A hip, knee and ankle rehabilitation exoskeleton smart machine for stroke patients

Abstract

The present invention provides a hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients, comprising a waist brace worn on the patient's waist, a knee brace tied to the patient's knee, an ankle brace sleeved on the patient's ankle, an electromyographic sensor for collecting electromyographic signals at the patient's calf muscles, and an angle sensor for collecting the patient's knee bending angle. The waist brace is movably connected to the knee brace through a first connecting piece, and the knee brace is movably connected to the ankle brace through a second connecting piece. A first electrical stimulation module is provided on the inner side of the waist brace, and a second electrical stimulation module is provided on the inner side of the knee brace. The electromyographic sensor is electrically connected to the first electrical stimulation module and the second electrical stimulation module through a signal transmission line. The first electrical stimulation module and the second electrical stimulation module are provided with an electromyographic signal collection unit, an electromyographic signal optimization unit and an electromyographic signal classification unit. The present invention can reduce the probability of accidentally touching an electrical stimulation switch, and help patients stabilize their upper body and improve their posture when walking.

Description

Technical Field

[0001] The present invention relates to the technical field of rehabilitation devices, and in particular to an intelligent exoskeleton machine for hip, knee and ankle rehabilitation for stroke patients. Background Art

[0002] Surface electromyography (sEMG) is a complex temporal and spatial synthesis of subcutaneous muscle electrical activity on the surface of the skin. Since EMG signals come from the electrical signals of the human body, they are direct and natural. Nowadays, EMG signals have become an important information source that can be used for research on muscle movement, muscle injury diagnosis, rehabilitation medicine, forensic identification, sports and other aspects. Usually, multi-channel EMG signals (EMG signals) measured from the skin on the surface of moving muscles are safe and non-invasive EMG signals that can be used for research on human movement and biomechanics.

[0003] In the prior art, when taking care of stroke patients, they are usually given rehabilitation braces. The rehabilitation braces will continuously stimulate the patient's knee joints through electrical stimulation when the patient walks, so that the patient's nerves are touched and the patient walks. This method has the following problems: 1. It is impossible to accurately obtain the patient's true intentions, and the electric stimulation switch is often accidentally touched, so that the patient is stimulated even when resting, resulting in a decreased sense of experience; 2. Since the patient usually has half of the body paralyzed, although only stimulating the patient's knee joint can enable the patient to walk, the patient's upper body is not stable enough and is not closely connected to the lower body, which will cause the patient's posture to deform when walking, which will affect the patient's body shape over time. Summary of the invention

[0004] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients, aiming to solve the technical problems mentioned in the background technology.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0006] A hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients, comprising a waist brace worn on the patient's waist, a knee brace tied to the patient's knee, an ankle brace sleeved on the patient's ankle, an electromyographic sensor for collecting electromyographic signals at the patient's calf muscles, and an angle sensor for collecting the patient's knee bending angle, wherein the waist brace is movably connected to the knee brace via a first connector, the knee brace is movably connected to the ankle brace via a second connector, a first electrical stimulation module is provided on the inner side of the waist brace, a second electrical stimulation module is provided on the inner side of the knee brace, the electromyographic sensor is electrically connected to the first electrical stimulation module and the second electrical stimulation module via a signal transmission line, and an electromyographic signal acquisition unit, an electromyographic signal optimization unit and an electromyographic signal classification unit are provided in the first electrical stimulation module and the second electrical stimulation module;

[0007] The electromyographic signal acquisition unit is used for:

[0008] Collecting initial electromyographic signals at the patient's calf by means of the electromyographic sensor;

[0009] Performing filtering processing on the initial electromyographic signal to eliminate high-frequency signal interference and power frequency signal interference, so as to obtain the electromyographic signal after the first screening;

[0010] Using an energy characteristic method to perform a purposeful screening and evaluation on the electromyographic signal after the first screening, to obtain an electromyographic signal after the second screening that reflects the patient's true purpose;

[0011] Through the generic channel decision method, the signal activity of all the electromyographic signals after the second screening at the same time is determined to obtain the electromyographic signals after the third screening, so as to improve the data information quality of the sliding window at the current time;

[0012] The electromyographic signal optimization unit is used for:

[0013] Vectorizing the electromyographic signal after the third screening to obtain a vectorized electromyographic signal;

[0014] According to the preset action scenario, weights are allocated to the vectorized electromyographic signals to meet the patient's actual action intention;

[0015] Performing position update iteration on the weighted vectorized electromyographic signal based on the mean rule to obtain the optimal solution vectorized electromyographic signal;

[0016] The optimal solution quantized electromyographic signal is locally checked to exclude the optimal solution quantized electromyographic signal that is trapped in a local optimal solution, so as to obtain a vectorized electromyographic signal for comparison.

[0017] According to one aspect of the above technical solution, the electromyographic signal classification unit is used for:

[0018] Inputting the comparison vectorized electromyographic signal into a classifier model;

[0019] The comparison vectorized electromyographic signal input into the classifier model is compared with the pre-trained real vectorized electromyographic signal to determine the patient's willingness based on the electromyographic signal at this moment, thereby driving the first electrical stimulation module and the second electrical stimulation module to send out corresponding stimulation signals.

[0020] According to one aspect of the above technical solution, the specific steps of using the energy characteristic method to perform purposeful screening and evaluation on the electromyographic signal after the first screening to obtain the electromyographic signal after the second screening that reflects the patient's true purpose include:

[0021] A sliding window is set, and the signal-to-noise ratio of the electromyographic signal after the first screening in the sliding window is improved by an energy operator to obtain an electromyographic signal after the signal-to-noise ratio is amplified. The formula is as follows:

[0022] ;

[0023] in, k represents the frame number of the electromyographic signal after the first screening, Indicates k The amplified signal-to-noise ratio of the frame is the electromyographic signal. x is the amplitude of each frame of the electromyographic signal after the first screening;

[0024] Obtain the interval energy of the amplified signal-to-noise ratio electromyographic signal in a period of time, and the absolute value mean of the first frame signal of the amplified signal-to-noise ratio electromyographic signal in the sliding window in this period, and set the logarithm with a base of ten, and optimize the amplified signal-to-noise ratio electromyographic signal through the logarithmic energy signal characterization formula, the formula is as follows:

[0025] ;

[0026] ;

[0027] ;

[0028] Among them, E k is the interval energy, c is the absolute value mean, SE is the logarithmic energy of one frame, and N is the length of the sliding window;

[0029] The logarithmic energy of each frame is plotted into a logarithmic energy feature diagram, and the logarithmic energy below a preset threshold is filtered out to obtain the electromyographic signal after the second screening.

[0030] According to one aspect of the above technical solution, the specific steps of determining the signal activity of all the electromyographic signals after the second screening at the same time by a generic channel decision method to obtain the electromyographic signals after the third screening to improve the data information quality of the sliding window at the current time include:

[0031] Based on the electromyographic signal after the second screening, obtaining an influencing parameter affecting the accuracy of the electromyographic signal after the second screening;

[0032] The influencing parameters include but are not limited to standard deviation, sample entropy and zero-crossing rate;

[0033] Marking the first frame of the electromyographic signal after the second screening within a moment as a silent segment, and constructing a correlation coefficient formula according to the influencing parameter to obtain the correlation coefficients of the remaining frames relative to the first frame;

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] in, discover represents the functional distance between two vectors, dr represents the correlation between two vectors, X and Y They represent two eigenvectors, f 1 、f 2 、 f 3 represent the characteristic function of standard deviation, the characteristic function of sample entropy and the characteristic function of zero-crossing rate respectively. k’ represents the frame number of the electromyographic signal after the second screening, x i , x j , x l , y i , y j , y l All represent points on each characteristic function;

[0042] The frames with correlation coefficients less than the preset correlation threshold are marked as active segments to obtain the electromyographic signals after the third screening.

[0043] According to one aspect of the above technical solution, the specific steps of weighting the vectorized electromyographic signal according to the preset action scenario to meet the patient's actual action intention include:

[0044] Based on the wavelet function, each vector in a set of the vectorized electromyographic signals is weighted according to a preset action scenario to calculate a weighted average value of the set of vectorized electromyographic signals;

[0045] ;

[0046] ;

[0047] Wherein, MU is the weighted average value, x n is the first one of the vectorized electromyographic signals n single vectors, P is the number of the single vectors, m is the expansion coefficient, is the mother wavelet, exp is the exponential function;

[0048] The weighted average value is taken into consideration in the vectorized electromyographic signal to obtain the vectorized electromyographic signal after weight distribution.

[0049] According to one aspect of the above technical solution, the specific steps of performing position update iteration on the weighted vectored electromyographic signal based on the mean rule to obtain the optimal solution of the vectored electromyographic signal include:

[0050] Based on the mean rule, the vectorized electromyographic signals after weight distribution are differentiated to obtain a first weighted average value of the vectorized electromyographic signals after differentiation under the two action scenarios. MH 1 and the second weighted average MH 2;

[0051] Among them, in one of the action scenarios:

[0052] ;

[0053] ;

[0054] Among them, in another action scenario:

[0055] ;

[0056] ;

[0057] in, q is the number of iterations, is the maximum number of iterations, 𝜀 is a constant, rand 1 and rand 2 is a normally distributed random value, f ( x ) is the fitness function, x a1 , x a2 and x a3 are the optimal solution, intermediate solution and worst solution under the action scenario, respectively. x bs , x bt and x ws are the optimal solution, the intermediate solution and the worst solution under another action scenario, respectively, and exp is an exponential function;

[0058] Using the information of the first weighted average value and the second weighted average value to move to the next solution of the vectorized electromyographic signal after differentiation, the diversity of the group of vector populations is enhanced, and then the optimal solution vectorized electromyographic signal is calculated;

[0059] ;

[0060] Here, r is a random number between 0 and 1.

[0061] According to one aspect of the above technical solution, the specific steps of locally checking the optimal solution quantized electromyographic signal and excluding the optimal solution quantized electromyographic signal that is trapped in the local optimal solution to obtain the vectorized electromyographic signal for comparison include:

[0062] According to a preset normal distribution range, searching for the optimal solution quantized electromyographic signal that meets the normal distribution range;

[0063] Optimizing the vectorized electromyographic signal after differentiation of each iteration by an optimization algorithm, and then optimizing the optimal solution vectorized electromyographic signal;

[0064] ;

[0065] in, is the optimized optimal solution to quantize the electromyographic signal, randq For the q During the iteration rand 1 and rand The mean between 2, φ is a random constant between (0, 1), For comprehensive , , The new solution vector of ;

[0066] All the optimized optimal solution vectorized electromyographic signals in the group are fused to obtain vectorized electromyographic signals for comparison.

[0067] According to one aspect of the above technical solution, the waist brace includes a waist sleeve, the two ends of the first connecting member are rotatably connected to the knee brace and the ankle brace respectively, the ankle brace includes a supporting member supporting the sole of the patient and a clamping member connected to the patient's shoe upper, the supporting member and the clamping member are integrally formed, the knee brace includes a knee sleeve, one end of the second connecting member is fixedly connected to the knee sleeve, and the other end of the second connecting member is rotatably connected to the ankle brace.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] By connecting the waist brace, knee brace and ankle brace respectively through the first connecting piece and the second connecting piece, and arranging the first electrical stimulation module and the second electrical stimulation module on the inner side of the waist brace and the knee brace respectively, the electromyographic signal of the patient's calf can be collected in real time through the electromyographic sensor, and in combination with the angle sensor, the useless signals are filtered out and the useful signals are retained in the electromyographic signal collection unit, the electromyographic signal optimization unit and the electromyographic signal classification unit, and the useful signals are analyzed to obtain the patient's true intention, and the patient's waist and knees are stimulated respectively through the first electrical stimulation module and the second electrical stimulation module, and the size of the stimulation can be changed according to the analysis result to help the patient achieve the corresponding action, thereby reducing the probability of accidentally touching the electrical stimulation switch and improving the patient's experience. At the same time, the introduction of the waist brace can help the patient stabilize the upper body and improve the posture when walking based on waist stimulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0071] Figure 1 It is a schematic structural diagram of a hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients in a first embodiment of the present invention at a first viewing angle;

[0072] Figure 2 It is a schematic diagram of the structure of the hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients in the first embodiment of the present invention at a second viewing angle;

[0073] Figure 3 It is a structural block diagram of the electromyographic signal acquisition unit, the electromyographic signal optimization unit and the electromyographic signal classification unit in the second embodiment of the present invention;

[0074] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0075] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0076] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0077] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0078] Embodiment 1

[0079] See also Figure 1 to Figure 2 , shown is a hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients in the first embodiment of the present invention, comprising a waist brace 10 worn on the patient's waist, a knee brace 20 tied to the patient's knee, an ankle brace 30 sleeved on the patient's ankle, an electromyographic sensor for collecting electromyographic signals at the patient's calf muscles, and an angle sensor for collecting the patient's knee bending angle. The waist brace 10 is movably connected to the knee brace 20 through a first connecting member 40, and the knee brace 20 is movably connected to the ankle brace 30 through a second connecting member 50. A first electrical stimulation module 11 is provided on the inner side of the waist brace 10, and a second electrical stimulation module 21 is provided on the inner side of the knee brace 20. The electromyographic sensor is electrically connected to the first electrical stimulation module 11 and the second electrical stimulation module 21 through a signal transmission line. The first electrical stimulation module 11 and the second electrical stimulation module 21 are provided with an electromyographic signal acquisition unit, an electromyographic signal optimization unit and an electromyographic signal classification unit.

[0080] It can be understood that the present invention connects the waist brace 10, the knee brace 20 and the ankle brace 30 respectively through the first connecting piece 40 and the second connecting piece 50, and respectively sets the first electrical stimulation module 11 and the second electrical stimulation module 21 on the inner side of the waist brace 10 and the knee brace 20. The electromyographic signal of the patient's calf can be collected in real time by the electromyographic sensor, and in combination with the angle sensor, the useless signals are filtered out and the useful signals are retained in the electromyographic signal collection unit, the electromyographic signal optimization unit and the electromyographic signal classification unit. The useful signals are analyzed to obtain the patient's true intention, and the patient's waist and knees are stimulated by the first electrical stimulation module 11 and the second electrical stimulation module 21 respectively. The size of the stimulation can be changed according to the analysis result to help the patient achieve the corresponding action, thereby reducing the probability of accidentally touching the electrical stimulation switch and improving the patient's experience. At the same time, the introduction of the waist brace 10 can help the patient stabilize the upper body and improve the posture when walking based on waist stimulation.

[0081] Furthermore, the waist brace 10 includes a waist sleeve, and both ends of the first connecting member 40 are rotatably connected to the knee brace 20 and the ankle brace 30 respectively. The ankle brace 30 includes a supporting member 31 for supporting the sole of the patient and a clamping member 32 connected to the patient's shoe upper, and the supporting member 31 and the clamping member 32 are integrally formed. The knee brace 20 includes a knee sleeve, and one end of the second connecting member 50 is fixedly connected to the knee sleeve, and the other end of the second connecting member 50 is rotatably connected to the ankle brace 30.

[0082] It can be understood that the sole of the patient's shoe is supported by the supporting member 31 and connected to the patient's shoe body by the clamping member 32, and then the knee sleeve and the ankle brace 30 are connected by the second connecting member 50, wherein the second connecting member 50 must be fixedly connected to the knee sleeve and rotatably connected to the ankle brace 30, so that when the patient stands on tiptoe while walking, the ankle can drive the knee to move through the second connecting member 50, which is more in line with the walking posture of the human body, and the waist brace 10 and the ankle brace 30 only need to have the first connecting member 40 hinged to the two.

[0083] Embodiment 2

[0084] Please refer to Figure 3 , shown is a hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients in the second embodiment of the present invention. This embodiment focuses on describing the processing process of the electromyographic signal acquisition unit 1, the electromyographic signal optimization unit 2 and the electromyographic signal classification unit 3 and their actual application in this machine.

[0085] The electromyographic signal acquisition unit 1 is used for:

[0086] Collecting initial electromyographic signals at the patient's calf by means of the electromyographic sensor;

[0087] Performing filtering processing on the initial electromyographic signal to eliminate high-frequency signal interference and power frequency signal interference, so as to obtain the electromyographic signal after the first screening;

[0088] It can be understood that high-frequency signals and industrial frequency signals are noise signals that are bound to be generated during the signal acquisition process, usually coming from external interference or emitted by machine components during the acquisition process. In this embodiment, a notch filter can be used for filtering, which is only the first screening;

[0089] Using an energy characteristic method to perform a purposeful screening and evaluation on the electromyographic signal after the first screening, to obtain an electromyographic signal after the second screening that reflects the patient's true purpose;

[0090] Through the generic channel decision method, the signal activity of all the electromyographic signals after the second screening at the same time is determined to obtain the electromyographic signals after the third screening, so as to improve the data information quality of the sliding window at the current moment.

[0091] Specifically, the specific steps of using the energy characteristic method to perform purposeful screening and evaluation on the first screened electromyographic signal to obtain the second screened electromyographic signal reflecting the patient's true purpose include:

[0092] A sliding window is set, and the signal-to-noise ratio of the electromyographic signal after the first screening in the sliding window is improved by an energy operator to obtain an electromyographic signal after the signal-to-noise ratio is amplified. The formula is as follows:

[0093] ;

[0094] in, k represents the frame number of the electromyographic signal after the first screening, Indicates k The amplified signal-to-noise ratio of the frame is the electromyographic signal. x is the amplitude of each frame of the electromyographic signal after the first screening; it can be understood that in addition to the high-frequency signal and the power frequency signal caused by external factors, there are also noise signals caused by continuous muscle movement, muscle shaking and static electricity caused by body sweat, so the signal-to-noise ratio of the electromyographic signal after the first screening can be amplified by this formula;

[0095] Obtain the interval energy of the amplified signal-to-noise ratio electromyographic signal in a period of time, and the absolute value mean of the first frame signal of the amplified signal-to-noise ratio electromyographic signal in the sliding window in this period, and set the logarithm with a base of ten, and optimize the amplified signal-to-noise ratio electromyographic signal through the logarithmic energy signal characterization formula, the formula is as follows:

[0096] ;

[0097] ;

[0098] ;

[0099] Among them, E k is the interval energy, c is the absolute value mean, SE is the logarithmic energy of one frame, and N is the length of the sliding window; it can be understood that the present invention represents the amplified signal-to-noise ratio electromyographic signal in the form of energy, and divides the amplified signal-to-noise ratio electromyographic signal into an interval energy E of a small segment. k , and then characterize the logarithmic energy of each time period by means of logarithmic energy;

[0100] The logarithmic energy of each frame is plotted into a logarithmic energy feature map, and the logarithmic energy below a preset threshold is filtered out, thereby obtaining an electromyographic signal after the second screening;

[0101] It is understandable that although the noise signals caused by continuous muscle movement, muscle shaking and static electricity from body sweat are loud, their energy is small. Therefore, these noises can be easily distinguished through the logarithmic energy method and then excluded to obtain the electromyographic signal after the second screening.

[0102] Furthermore, the specific steps of determining the signal activity of all the electromyographic signals after the second screening at the same time by the generic channel decision method to obtain the electromyographic signals after the third screening to improve the data information quality of the sliding window at the current time include:

[0103] Based on the electromyographic signal after the second screening, obtaining an influencing parameter affecting the accuracy of the electromyographic signal after the second screening;

[0104] Among them, the influencing parameters include but are not limited to standard deviation, sample entropy and zero-crossing rate; it is understandable that since the electromyographic signal has the characteristics of short-term, weak, random, low-frequency, etc., it is necessary to take these influencing parameters into account. It should be noted that the expressions of these influencing parameters are prior art and will not be repeated here;

[0105] Marking the first frame of the electromyographic signal after the second screening within a moment as a silent segment, and constructing a correlation coefficient formula according to the influencing parameter to obtain the correlation coefficients of the remaining frames relative to the first frame;

[0106] ;

[0107] ;

[0108] ;

[0109] ;

[0110] ;

[0111] ;

[0112] ;

[0113] in, discover represents the functional distance between two vectors, dr represents the correlation between two vectors, X and Y They represent two eigenvectors, f 1 、f 2 、 f 3 represent the characteristic function of standard deviation, the characteristic function of sample entropy and the characteristic function of zero-crossing rate respectively. k’ represents the frame number of the electromyographic signal after the second screening, x i , x j , x l , y i , y j , y l All represent points on each characteristic function;

[0114] The frames with correlation coefficients less than the preset correlation threshold are marked as active segments to obtain the electromyographic signals after the third screening.

[0115] It is understandable that we first establish the initial two eigenvectors X and Y , and then use the distance formula discover Calculate the distance between the eigenvectors and then get the correlation coefficient between the two eigenvectors dr , since the first frame is used as the reference point and the first frame is marked as a silent segment, the reason is that the patient's muscles will move inadvertently in life, but the patient has no intention of moving. Therefore, if the correlation coefficient between most of the subsequent frames and the first frame is higher, it means that the subsequent frames are very close to the silent segment, and it can be determined that the electromyographic signal does not reflect the patient's intention to act, and the first electrical stimulation module and the second electrical stimulation module do not send out electrical stimulation or send out weak electrical stimulation to better meet the patient's comfort needs; if the correlation coefficient between most of the subsequent frames and the first frame is low, the subsequent frames are marked as active segments to obtain the electromyographic signal after the third screening for the next step of analysis.

[0116] Furthermore, the electromyographic signal optimization unit 2 is used for:

[0117] Vectorizing the electromyographic signal after the third screening to obtain a vectorized electromyographic signal;

[0118] According to the preset action scenario, weights are allocated to the vectorized electromyographic signals to meet the patient's actual action intention;

[0119] Performing position update iteration on the weighted vectorized electromyographic signal based on the mean rule to obtain the optimal solution vectorized electromyographic signal;

[0120] The optimal solution quantized electromyographic signal is locally checked to exclude the optimal solution quantized electromyographic signal that is trapped in a local optimal solution, so as to obtain a vectorized electromyographic signal for comparison.

[0121] Specifically, the specific steps of weighting the vectorized electromyographic signals according to the preset action scenario to meet the patient's actual action intention include:

[0122] Based on the wavelet function, each vector in a set of the vectorized electromyographic signals is weighted according to a preset action scenario to calculate a weighted average value of the set of vectorized electromyographic signals;

[0123] ;

[0124] ;

[0125] Wherein, MU is the weighted average value, x n is the first one of the vectorized electromyographic signals n single vectors, P is the number of the single vectors, m is the expansion coefficient, is the mother wavelet, exp is the exponential function;

[0126] The weighted average value is taken into consideration in the vectorized electromyographic signal to obtain the vectorized electromyographic signal after weight distribution.

[0127] It is understandable that in the process of extracting the electromyographic signal, it is necessary not only to consider the patient's true intention and purpose, but also to consider that the purpose of the machine itself is to stimulate the patient's nerves to assist the patient in walking. Therefore, according to the position of the angle sensor, the patient's action scenario at this time (standing or sitting) can be analyzed, and corresponding weights can be given in different action scenarios. For example, when it is detected that the patient is sitting, in order to move the legs, the patient will often twist the ankle 360 ​​degrees. At this time, at the electromyographic signal level, the behavior of "standing up" can be given a lower weight to turn off or reduce the electrical stimulation, giving the patient a more comfortable experience. The intensity of the electrical stimulation will only be increased when the electromyographic signal is strong. For example, when the patient is in a standing posture, patients with severe stroke have poor control over their muscles, and they themselves cannot even clearly perceive the strength of their muscles. Therefore, a higher weight can be given to the behavior of "walking", giving priority to the behavior of "walking".

[0128] Furthermore, the specific steps of performing position update iteration on the weighted vectored electromyographic signal based on the mean rule to obtain the optimal solution of the vectored electromyographic signal include:

[0129] Based on the mean rule, the vectorized electromyographic signals after weight distribution are differentiated to obtain a first weighted average value of the vectorized electromyographic signals after differentiation under the two action scenarios. MH 1 and the second weighted average MH 2;

[0130] Among them, in one of the action scenarios:

[0131] ;

[0132] ;

[0133] Among them, in another action scenario:

[0134] ;

[0135] ;

[0136] in, q is the number of iterations, is the maximum number of iterations, 𝜀 is a constant, rand 1 and rand 2 is a normally distributed random value, f ( x ) is the fitness function, x a1 , x a2 and x a3are the optimal solution, intermediate solution and worst solution under the action scenario, respectively. x bs , x bt and x ws are the optimal solution, the intermediate solution and the worst solution under another action scenario, respectively, and exp is an exponential function;

[0137] Using the information of the first weighted average value and the second weighted average value to move to the next solution of the vectorized electromyographic signal after differentiation, the diversity of the group of vector populations is enhanced, and then the optimal solution vectorized electromyographic signal is calculated;

[0138] ;

[0139] Here, r is a random number between 0 and 1.

[0140] It is understandable that since the evaluation of vectorized electromyographic signals is more complicated, it requires stronger search capabilities and faster search speeds. Therefore, based on the mean rule, the vectorized electromyographic signals are continuously iterated to converge to obtain the optimal solution. The main idea is to obtain the first weighted average value under the two action scenarios by differentiating the vectorized electromyographic signals. MH 1st and 2nd weighted average MH 2. Then continuously iterate to find the optimal solution that reflects the patient’s true intention in the two action scenarios.

[0141] Furthermore, the specific steps of locally checking the optimal solution quantized electromyographic signal to exclude the optimal solution quantized electromyographic signal that is trapped in the local optimal solution to obtain the vectorized electromyographic signal for comparison include:

[0142] According to a preset normal distribution range, searching for the optimal solution quantized electromyographic signal that meets the normal distribution range;

[0143] Optimizing the vectorized electromyographic signal after differentiation of each iteration by an optimization algorithm, and then optimizing the optimal solution vectorized electromyographic signal;

[0144] ;

[0145] in, is the optimized optimal solution to quantize the electromyographic signal, randq For the q During the iteration rand 1 and rand The mean between 2, φ is a random constant between (0, 1), For comprehensive , , The new solution vector of ;

[0146] All the optimized optimal solution vectorized electromyographic signals in the group are fused to obtain vectorized electromyographic signals for comparison.

[0147] It can be understood that in order to prevent the vectorized electromyographic signal from falling into a local optimal solution during the iteration process, the vectorized electromyographic signal is improved by an optimization algorithm to obtain the optimal solution vectorized electromyographic signal, and all the optimized optimal solution vectorized electromyographic signals in the group are fused to obtain the vectorized electromyographic signal for comparison.

[0148] Furthermore, the electromyographic signal classification unit 3 is used for:

[0149] Inputting the comparison vectorized electromyographic signal into a classifier model;

[0150] The comparison vectorized electromyographic signal input into the classifier model is compared with the pre-trained real vectorized electromyographic signal to determine the patient's willingness based on the electromyographic signal at this moment, thereby driving the first electrical stimulation module and the second electrical stimulation module to send out corresponding stimulation signals.

[0151] It can be understood that the classifier model has pre-trained real vectorized EMG signals, and the real vectorized EMG signals have a variety of action scenarios. Based on the comparison of the vectorized EMG signals, the closest real vectorized EMG signals are found. In this way, the first electrical stimulation module and the second electrical stimulation module are controlled to make corresponding responses (the size of the released electrical stimulation intensity) according to the various action scenarios in the real vectorized EMG signals to assist the daily behavior of stroke patients.

[0152] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0153] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients, characterized in that: The invention comprises a waist brace worn on the patient's waist, a knee brace tied on the patient's knee, an ankle brace sleeved on the patient's ankle, an electromyographic sensor for collecting electromyographic signals at the patient's calf muscles, and an angle sensor for collecting the patient's knee bending angle, wherein the waist brace is movably connected to the knee brace through a first connecting piece, and the knee brace is movably connected to the ankle brace through a second connecting piece, a first electrical stimulation module is provided on the inner side of the waist brace, a second electrical stimulation module is provided on the inner side of the knee brace, the electromyographic sensor is electrically connected to the first electrical stimulation module and the second electrical stimulation module through a signal transmission line, and an electromyographic signal acquisition unit, an electromyographic signal optimization unit, and an electromyographic signal classification unit are provided in the first electrical stimulation module and the second electrical stimulation module; The electromyographic signal acquisition unit is used for: Collecting initial electromyographic signals at the patient's calf by means of the electromyographic sensor; Performing filtering processing on the initial electromyographic signal to eliminate high-frequency signal interference and power frequency signal interference, so as to obtain the electromyographic signal after the first screening; Using an energy characteristic method to perform a purposeful screening and evaluation on the electromyographic signal after the first screening, to obtain an electromyographic signal after the second screening that reflects the patient's true purpose; Through the generic channel decision method, the signal activity of all the electromyographic signals after the second screening at the same time is determined to obtain the electromyographic signals after the third screening, so as to improve the data information quality of the sliding window at the current time; The specific steps of using the energy characteristic method to purposefully screen and evaluate the electromyographic signal after the first screening to obtain the electromyographic signal after the second screening that reflects the patient's true purpose include: A sliding window is set, and the signal-to-noise ratio of the electromyographic signal after the first screening in the sliding window is improved by an energy operator to obtain an electromyographic signal after the signal-to-noise ratio is amplified. The formula is as follows: ； in, k represents the frame number of the electromyographic signal after the first screening, Indicates k The amplified signal-to-noise ratio of the frame is the electromyographic signal. x is the amplitude of each frame of the electromyographic signal after the first screening; Obtain the interval energy of the amplified signal-to-noise ratio electromyographic signal in a period of time, and the absolute value mean of the first frame signal of the amplified signal-to-noise ratio electromyographic signal in the sliding window in this period, and set the logarithm with a base of ten, and optimize the amplified signal-to-noise ratio electromyographic signal through the logarithmic energy signal characterization formula, the formula is as follows: ； ； Among them, E k is the interval energy, c is the absolute value mean, SE is the logarithmic energy of a frame, and N is the length of the sliding window; The logarithmic energy of each frame is plotted into a logarithmic energy feature map, and the logarithmic energy below a preset threshold is filtered out, thereby obtaining an electromyographic signal after the second screening; The specific steps of determining the signal activity of all the electromyographic signals after the second screening at the same time by the generic channel decision method to obtain the electromyographic signals after the third screening to improve the data information quality of the sliding window at the current time include: Based on the electromyographic signal after the second screening, obtaining an influencing parameter affecting the accuracy of the electromyographic signal after the second screening; The influencing parameters include but are not limited to standard deviation, sample entropy and zero-crossing rate; Marking the first frame of the electromyographic signal after the second screening within a moment as a silent segment, and constructing a correlation coefficient formula according to the influencing parameter to obtain the correlation coefficients of the remaining frames relative to the first frame; ； ； ； ； ； ； ； in, discover represents the functional distance between two vectors, dr represents the correlation between two vectors, X and Y They represent two eigenvectors, f 1 、f 2 、 f 3 represent the characteristic function of standard deviation, the characteristic function of sample entropy and the characteristic function of zero-crossing rate respectively. k’ represents the frame number of the electromyographic signal after the second screening, x i , x j , x l , y i , y j , y l All represent points on each characteristic function; Mark the frames with correlation coefficients less than a preset correlation threshold as active segments to obtain the electromyographic signals after the third screening; The electromyographic signal optimization unit is used for: Vectorizing the electromyographic signal after the third screening to obtain a vectorized electromyographic signal; According to the preset action scenario, weights are allocated to the vectorized electromyographic signals to meet the patient's actual action intention; Performing position update iteration on the weighted vectorized electromyographic signal based on the mean rule to obtain the optimal solution vectorized electromyographic signal; The optimal solution vectorized electromyographic signal is locally checked to exclude the optimal solution vectorized electromyographic signal that is trapped in a local optimal solution, so as to obtain a vectorized electromyographic signal for comparison.

2. The hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients according to claim 1, characterized in that: The electromyographic signal classification unit is used for: Inputting the comparison vectorized electromyographic signal into a classifier model; The comparison vectorized electromyographic signal input into the classifier model is compared with the pre-trained real vectorized electromyographic signal to determine the patient's willingness based on the electromyographic signal at this moment, thereby driving the first electrical stimulation module and the second electrical stimulation module to send out corresponding stimulation signals.

3. The hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients according to claim 2, characterized in that: The specific steps of weighting the vectorized electromyographic signals according to the preset action scenario to meet the patient's actual action intention include: Based on the wavelet function, each vector in a set of the vectorized electromyographic signals is weighted according to a preset action scenario to calculate a weighted average value of the set of vectorized electromyographic signals; ； ； Wherein, MU is the weighted average value, x n is the first one of the vectorized electromyographic signals n single vectors, P is the number of the single vectors, m is the expansion coefficient, is the mother wavelet, exp is the exponential function; The weighted average value is taken into account in the vectorized electromyographic signal to obtain the vectorized electromyographic signal after weight distribution.

4. The hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients according to claim 3, characterized in that: The specific steps of performing position update iteration on the weighted vectored electromyographic signal based on the mean rule to obtain the optimal solution of the vectored electromyographic signal include: Based on the mean rule, the vectorized electromyographic signals after weight distribution are differentiated to obtain a first weighted average value of the vectorized electromyographic signals after differentiation under the two action scenarios. MH 1 and the second weighted average MH 2; Among them, in one of the action scenarios: ； ； Among them, in another action scenario: ； ； in, q is the number of iterations, is the maximum number of iterations, 𝜀 is a constant, rand 1 and rand 2 is a normally distributed random value, f ( x ) is the fitness function, x a1 , x a2 and x a3 are the optimal solution, intermediate solution and worst solution under the action scenario, respectively. x bs , x bt and x ws are the optimal solution, the intermediate solution and the worst solution under another action scenario, respectively, and exp is an exponential function; Using the information of the first weighted average value and the second weighted average value to move to the next solution of the vectorized electromyographic signal after differentiation, the diversity of the group of vector populations is enhanced, and then the optimal solution vectorized electromyographic signal is calculated; ； Here, r is a random number between 0 and 1.

5. The hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients according to claim 4, characterized in that: The specific steps of locally checking the optimal solution vectorized electromyographic signal to exclude the optimal solution vectorized electromyographic signal that is trapped in the local optimal solution to obtain the vectorized electromyographic signal for comparison include: According to a preset normal distribution range, searching for the optimal solution quantized myoelectric signal that meets the normal distribution range; Optimizing the vectorized electromyographic signal after differentiation of each iteration by an optimization algorithm, and then optimizing the optimal solution vectorized electromyographic signal; ； in, is the optimized optimal solution to quantize the electromyographic signal, randq For the q During the iteration rand 1 and rand The mean between 2, φ is a random constant between (0, 1), For comprehensive , , The new solution vector of ; All the optimized optimal solution vectorized electromyographic signals in the group are fused to obtain vectorized electromyographic signals for comparison.

6. The hip, knee and ankle rehabilitation exoskeleton intelligent machine for stroke patients according to claim 1, characterized in that: The waist brace includes a waist sleeve, two ends of the first connecting member are rotatably connected to the knee brace and the ankle brace respectively, the ankle brace includes a supporting member supporting the sole of the patient's shoe and a clamping member connected to the patient's shoe upper, the supporting member and the clamping member are integrally formed, the knee brace includes a knee sleeve, one end of the second connecting member is fixedly connected to the knee sleeve, and the other end of the second connecting member is rotatably connected to the ankle brace.

Images