the gait labels include the classification labels of the stance phase and the swing phase. The process of adding the classification labels of the stance phase and the swing phase to the gait-related data of a certain patient includes: Obtain the sagittal plane angular velocity sequence collected from this patient and calculate its mean value ξ and maximum value ω max ; Generate an adaptive coefficient according to the pathological weight λ of this patient ; β min and β max are respectively set positive values, and their optimal values can be determined according to experiments. For example, β min = 0.01, β max = 0.3; the pathological weight λ is the ratio of the total score of the pathological characteristics of this patient to the highest total score of each evaluation index; Calculate the angular velocity detection threshold ; Combined with the sagittal plane angular velocity curve and the angular velocity detection threshold ω(λ), identify the heel strike moment HS and the toe off moment TO. Find the maximum value point exceeding the angular velocity detection threshold ω(λ) in the curve as the threshold point. Take the first minimum value point before the threshold point in the curve as the moment TO, and the first minimum value point after the threshold point as the moment HS. If the current gait-related data is in the time interval from the moment TO to the moment HS, it is marked as the stance phase. If the current gait-related data is in the time interval from the moment HS to the moment TO, it is marked as the swing phase.

[0091] Specifically, this method analyzes the sagittal plane angular velocity curve of the calf to find its periodic change characteristics. To identify whether the current gait-related data corresponds to the swing phase or the stance phase, two types of moments in the curve need to be identified, namely the heel strike moment HS and the toe off moment TO. In the time interval from the moment TO to its adjacent moment HS, the lower limb is in the stance phase. In the time interval from the moment HS to the adjacent moment TO, the lower limb is in the swing phase.

[0092] In this embodiment, considering that there are significant gait differences among different patients, by introducing the pathological characteristics of the patient to guide the determination of the angular velocity detection threshold, the threshold can be adaptively adjusted according to the patient's pathological condition. Thus, the moments of heel strike and toe off can be marked more accurately, and the stance phase and the swing phase can be accurately divided, and then a more accurate real-time abnormal gait detection system can be trained.

[0093] Embodiment 3 The present invention also provides a rehabilitation exoskeleton device, which includes a multi-node wearable data acquisition system, a driving device, and a controller. The controller includes the abnormal gait real-time detection system as described in Embodiment 1. Among them, the multi-node wearable data acquisition system is used to collect gait-related data of the patient and transmit it to the controller; the controller is used to input the gait-related data and the pathological data of the patient into the abnormal gait real-time detection system, and control the driving device to assist the corresponding lower limb movement after obtaining the current gait recognition result.

[0094] Specifically, the driving device may consist of two motors. Each motor is connected to an independent Bowden cable, and the two Bowden cables are respectively connected to the instep and the heel of the wearer. When receiving a control instruction, the motor rotates to drive the Bowden cable to contract or relax, so as to assist the wearer in performing dorsiflexion and plantar flexion, thereby assisting the lower limb movement.

[0095] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification. It should be noted that the "in one embodiment", "for example", "again, for example", etc. of the present invention are intended to illustrate the present invention, rather than to limit the present invention.

[0096] The above-described embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent application. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. A real-time abnormal gait detection system, characterized in that: include: The data acquisition module is used to acquire the gait-related data of the same lower limb of the patient to construct multidimensional features. The multidimensional features of multiple consecutive moments constitute a multidimensional time series X e ; A pathological data acquisition module, used to acquire a patient's pathological feature Y, where the pathological feature is a score set of different evaluation indicators of gait; The hybrid normalization processing module is used to calculate the ratio of the total score of feature Y to the highest total score of each evaluation index as the pathological weight λ. e Each dimension of the feature is mixed and standardized to obtain the sequence X u , the hybrid normalization processing includes making the feature obey the standard normal distribution to obtain a first processing result and standardizing the feature amplitude ratio to obtain a second processing result and weighting the two processings, the weight of the first processing result is positively correlated with λ, and the weight of the first processing result is negatively correlated with λ; The low-pass filter bank design module is used to map the feature Y into the dynamic frequency adjustment amount of M groups of low-pass filters through M fully connected layer networks, and fuse the dynamic frequency adjustment amount of each low-pass filter in the filter bank with its basic frequency to obtain the cutoff frequency of each low-pass filter in the corresponding filter bank. M>1 is the feature dimension; Adaptive filtering module, used to use M groups of low-pass filters to filter the sequence X u Convolution processing is performed, and M groups of low-pass filters correspond to M dimensions one by one. Each group of low-pass filters processes the feature sequence of the corresponding dimension. The feature sequence of each dimension is convolved by the N low-pass filters in the corresponding low-pass filter group to obtain N convolution sequences of the dimension. The N convolution sequences of the same dimension are first passed through the channel attention network and then fused to obtain the enhanced sequence of the dimension, N ≥ 1; The recognition module is used to make the multidimensional time series X composed of the enhanced sequence based on M dimensions z Through the deep learning network, the gait recognition results of the corresponding lower limbs are obtained.

2. The abnormal gait real-time detection system according to claim 1, characterized in that: The gait-related data include knee joint angle, ankle joint angle, and three-axis angular velocity and three-axis acceleration collected at specific positions, and the specific positions include the front of the thigh, the front of the calf, and the instep.

3. The abnormal gait real-time detection system according to claim 1, characterized in that: Different gait assessment indicators include take-off, foot lift height, step length, gait symmetry, step continuity, walking path, trunk stability and step width.

4. The abnormal gait real-time detection system according to claim 1, characterized in that: The gate mechanism is used to fuse the dynamic frequency adjustment of each low-pass filter in the filter group with its basic frequency to obtain the cutoff frequency of each low-pass filter in the corresponding filter group. The fusion formula is: ； In the formula, f n c is the cutoff frequency of the nth low-pass filter in any low-pass filter bank, f n a is the fundamental frequency of the nth low-pass filter, f n ' is the dynamic frequency adjustment of the nth low-pass filter, α n is a trainable scaling factor, f max 、f min are the maximum and minimum frequencies set respectively, and sigmoid() is the activation function.

5. The abnormal gait real-time detection system according to claim 1, characterized in that: The system further comprises: The adaptive adjustment module is used to map feature Y to feature h2 through the fully connected layer network, and to linearly map feature h2 through two linear mapping networks to obtain the M-dimensional feature amplitude adjustment ρ and the M-dimensional baseline offset correction ε. The adjustment ρ is used to adjust the sequence X f After weighting, it is then combined with the sequence X f , and the baseline offset correction ε are fused to obtain the sequence X Z ; where sequence X f is composed of M-dimensional enhanced sequences directly combined, and the i-th component in the adjustment amount ρ is used as X f The weighted weight of the i-th dimension enhanced sequence in , when fused X f Each eigenvalue of the i-th dimension enhanced sequence in is summed with the i-th component of the correction value ε.

6. The abnormal gait real-time detection system according to claim 1, characterized in that: The gait recognition includes single-task recognition or multi-task recognition; The single-task identification includes: classification of stance phase and swing phase, or identification of continuous gait phase parameters; The multi-task recognition includes: classification of the stance phase and the swing phase, and recognition of continuous gait phase parameters.

7. The abnormal gait real-time detection system according to claim 1, characterized in that: The gait recognition result includes the recognition of continuous gait phase parameters, and the deep learning network outputs the gait phase parameters in the form of a two-dimensional polar coordinate vector.

8. A training method for the abnormal gait real-time detection system according to any one of claims 1 to 7, characterized in that: include: Obtaining a training data set, each set of training data includes pathological characteristics of a patient and gait-related data of the patient; Adding a gait label to the gait-related data in each set of training data; The training data is input into the system; the gait-related data of the same lower limb of the patient is obtained through the data acquisition module in the system to construct a multidimensional feature, and the multidimensional features of multiple continuous moments constitute a multidimensional time series Xe; the pathological feature Y of the patient is obtained through the pathological data acquisition module in the system, and the pathological feature is a score set of different evaluation indicators of gait; the ratio of the total score of feature Y to the highest total score of each evaluation indicator is calculated as the pathological weight λ by the mixed standardization processing module in the system, and each dimensional feature in the sequence Xe is subjected to mixed standardization processing to obtain a sequence Xu, wherein the mixed standardization processing includes making the feature obey the standard normal distribution to obtain a first processing result and standardizing the feature amplitude ratio to obtain a second processing result and weighted summing the two processing, wherein the weight of the first processing result is positively correlated with λ, and the weight of the first processing result is negatively correlated with λ; the feature Y is mapped to M fully connected layer networks respectively through M fully connected layer networks through the low-pass filter group design module in the system. The dynamic frequency adjustment amount of each low-pass filter in the filter group is merged with its basic frequency to obtain the cutoff frequency of each low-pass filter in the corresponding filter group, where M>1 is the feature dimension; The sequence Xu is convolved using M groups of low-pass filters through the adaptive filtering module in the system, where the M groups of low-pass filters correspond to the M dimensions one by one, and each group of low-pass filters processes the feature sequence of the corresponding dimension. The feature sequence of each dimension is convolved with the N low-pass filters in the corresponding low-pass filter group to obtain N convolution sequences of the dimension. The N convolution sequences of the same dimension are first passed through the channel attention network and then fused to obtain an enhanced sequence of the dimension, where N≥1; The multidimensional time series Xz composed of the enhanced sequences of the M dimensions is passed through a deep learning network through the recognition module in the system to obtain the gait recognition result of the corresponding lower limbs; The loss between the gait recognition result and the gait label is calculated to train the system until the loss between the gait recognition result output by the system and the gait label converges.

9. The training method according to claim 8, characterized in that: The gait recognition result includes the classification results of the stance phase and the swing phase; the gait label includes the classification label of the stance phase and the swing phase; The process of adding classification labels of the stance phase and swing phase to a patient's gait-related data includes: Obtain the sagittal angular velocity sequence collected by the patient and calculate its mean ξ and maximum ω max ; Generate an adaptive coefficient based on the patient's pathological weight λ β min and β max are the set values ​​respectively; the pathological weight λ is the ratio of the total score of the patient's pathological characteristics to the highest total score of each evaluation index; Calculate angular velocity detection threshold ; Combine the sagittal plane angular velocity curve and the angular velocity detection threshold ω(λ) to identify the heel strike HS and toe lift-off TO, find the maximum point in the curve that exceeds the angular velocity detection threshold ω(λ) as the threshold point, take the first minimum point in front of the threshold point in the curve as the moment TO, and the first minimum point after the threshold point as the moment HS. If the current gait-related data is in the time interval from moment TO to moment HS, it is marked as the support phase. If the current gait-related data is in the time interval from moment HS to moment TO, it is marked as the swing phase.

10. A rehabilitation exoskeleton device, characterized in that: It comprises a multi-node wearable data acquisition system, a driving device and a controller, wherein the controller comprises the abnormal gait real-time detection system according to any one of claims 1 to 7; The multi-node wearable data acquisition system is used to collect gait-related data of the patient and transmit it to the controller; The controller is used to input the gait-related data and the patient's pathological data into the abnormal gait real-time detection system, and control the driving device to assist the corresponding lower limb movement after obtaining the current gait recognition result.

Images