Exoskeleton wearable fatigue evaluation system and method based on surface electromyography and gait

Abstract

The present invention belongs to the field of wearable fatigue evaluation technology, specifically, an exoskeleton wearable fatigue evaluation system and method based on surface electromyography and gait, and an exoskeleton comprehensive performance evaluation method combining muscle fatigue threshold, biomechanical analysis, and subjective fatigue self-perception scale. Different from the traditional single measurement method of surface electromyography signal or blood oxygen saturation, this method will effectively improve the accuracy of passive load lower limb exoskeleton performance evaluation; through motion capture, the subject's gait is compared and analyzed to obtain spatial position information and muscle force status, and the stability of the lower limb knee joint is calculated. At the same time, the subject's sEMG is collected, pre-processed and the muscle fatigue threshold is calculated. Finally, the EMG is analyzed by combining the sRPE score and the knee joint offset variance. FT The results showed that the evaluation method is effective for the evaluation of passive lower limb exoskeleton. FT The average arrival time was delayed by 42.9%, lower limb stability was improved by 75.8%, and subjective fatigue was alleviated by 30.3%.

Description

Technical Field

[0001] The present invention relates to the technical field of wearable fatigue assessment, and in particular to an exoskeleton wearable fatigue assessment system and method based on surface electromyography and gait. Background Art

[0002] Passive lower limb exoskeletons are currently primarily used for assisted training and walking. Lower limb movement is a complex, high-degree-of-freedom movement performed collaboratively by skeletal muscles, bones, and joints, and subjective and objective fatigue perceptions interact during lower limb movement [1]. Currently, most fatigue assessments during exoskeleton use are analyzed using only a single-dimensional evaluation metric, resulting in biased and inaccurate evaluation results.

[0003] Currently, sEMG and EMG FT It is widely used in traditional motion research. Due to the complex motion conditions during gait, the signal-to-noise ratio of biological signals will be reduced. If sEMG signals are used alone as the evaluation standard for exoskeleton, the effectiveness of the results will be reduced. Therefore, additional data is needed for comparison and verification to improve the accuracy of the evaluation.

[0004] Therefore, to address the above problems, an exoskeleton wearable fatigue assessment system and method based on surface electromyography and gait is proposed. Summary of the Invention

[0005] The present invention combines motion capture technology, EMG FT And the sRPE scale score is used to comprehensively evaluate the fatigue level of people when using the lower limb free exoskeleton.

[0006] The present invention is specifically achieved through the following technical solutions:

[0007] Exoskeleton wearable fatigue evaluation system based on surface electromyography and gait, including test method unit, fatigue feature calculation and extraction unit and test result analysis unit;

[0008] The experimental method unit includes experimental content and objectives, subjects and environment, and experimental design; the experimental content and objectives include experimental equipment and sRPE subjective scale selection; the collected data of the experimental design includes lower limb position information, sEMG raw signals, and sRPE subjective scale scores;

[0009] The fatigue feature calculation and extraction unit includes EMG FT Algorithms, lower limb stability analysis methods, and subjective scoring calculation methods;

[0010] The test result analysis unit includes EMG FT Calculation results, knee joint stability analysis, sRPE control feedback and characteristic value comparison.

[0011] Preferably, the EMG FT The specific steps of the algorithm are as follows:

[0012] In the first step, the sEMG raw data of groups A and B collected in the experiment were preprocessed using sEMG signals and processing methods to establish sEMG datasets.

[0013] In the second step, the preprocessed sEMG dataset is windowed and the RMS dataset is obtained, as shown in formula (1):

[0014]

[0015] In formula (1), N t is the number of sEMG signal data in a single tim, i is the data sequence number, E i is the i-th data sequence number in the sEMG dataset;

[0016] The third step is to divide the obtained RMS data set into two parts, M and N. Since the amount of data in this experiment is large, 15 RMS data points are taken as a group, recorded as M1, and the remaining data points are recorded as N1. M1 and N1 are used as fitting group 1, and 15+1 RMS data points are taken as M2, and the remaining data are taken as N2. M2 and N2 are used as fitting group 2, and so on. When N n When the number of RMS data points of M is equal to the RMS data points of M1, n , N n As the fitting group n, the first-order least squares fitting is performed on the n fitting groups to obtain two fitting straight lines and use the formula kM n ·kN n Calculate the product of the slopes of each set of fitted lines.

[0017] The fourth step is to take the time corresponding to the intersection of the fitted straight line in the set of data with the largest slope product calculated in the third step as the EMG FT .

[0018] Preferably, the lower limb stability analysis method is:

[0019] Formula (2) was used to calculate the variance of the knee joint offset distance for the data collected in experiment 1.0.

[0020]

[0021] In the formula is the average value of the knee joint deviation angle obtained in the experiment, X i is the i-th offset data; finally, the least square method is used to fit the variance to obtain the slope of each set of fitting lines, which is used as the characteristic value of knee joint stability. The size of this characteristic value reflects the stability of the knee joint.

[0022] Preferably, the subjective scoring calculation method divides the fatigue level into ten levels. Considering that all subjects are weight-bearing and the training intensity is moderate, in order to improve the accuracy, strenuous and very strenuous are expressed in two levels to improve the accuracy of the state description.

[0023] Preferably, according to EMG FT The preprocessing method described in the algorithm performs low-pass filtering and notch filtering on the original sEMG signal, and randomly selects a subject's original sEMG signal preprocessing result to display; through EMG FT The algorithm performs EMG on the pre-processed sEMG signal FT Calculate and randomly select a subject to display the experimental results.

[0024] Preferably, the knee joint stability analysis performs stability calculation on the knee joint. Since the experiment time is long and the amount of data obtained is large, the two groups of data A and B are windowed and their variance is calculated, assuming tim to be 2.5s and mov to be 5s; the variance of the subject's knee joint offset is obtained, and the variance is fitted with a least squares curve, and the slope K of the straight line is calculated at the same time, and the experimental results of a subject are randomly selected to display.

[0025] Preferably, the sRPE control feedback is collected by performing subjective fatigue scores on Groups A and B according to the CR-10 scale, and feature values ​​are extracted for each group, and a group of subject data is randomly selected for display.

[0026] Preferably, the characteristic value comparison summarizes the three groups of characteristic values ​​and performs group comparison, and the auxiliary lifting effect of the exoskeleton on the subject under the load state is demonstrated.

[0027] The present invention is beneficial in that:

[0028] 1. In a load comparison experiment of passive lower limb exoskeleton on 18 subjects, the sRPE score and knee joint stability were used to evaluate EMG. FT Subjective and objective verification was carried out and the three groups of data showed the same trend under different grouping conditions. Their eigenvalues ​​were strongly correlated and could be used as evaluation criteria.

[0029] 2. Feature extraction from each set of experimental data revealed that the passive lower-limb exoskeleton used in this experiment, when used expertly, can effectively reduce the muscle load incurred by the human body during walking under load. However, according to subject feedback, while the backboard and joints alleviate the user's overall walking pressure, significant localized pressure on the shoulders can affect subjective fatigue scores. Therefore, optimizing the human-machine relationship may improve subjective fatigue caused by localized pressure.

[0030] 3. In the experiment, both groups A(3) and B(3) showed higher lower limb stability and later EMG arrival. FT The characteristics of the exoskeleton can be judged as the subjects entering the fatigue state later, so better physical fitness will effectively improve the effectiveness of the exoskeleton. At the same time, all subjects mentioned that the exoskeleton's back weight plate affects their flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 It is a flowchart of the experimental process of the present invention;

[0033] Figure 2 Schematic diagram of gait cycle division of the present invention;

[0034] Figure 3 is a sensor distribution diagram of the present invention;

[0035] Figure 4 This is a schematic diagram of the experimental state and environment of the subject of the present invention;

[0036] Figure 5 EMG of the present invention FT Calculation principle diagram;

[0037] Figure 6 is a schematic diagram of the knee joint abduction angle of the present invention;

[0038] Figure 7 The time domain and frequency domain diagrams of the sEMG signal preprocessing process of the present invention;

[0039] Figure 8 EMG of the present invention FT Schematic diagram of variance calculation results;

[0040] Figure 9 Schematic diagram of the knee joint offset variance fitting line and the slope of the fitting line according to the present invention. DETAILED DESCRIPTION

[0041] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0042] Example 1: Experimental method

[0043] In order to comprehensively evaluate and analyze the fatigue state during the use of lower limb exoskeleton from subjective and objective perspectives and realize effective fatigue monitoring during the use of exoskeleton, this paper designed a gait analysis comparative experiment. By collecting the lower limb position information, sEMG signals and sRPE scores of 18 subjects during the experiment, and in order to explore the influence of exoskeleton use proficiency and wearer strength on the effect of lower limb exoskeleton use, this paper added a skilled control group and a strong control group in the comparative experiment. The experimental methods are as follows: Figure 1 shown.

[0044] 1.1 Experimental content and objectives

[0045] This experiment verifies the effectiveness of the method and evaluates the experimental exoskeleton by comparing the data differences of different feature groups when wearing and not wearing exoskeleton. The subjects are grouped according to their status to determine the data trend and analyze the impact of objective factors such as familiarity and strength on the effectiveness of exoskeleton use. Before the experiment, gait analysis is performed through motion capture to find the relevant force-generating muscles and select the sEMG collection location. In the experiment, the sEMG data, motion capture data, and sRPE scores of the subjects will be collected simultaneously, which will be used to analyze the muscle fatigue of the subjects, the stability of the lower limbs, and the comprehensive evaluation of subjective fatigue. Finally, the experimental results are obtained by comparing the feature groups. The goals of this experiment are: ① Verify EMG FT The accuracy of the algorithm for fatigue analysis of passive lower limb exoskeleton. ② Observe the impact of the subjects' physical strength and proficiency in using the exoskeleton on its effectiveness.

[0046] 1.1.1 Experimental equipment

[0047] The experiment used a Trigno fully wireless physiological sensor, which synchronously collects sEMG and inertial measurement unit (IMU) signals. The sEMG sampling frequency was 2000Hz, and the IMU signal sampling frequency was 75Hz, with angle acquisition accuracy to 5 decimal places. The exoskeleton used was a passive power-assisted exoskeleton. The treadmill used in the experiment was a KPOWER K160A, with each speed setting at 1 km / h.

[0048] 1.1.2 Selection of subjective sRPE scale

[0049] To improve the reliability of the experiment, the CR-10 scale, a 10-level scale based on subjective fatigue estimation, was selected. Unlike the Borg 6-20 scale, which roughly corresponds to heart rate (60-200 bpm), the CR-10 scale is more reliable and effective. Its correlation with objective indicators has been validated by extensive data, demonstrating its stability and broad applicability. This method has been proven to quantify subjective fatigue in athletes or subjects of various ages and skill levels, and the subjects in this experiment fully met the evaluation requirements of this scale.

[0050] According to the Training Impulse (TRIMP) concept, a subject's internal load can be calculated as "internal load = perceived exertion score x exercise duration." Perceived exertion scores are determined by the body's internal perception of the load during exercise, as well as a combination of psychological factors. Therefore, systematic training of subjects can reduce scoring errors and improve data accuracy. This experiment involved subjects training on a treadmill with varying loads to provide a preliminary understanding of the subjective impact of different scoring levels.

[0051] 1.2 Test subjects and environment

[0052] Eighteen healthy male subjects with no unhealthy habits were selected for this study. They were aged (25 ± 2) years, had a height of (168.0 ± 3.0) cm, and weighed (62.0 ± 5.0) kg. All subjects were in good health and had no muscle injuries or related diseases. They had not participated in strenuous exercise in the past week. Before the formal experiment, the subjects were informed of the experimental content and instructed to learn the sRPE subjective scale to assess exercise intensity. All subjects participated in this experiment voluntarily and were familiar with the experimental protocol, procedures, and precautions before the experiment began.

[0053] To compare fatigue levels between wearing and not wearing the exoskeleton, 18 subjects underwent two gait analysis experiments, Group A and Group B. Group A involved wearing the exoskeleton, while Group B did not. The two experiments were repeated 48 hours apart. To enhance experimental reliability, Groups A and B were repeated three times, with a 72-hour interval between each experiment. To investigate the impact of exoskeleton proficiency and wearer strength on the effectiveness of lower-limb exoskeleton use, subjects were categorized based on their weekly physical training hours and total exoskeleton use time. The 18 subjects were divided into three groups, each consisting of six participants: a normal group (less than 3 hours of weekly physical training and 0 hours of exoskeleton use); a skilled group (less than 3 hours of weekly physical training and 20 hours of exoskeleton use); and a strong group (more than 7 hours of weekly exercise and 0 hours of exoskeleton use). The experimental load consisted of a 20kg standard weight sandbag placed in a backpack. All subjects also underwent a quantified subjective fatigue training session lasting more than 4 hours. The subject groups are shown in Table 1.

[0054] Table 1 Grouping of subjects

[0055]

[0056] 1.3 Experimental Design

[0057] In order to understand the force exerted by the lower limb muscles during gait, the gait was divided into cycles through motion capture before the experiment. Taking the right unilateral leg of the human body as an example for gait analysis, when a complete gait cycle is performed, there are two obvious characteristics: the support phase and the swing phase. The support phase uses the target leg as the main supporting leg, and the target leg is not the main force-bearing leg in the swing phase. The support phase accounts for 60% of the entire gait cycle, and the swing phase accounts for 40%. In a complete cycle, the gait is divided into eight main characteristics for data collection nodes: first touchdown period-weight-bearing reaction period-mid-support phase-end of support phase-early swing phase-early swing phase-mid-swing phase-end of swing phase, such as Figure 2 . According to the results of gait analysis, the muscle force of the lower limbs during walking can be divided into the ground contact period: thigh muscles + buttocks muscles; load-bearing reaction period to the middle of the support phase: calf muscles + part of the thigh muscles; the end of the support phase to the end of the swing phase: thigh muscles + buttocks muscles. Therefore, the vastus medialis and gastrocnemius muscles are selected for data collection, as these parts have less friction with clothing and are the main force-generating muscles. According to the force analysis of the lower limb gait, this experiment uses 14 sensors for data collection, of which sensors 1 to 9 and 14 enable the IMU channel to collect motion capture position information, and sensors 10 to 13 collect sEMG raw signals. The corresponding parts of sensors 10 to 13 are: No. 10 left vastus medialis muscle, No. 11 left gastrocnemius muscle; No. 12 right vastus medialis muscle, No. 13 right gastrocnemius muscle. Wearing position is as follows Figure 3 .

[0058] (1) Group A subjects

[0059] The subjects in this group wore the exoskeleton and carried a 20kg backpack on the backrest. To prevent the IMU sensors from slipping due to wearing the exoskeleton, the lateral thigh sensors 03 and 04 and the calf sensors 02 and 05 were reinforced with electrical tape, and the remaining IMU and sEMG sensors were fixed with double-sided stickers. Figure 4 As shown in (a).

[0060] (2) Group B subjects

[0061] The subjects in this group were not wearing exoskeletons and were carrying a 20kg backpack. Therefore, the sensors were fixed in place using double-sided stickers. Figure 4 (b) shown.

[0062] Before the experiment, to ensure that the subjects were in a non-fatigued state, the interval between the experiments of Groups A and B was set to 48 hours. During the rest period, the subjects were not allowed to engage in strenuous exercise. Before the experiment, in order to reduce the error caused by signal drift and improve the accuracy of sEMG data, the surface hair of the subjects' rectus femoris and gastrocnemius muscles was shaved and the surface skin was cleaned with 75% alcohol. In order to avoid accidents caused by the subjects' first operation and weight-bearing discomfort, all test subjects were required to try on the experimental equipment and test the weight within 1 minute before the start. After the preparation work was completed, according to Figure 4 Eighteen subjects were assigned sensors and performed a 6-minute gait training experiment. Each participant completed the 6-minute gait training experiment, and each of the 18 subjects completed the experiment in sequence, which was recorded as one experiment. Groups A and B each underwent three experiments, with a 72-hour interval between each experiment. Three types of raw data were collected simultaneously: lower limb position information, raw sEMG signals, and subjective sRPE scores. The data were categorized and organized according to the experimental group, and significantly abnormal data were removed.

[0063] Example 2: Fatigue feature calculation and extraction

[0064] In order to conduct a subjective and objective comprehensive analysis of the fatigue state of the subjects during the gait training of the lower limb wearing exoskeleton, this paper uses EMG to analyze the fatigue state of the subjects. FT The algorithm, lower limb stability analysis method and sRPE score calculation method are used to calculate and analyze the lower limb position information obtained from the experiment, the original sEMG signal, and the original data of the sRPE subjective scale score to realize the exoskeleton wear fatigue evaluation based on surface electromyography and gait.

[0065] 2.1 EMG FT algorithm

[0066] sEMG is a technique that captures biological signals generated by muscles during biological activities through electrodes, amplifies and records the captured signals to obtain an orderly one-dimensional time series signal. sEMG signal acquisition has the characteristics of low risk, multi-target and non-invasive. FT It is a method to determine the muscle fatigue threshold by calculating the sEMG signal. Compared with other traditional sEMG signal fatigue feature analysis methods, EMG FT The time point when the muscle enters the anaerobic threshold is obtained by windowing the root mean square value of the sEMG signal, which can better reflect the muscle state and characteristic state of the monitored part. FT The fatigue threshold of the vastus medialis muscle of the subjects in the two experimental groups A and B was calculated to obtain a quantitative basis for the fatigue degree of the subjects.

[0067] 2.1.1 EMG FT calculate

[0068] During EMG FT The original sEMG needs to be preprocessed before calculation. Since the human sEMG is concentrated between 0 and 500 Hz, a low-pass filter is used to remove the noise generated by sensor touch and friction during the experiment; and then a 49.5-50.0 Hz notch filter is used to remove the power frequency noise. Since the sampling frequency is 2000 Hz and the amount of sample data is large, in order to effectively calculate the root mean square (RMS) value, the preprocessed sEMG data is windowed. Since the experimental duration is fixed and the sample size is large, a fixed-length moving window (mov) and time window (tim) are set to improve the accuracy of the results. According to the time length, this article sets mov to 2s and tim to 1s. A 6-minute experiment will obtain 360 data samples. In order to quantitatively analyze the degree of muscle fatigue, the preprocessed sEMG signal is subjected to EMG. FT The specific calculation steps are as follows: First, using the sEMG signal and processing methods in 2.1.1, preprocess the two groups of sEMG raw data collected in the experiment, A and B, and establish an sEMG dataset. Second, perform windowing calculation on the preprocessed sEMG dataset to obtain the RMS dataset, as shown in Equation (1).

[0069]

[0070] In formula (1), Nt is the number of sEMG signal data within a single tim, i is the data sequence number, and Ei is the i-th data sequence number in the sEMG data set. The third step is to divide the obtained RMS data set into two parts, M and N. Since the amount of data in this experiment is large, 15 RMS data points are taken as a group, recorded as M1, and the remaining data points are recorded as N1. M1 and N1 are used as fitting group 1. Then 15+1 RMS data points are taken as M2, the remaining data are taken as N2, and M2 and N2 are used as fitting group 2. Similarly, when the number of RMS data points of Nn is equal to the RMS data points of M1, Mn and Nn are used as fitting group n. Perform first-order least squares fitting on the n fitting groups to obtain two fitting straight lines and calculate the slope product of each group of fitting straight lines using the formula kMn·kNn. The fourth step is to take the time corresponding to the intersection of the fitting straight lines in the group of data with the largest slope product calculated in the third step as the EMG. FT . Figure 5 EMG FT Calculation principle.

[0071] 2.2 Lower limb stability analysis method

[0072] In order to analyze the stability of the knee joint during gait, motion capture was performed on two groups of subjects, A and B. In this experiment, IMU sensors at the thigh and calf were used to output the knee abduction angle with the Isen module. The angle accuracy of the IMU sensors used in the experiment was 5 decimal places, which met the experimental requirements. The abduction angle is α, and the coordinate plane is as follows: Figure 6 As shown in the figure. During gait, knee joint movement is carried out by the coordinated efforts of multiple muscles. When the lower limb muscles enter a fatigue state, it will affect the stability of the knee joint and cause the movement to deviate. Variance is a characteristic value that measures the degree of dispersion of a set of data. It can reflect the fluctuation of the data set, that is, the stability of the data set, for a single variable data set. Therefore, the variance of the lower limb knee abduction angle is positively correlated with the fatigue state of the lower limb. Therefore, this paper uses the knee joint offset angle for variance calculation, which will effectively reflect the lower limb stability and fatigue level of the subjects in the experiment. The variance of the knee joint offset distance is calculated using formula (2) for the data collected in the 1.0 experiment.

[0073]

[0074] In the formula is the average knee joint offset angle obtained in the experiment, and Xi is the i-th offset data. Finally, the least squares method is used to fit the variance to obtain the slope of each fitted line. This slope is used as the characteristic value of knee joint stability. The size of this characteristic value reflects the stability of the knee joint.

[0075] 2.3 Subjective scoring calculation method

[0076] Lower limb fatigue is a multidimensional phenomenon characterized by the interplay of subjective and objective perceptions. Comprehensively evaluating exoskeleton fatigue requires multiple sRPE (severely exertion) score responses from subjects. Therefore, this experiment used the modified CR-10 scale described in 1.1.2 to collect fatigue scores. Fatigue levels were categorized into ten levels, as shown in Table 2. Given that all subjects were weight-bearing and exercised at a moderate intensity, to improve accuracy, fatigue scores were expressed as two levels: effortful and very effortful. Based on this table, fatigue scores were collected from each subject four times during the six-minute experiment: at 0, 2, 4, and 6 minutes. The mean of each group was used as the characteristic sRPE score.

[0077] Table 2 CR-10 scale comparison table for experiments

[0078]

[0079] 3 Experimental results analysis

[0080] 3.1 EMG FT Calculation results

[0081] According to the preprocessing method described in 2.1.1, the raw sEMG signal was low-pass filtered and notch filtered. The preprocessing results of the raw sEMG signal of a random subject were displayed as follows: Figure 7 .

[0082] Through EMG in 2.1 FT The algorithm performs EMG on the pre-processed sEMG signal FT Calculate and randomly select a subject to display the experimental results, such as Figure 8 , EMG of each group FT The mean values ​​are shown in Table 3.

[0083] Table 3 EMG FT Mean Tab.3Fig.3Mean EMG FT

[0084]

[0085] Depend on Figure 8 It can be seen that in the two experiments of group A and group B, the exoskeleton can effectively delay the time for the vastus medialis muscle of the subjects to reach fatigue under load, and the effect of the exoskeleton on the EMG of group A(2) and group A(3) is greater than that of group B(2) and group B(3). FT However, the effect was not good for Group A(1) and Group B(2). Compared with Group A(1), Group A(1) EMG FTIt was only 1.3s later than the average of group B(1). This result shows that it is difficult for unskilled people to effectively relieve fatigue by using exoskeletons. Compared with group A(2), group A(2) EMG FT Compared with group B(2), the average delay was 51.7s. From this comparison, it can be seen that proficient use of the exoskeleton will greatly delay EMG. FT ; Comparison between Group A(3) and Group B(3), EMG of Group A(3) FT Compared with group B(3), which was delayed by an average of 44.6s, this group's results show that people with higher physical fitness can also use exoskeletons to delay fatigue.

[0086] In the experiment of group A, compared with group A(1), the subjects in group A(2) improved their exoskeleton usage by 211%, and the subjects in group A(3) improved their exoskeleton usage by 333%. Group A(2) improved its usage by 41% compared with group A(3). The horizontal comparison of group A showed that physical fitness and proficiency were the relevant factors that determined the effectiveness of exoskeleton usage. Under the same physical fitness conditions, proficiency played a decisive role.

[0087] In the experiment of group B, EMG of group B(2) FT The difference between group B(1) and group B(3) was 6%. FT Delayed by 157%. EMG of group B(3) compared with group B(2) FT Delayed by 140%. Compared with Group A, the results showed that the subjects with better physical fitness reached fatigue time later. At the same time, the physical fitness of Group B(2) was equivalent to that of Group B(1). Without the help of exoskeleton, EMG FT quite.

[0088] 3.2 Knee joint stability analysis

[0089] Perform the stability calculations in 2.2 on the knee joint. Since this experiment takes a long time and requires a large amount of data, window the data of groups A and B and calculate their variances. Set tim to 2.5s and mov to 5s. Obtain the variance of the subject's knee joint offset and perform a least squares curve fit on the variance. Calculate the slope K of the line. Randomly select a subject for the experimental results to display, as shown below. Figure 9 Comparing the results of Groups A and B, the lower limb stability during gait load training with the exoskeleton was higher than that without the exoskeleton. The slope of the variance fitting line of Group A was smaller than that of Group B, proving that the exoskeleton can effectively alleviate the increased knee joint vibration caused by fatigue.

[0090] A longitudinal comparison of Group A(1) and Group B(1) showed that the eigenvalue K decreased by 67.8%; the eigenvalue K decreased by 77.2% between Group A(2) and Group B(2); and the eigenvalue K decreased by 84.1% between Group A(3) and Group B(3). The comparison of the results in this group showed that the exoskeleton had a stable improvement in the stability of the lower limbs of the members of Groups A(1), A(2), and A(3), and had little correlation with factors such as physical fitness and proficiency.

[0091] A horizontal comparison of Group A showed that the eigenvalue K of Group A(2) decreased by 25.0% compared to Group A(1); the eigenvalue K of Group A(3) decreased by 60.7% compared to Group A(1); and the eigenvalue K of Group A(3) increased by 91.0% compared to Group A(2). These results indicate that high proficiency has a lower impact on lower limb stability, but good physical fitness has a greater impact on lower limb stability.

[0092] In a horizontal comparison of Group B, the eigenvalue K of Group B(2) increased by 8.0% compared to Group B(1); the eigenvalue K of Group B(3) decreased by 20.7% compared to Group B(1); and the eigenvalue K of Group B(3) increased by 25.0% compared to Group B(2). In the load-bearing gait experiment without an exoskeleton, physical fitness is directly related to lower limb stability.

[0093] 3.3sRPE control feedback

[0094] Subjective fatigue scores of Groups A and B were collected using the CR-10 scale described in 2.3. Feature values ​​were extracted for each group, and a group of subject data was randomly selected for presentation. The results are shown in Tables 4 and 5.

[0095] Table 4 sRPE score table for group A wearing exoskeleton and bearing weight

[0096]

[0097] Table 5 sRPE score table for group B wearing exoskeleton and bearing weight

[0098]

[0099] Scores were collected every 2 minutes. The results showed that wearing the exoskeleton to carry out the load test reduced the sRPE score by an average of 28.5%, and both groups A(3) and B(3) maintained a relatively relaxed subjective feeling. At the same time, the test results of group A(2) showed that this group also achieved a relatively relaxed state through the exoskeleton.

[0100] In Group A, when using the exoskeleton, the characteristic value of Group A(2) decreased by 43.5% compared to Group A(1); the characteristic value of Group A(3) decreased by 50.5% compared to Group A(1); and the characteristic value of Group A(3) decreased by 12.6% compared to Group A(2). In Group B, the characteristic value of Group B(2) decreased by 15.1% compared to Group B(1); the characteristic value of Group B(3) decreased by 53.1% compared to Group B(1); and the characteristic value of Group B(3) decreased by 44.8% compared to Group B(2).

[0101] The results showed that the exoskeleton had a subjective fatigue-reducing effect on unskilled users, and that the fatigue-reducing effect was further amplified after skilled use.

[0102] 3.4 Eigenvalue Comparison

[0103] The three groups of characteristic values ​​were summarized and compared in groups. The auxiliary lifting effect of the exoskeleton on the subjects under load is shown in Table 6.

[0104] Table 6 Comparison of three eigenvalue results

[0105]

[0106] As shown in Table 6, the exoskeleton used in this experiment played an auxiliary role for the subjects in three aspects. In terms of lower limb fatigue, proficient use of the exoskeleton will effectively delay EMG FT , an average delay of 42.9%. Although Group A(3) also delayed EMG by wearing exoskeleton FT , but compared with group B(3), its improvement effect was lower than that of groups A(2) and B(2); in terms of lower limb stability, the exoskeleton performed outstandingly, improving the stability by 75.8%. At the same time, the effects of groups A(3) and B(3) were relatively outstanding, so the exoskeleton can effectively improve the stability of the lower limbs and effectively relieve fatigue; in terms of subjective fatigue, the exoskeleton reduced the subjective fatigue of the subjects by an average of 30.3%. At the same time, skilled use of the exoskeleton can further reduce the subjective fatigue.

[0107] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention, and such changes and modifications fall within the scope of the invention as claimed.

Claims

1. Exoskeleton wearable fatigue assessment system based on surface electromyography and gait, characterized by: It includes test method unit, fatigue characteristics calculation and extraction unit and test result analysis unit; The experimental method unit includes experimental content and objectives, subjects and environment, and experimental design; the experimental content and objectives include experimental equipment and sRPE subjective scale selection; the collected data of the experimental design includes lower limb position information, sEMG raw signals, and sRPE subjective scale scores; The fatigue feature calculation and extraction unit includes EMG FT Algorithms, lower limb stability analysis methods, and subjective scoring calculation methods; The test result analysis unit includes EMG FT Calculation results, knee stability analysis, sRPE control feedback and eigenvalue comparison; The EMG FT The specific steps of the algorithm are as follows: In the first step, the sEMG signal preprocessing method is used to preprocess the two groups of sEMG raw data collected in the experiment, A and B, and establish the sEMG dataset; In the second step, the preprocessed sEMG dataset is windowed and the RMS dataset is obtained, as shown in formula (1): In formula (1), N t is the number of sEMG signal data in a single tim, i is the data sequence number, E i is the i-th data sequence number in the sEMG dataset; The third step is to divide the obtained RMS data set into two parts, M and N. Since the amount of data in this experiment is large, 15 RMS data points are taken as a group, recorded as M1, and the remaining data points are recorded as N1. M1 and N1 are used as fitting group 1, and 15+1 RMS data points are taken as M2, and the remaining data are taken as N2. M2 and N2 are used as fitting group 2, and so on. When the number of RMS data points of Nn is equal to the RMS data points of M1, M is taken as the fitting group 1. n , N n As the fitting group n, the first-order least squares fitting is performed on the n fitting groups to obtain two fitting straight lines and use the formula kM n· kN n Calculate the product of the slopes of each set of fitted lines; The fourth step is to take the time corresponding to the intersection of the fitted straight line in the set of data with the largest slope product calculated in the third step as the EMG FT ; The lower limb stability analysis method is: The variance of the knee joint offset distance was calculated using formula (2) for the data collected in experiment 1.0; In the formula is the average value of the knee joint deviation angle obtained in the experiment, X i is the i-th offset data; finally, the least square method is used to perform curve fitting on the variance and the slope of each set of fitting lines is obtained. The slope is used as the characteristic value of knee joint stability, and the size of this characteristic value reflects the stability of the knee joint; The subjective scoring calculation method divides the fatigue level into ten levels. Considering that all subjects are weight-bearing and the training intensity is moderate, in order to improve the accuracy, strenuous and very strenuous are expressed as two levels to improve the accuracy of the state description.

2. The exoskeleton wearable fatigue assessment system based on surface electromyography and gait according to claim 1 is characterized in that: According to EMG FT The preprocessing method described in the algorithm performs low-pass filtering and notch filtering on the original sEMG signal, and randomly selects a subject's original sEMG signal preprocessing result to display; through EMG FT The algorithm performs EMG on the pre-processed sEMG signal FT Calculate and randomly select a subject to display the experimental results.

3. The exoskeleton wearable fatigue assessment system based on surface electromyography and gait according to claim 1 is characterized in that: The knee joint stability analysis calculates the stability of the knee joint. Since the experiment lasted a long time and a large amount of data was obtained, the two groups of data, A and B, were windowed and their variances were calculated. tim was set to 2.5s and mov was set to 5s. The variance of the subject's knee joint offset was obtained and a least squares curve fit was performed on the variance. The slope K of the straight line was also calculated. The experimental results of a random subject were displayed.

4. The exoskeleton wearable fatigue assessment system based on surface electromyography and gait according to claim 1 is characterized in that: The sRPE control feedback was collected by collecting subjective fatigue scores of Groups A and B according to the CR-10 scale, and feature values ​​were extracted for each group, and a group of subject data was randomly selected for display.

5. The exoskeleton wearable fatigue assessment system based on surface electromyography and gait according to claim 1 is characterized in that: The characteristic value comparison summarizes the three groups of characteristic values ​​and compares them in groups, and the auxiliary lifting effect of the exoskeleton on the subject under the load state is demonstrated.

Images