An exoskeleton gait parameter optimization method for lower limb rehabilitation training

Abstract

The application discloses a kind of exoskeleton gait parameter optimization methods for lower limb rehabilitation training.First, the initial gait of lower limb rehabilitation exoskeleton is set, and the exoskeleton-gait fitness under current gait is obtained after user wears lower limb rehabilitation exoskeleton and carries out walking experiment;Again, according to the exoskeleton-gait fitness corresponding to the gait of current walking experiment, the gait of gait library that has been walked experiment is clustered using mixed gaussian model, and the gait clustering cluster is obtained, and then the utility degree of corresponding gait in gait library is updated;Then, the gait in gait library is sampled, and the sampling gait set is obtained, and then the gait with the maximum utility degree is selected from the sampling gait set, and the gait is continuously optimized, and finally, the gait parameters corresponding to the final gait are used as optimal gait parameters.The application can predict the lower limb exoskeleton gait parameters that meet the needs of user, and the calculation cost is low, the convergence speed is fast, and the robustness is high.

Description

Technical Field

[0001] This invention belongs to the field of rehabilitation robot technology and is a method for optimizing skeletal gait parameters, specifically an exoskeleton gait parameter optimization method for lower limb rehabilitation training. Background Technology

[0002] Patients with spinal cord injuries, strokes, and other conditions often experience central nervous system damage, with only 3-10% regaining the ability to walk independently for 50 meters within three years of the injury. Long-term, standardized, and scientific rehabilitation exercise training can rebuild the relevant functional areas of the nervous system, enabling patients to regain their independent motor abilities. China is rapidly entering a deeply aging society, and disabled and semi-disabled elderly individuals require rehabilitation equipment and basic rehabilitation services. Lower limb rehabilitation exoskeletons can provide systematic rehabilitation training without relying on physical therapists, and gait design is one of the key technologies of lower limb rehabilitation exoskeletons. Gait refers to the trajectory of walking driven by the lower limb rehabilitation exoskeleton; personalized gait design for each patient plays a crucial role in improving the efficiency of rehabilitation training and accelerating lower limb recovery.

[0003] However, current technology lacks consideration for personalized gait parameters. Because different patients have different leg structure parameters and varying degrees of motor function impairment, effective gait optimization methods tailored to individual patients are necessary to achieve ideal rehabilitation results. Generating gait solely based on patient body characteristic parameters cannot fully reflect the user's preference for gait parameters; therefore, it is necessary to utilize user feedback on gait preferences to optimize the user's gait parameters. Summary of the Invention

[0004] To address the problems in the background art, this invention proposes a method for optimizing exoskeleton gait parameters for lower limb rehabilitation training. By quantifying the exoskeleton gait, clustering the gait using a Gaussian mixture model based on the gait-exoskeleton fit, calculating the gait utility using a multi-armed gambling machine algorithm, predicting the posterior probability that the gait meets user needs, and optimizing the gait parameters. This invention can predict lower limb exoskeleton gait parameters that meet user needs, and it has low computational cost, fast convergence speed, and high robustness.

[0005] To achieve the above objectives, the technical solution of the present invention includes:

[0006] 1) Set the initial gait of the lower limb rehabilitation exoskeleton. After the user wears the lower limb rehabilitation exoskeleton and conducts a walking experiment, obtain the exoskeleton-gait fit degree under the current gait.

[0007] 2) Based on the exoskeleton-gait fit of the gait in the current walking experiment, use a Gaussian mixture model to cluster the gait in the gait database to obtain gait clusters;

[0008] 3) Calculate the utility of gait clusters and update the utility of the corresponding gait in the gait library;

[0009] 4) Sample the gaits in the gait library based on the current utility of the gait to obtain a sampled gait set, and then select the gait with the highest utility from the gait set;

[0010] 5) Based on the gait with the highest current utility, the user wears the lower limb rehabilitation exoskeleton and conducts a walking experiment to obtain the exoskeleton-gait fit under the current gait. Repeat steps 2)-4) until the preset number of rounds is reached, and take the gait parameters corresponding to the final gait as the optimal gait parameters.

[0011] Specifically, 2) refers to:

[0012] 2.1) The Bayesian information criterion operator is used to calculate the Bayesian information criterion operator before and after doubling the number of clusters in the Gaussian mixture model. The difference between the two is used to obtain the operator difference. The final number of clusters in the Gaussian mixture model is obtained by adjusting the operator difference.

[0013] 2.2) Based on the number of clusters in the current Gaussian mixture model and the exoskeleton-gait fit of the gait in the current walking experiment, the Gaussian mixture model is used to cluster the gaits in the walking experiment in the gait library to obtain the initial gait clusters;

[0014] 2.3) Based on the minimum coverage area, adjust the coverage area of ​​the current gait cluster to obtain the updated gait cluster and use it as the final gait cluster.

[0015] In section 2.1), the final number of clusters in the Gaussian mixture model is obtained by adjusting the operator differences, specifically as follows:

[0016] The operator difference is obtained by subtracting the doubled Bayesian information criterion operator from the original Bayesian information criterion operator. When the operator difference is less than 0, the number of clusters in the current Gaussian mixture model is doubled to obtain the final number of clusters in the Gaussian mixture model; otherwise, the number of clusters in the current Gaussian mixture model remains unchanged.

[0017] The formula for the Gaussian mixture model of gait is as follows:

[0018] G = {(α1, θ1)} (L，H，W，T，f) (α2, θ3) (L，H，W，T，f) , …, (α K θ K ) (L，H，W，T，f)}

[0019] Where G represents the Gaussian mixture model of gait, (α) K θ K ) (L，H，W，T，f)Let L represent the Kth Gaussian distribution cluster in the gait mixture Gaussian model, H represent the stride length, H represent the stride height, W represent the stride width, T represent the gait period, f represent the exoskeleton-gait fit, and α represent the gait pattern. K Let θ be the probability of the Kth gait following a Gaussian distribution. K Let K represent the probability density function of the Gaussian distribution of the Kth gait, where K is the number of clusters in the Gaussian mixture model.

[0020] In section 2.3), for each initial gait cluster, the coverage area of ​​each initial gait cluster is calculated using the following formula:

[0021]

[0022] Where Cov represents the coverage area of ​​each initial gait cluster, σ i Let represent the standard deviation of the i-th dimension of the gait cluster, and N represent the dimension of the gait parameters.

[0023] If the coverage area of ​​the current gait cluster is less than the minimum coverage area, then the current minimum coverage area is used as the coverage area of ​​the current gait cluster, and the current gait cluster is updated accordingly; otherwise, the coverage area of ​​the current gait cluster remains unchanged.

[0024] The formula for calculating the minimum coverage area is as follows:

[0025] minTh=Th×d t

[0026] Where minTh is the minimum coverage area, Th is the current minimum coverage area, d is the convergence rate, and t is the number of iterations.

[0027] The gait parameters include stride length, stride width, stride height, and gait period.

[0028] In step 3), gait clusters are used as the arms of the multi-armed gambling machine algorithm, and the utility corresponding to the gait clusters is calculated using the multi-armed gambling machine algorithm.

[0029] In the multi-armed gambling machine algorithm described in 3), the utility calculation formula for gait clustering is as follows:

[0030]

[0031] Where P(f|D, GMM) represents the utility of the gait cluster, D represents the set of gaits from the walking experiment, GMM represents the Gaussian mixture model of gait, and K is the number of clusters in the Gaussian mixture model. Let represent the probability of the k-th gait in the t-th round being a Gaussian distribution. Let represent the Gaussian mixture model of the gait in round t. This represents the Gaussian distribution of the k-th gait in the t-th round. Let φ(f) represent the covariance matrix of the Gaussian distribution of the k-th gait in the t-th round. i t , σ) represents the cumulative distribution function of the Gaussian distribution formed by the gait data from the walking experiment, and σ is the standard deviation of the noise value of the Gaussian distribution.

[0032] The beneficial effects of this invention are as follows:

[0033] This invention can predict lower limb exoskeleton gait parameters that meet user needs, and has low computational cost, fast convergence speed, and high robustness. Attached Figure Description

[0034] Figure 1 A detailed flowchart for optimizing exoskeleton gait parameters for lower limb rehabilitation training.

[0035] Figure 2 This is a schematic diagram of the stride length and stride width parameters of gait.

[0036] Figure 3 This is a schematic diagram of the stride height and gait period parameters.

[0037] Figure 4 A flowchart of the overall method for optimizing exoskeleton gait parameters for lower limb rehabilitation training. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0039] like Figure 1 and Figure 4 As shown, the present invention includes the following steps:

[0040] 1) Set the initial gait of the lower limb rehabilitation exoskeleton. After the user wears the lower limb rehabilitation exoskeleton and conducts a walking experiment, the exoskeleton-gait fit degree under the current gait is obtained. In specific implementation, the user's subjective evaluation of the exoskeleton under the current gait is used as the exoskeleton-gait fit degree; the exoskeleton-gait fit degree is a numerical type.

[0041] In this embodiment, a total of 8 initial experimental gait groups were conducted, and the results are shown in Table 1:

[0042] Table 1 shows the initial gait parameters.

[0043] Serial Number Step length Step High stride width Gait cycle Evaluation value 1 1.24 0.09 0.18 3.90 31 2 1.26 0.09 0.19 2.58 39 3 1.29 0.09 0.20 1.99 75 4 1.29 0.09 0.18 1.75 74 5 1.36 0.13 0.21 1.27 80 6 1.35 0.10 0.18 1.03 76 7 1.46 0.12 0.18 0.91 60 8 1.50 0.13 0.20 0.79 41

[0044] 2) Based on the exoskeleton-gait fit of the gait observed in the current walking experiments, a Gaussian mixture model is used to cluster the gait patterns from the walking experiments in the gait database, obtaining gait clusters. The gait parameters for each group of gait patterns include stride length, stride width, stride height, and gait cycle, such as... Figure 2 and Figure 3 As shown.

[0045] 2) Specifically:

[0046] 2.1) The Bayesian information criterion operator is used to calculate the Bayesian information criterion operator before and after doubling the number of clusters in the Gaussian mixture model. The difference between the two is used to obtain the operator difference. The final number of clusters in the Gaussian mixture model is obtained by adjusting the operator difference.

[0047] In section 2.1), the final number of clusters in the Gaussian mixture model is obtained by adjusting the operator differences, specifically as follows:

[0048] The operator difference is obtained by subtracting the doubled Bayesian information criterion operator from the original Bayesian information criterion operator, i.e., ΔBIC = BIC. K -BIC 2K Among them, BIC K The Bayesian information criterion operator before doubling, BIC 2K This represents the doubled Bayesian information criterion operator. When the operator difference ΔBIC is less than 0, it indicates that the existing number of clusters is insufficient to accurately fit the model. In this case, the number of clusters in the current Gaussian mixture model is doubled to obtain the final number of clusters in the Gaussian mixture model. Otherwise, the number of clusters in the current Gaussian mixture model remains unchanged, i.e., it remains at K.

[0049] In this embodiment, the calculation formulas for the Bayesian information criterion operator before and after doubling the number of clusters in the Gaussian mixture model are as follows:

[0050]

[0051] ΔBIC=BIC2-BIC4=-74.89-(-124.20)=47.31

[0052] Since ΔBIC > 0, it indicates that the existing number of clusters is sufficient to accurately fit the data, and there is no need to increase the number of clusters.

[0053] 2.2) Based on the number of clusters in the current Gaussian mixture model and the exoskeleton-gait fit of the gait in the current walking experiment, the Gaussian mixture model is used to cluster the gaits in the walking experiment in the gait library to obtain the initial gait clusters;

[0054] The formula for the Gaussian mixture model of gait is as follows:

[0055] G = {(α1, θ1)} (L，H，W，T，f) (α2, θ3) (L，H，W，T，f) , …, (α K θ K ) (L，H，W，T，f)}

[0056] Where G represents the Gaussian mixture model of gait, (α) K θ K ) (L，H，W，T，f) Let g = {L, H, W, T, f} represent the Kth Gaussian distribution cluster in the gait mixture Gaussian model, where g represents a set of gaits, L is stride length, H is stride height, W is stride width, T is gait period, f is exoskeleton-gait fit, and α is the exoskeleton-gait fit. K Let θ be the probability of the Kth gait following a Gaussian distribution. K Let θ represent the probability density function of the Gaussian distribution of the Kth gait, satisfying θ K =(μ K , ∑ K ), μ K Let ∑ be the mean of the Gaussian distribution of the Kth gait. K Let K be the covariance matrix of the Kth gait Gaussian distribution. K is the number of clusters in the Gaussian mixture model. Initialize K = 2, and set parameters to random.

[0057] Solve for the parameters of the gait mixture Gaussian model using the expectation-maximization algorithm:

[0058] α = [0.625 0.375]

[0059]

[0060]

[0061] 2.3) Based on the minimum coverage area, adjust the coverage area of ​​the current gait cluster to obtain the updated gait cluster and use it as the final gait cluster.

[0062] For each initial gait cluster, the coverage area of ​​each initial gait cluster is calculated using the following formula:

[0063]

[0064] Where Cov represents the coverage area of ​​each initial gait cluster, σ i Let represent the standard deviation of the i-th dimension of the gait cluster, and N represent the dimension of the gait parameters.

[0065] In the early stages of gait parameter optimization, due to the small number of sampling points and inaccurate model fitting, if the coverage area of ​​the current gait cluster is less than the minimum coverage area, it may be impossible to explore unknown regions. In this case, the minimum coverage area of ​​the Gaussian model should be limited, and the current minimum coverage area should be used as the coverage area of ​​the current gait cluster to update the current gait cluster; otherwise, the coverage area of ​​the current gait cluster remains unchanged.

[0066] The formula for calculating the minimum coverage area is as follows:

[0067] minTh=Th×d t

[0068] Where minTh is the minimum coverage area, Th is the current minimum coverage area, d is the convergence rate, and t is the number of iterations.

[0069] In this embodiment, the formulas for calculating the coverage area and minimum coverage region of the gait cluster are as follows:

[0070]

[0071] The initial minimum coverage area Th is 1.25e-5, and the convergence rate is 0.8.

[0072] minTh=Th×d t =1e-5

[0073] Since Cov < minTh, the model variance is corrected. The minimum variance is log51e-5 = 0.1, and the corrected variance is [0.1, 0.1, 0.1, 0.1, 46.4].

[0074] 3) Calculate the utility of gait clusters and update the utility of the corresponding gait in the gait library;

[0075] Gait clusters are used as arms in a multi-armed gambling machine algorithm, and the utility of each gait cluster is calculated using this algorithm. The formula for calculating the utility of a gait cluster is as follows:

[0076]

[0077] Where P(f|D, GMM) represents the utility of the gait cluster, D represents the set of gaits from the walking experiment, GMM represents the Gaussian mixture model of gait, and K is the number of clusters in the Gaussian mixture model. Let represent the probability of the k-th gait in the t-th round being a Gaussian distribution. Let represent the Gaussian mixture model of the gait in round t. This represents the Gaussian distribution of the k-th gait in the t-th round. Let φ(f) represent the covariance matrix of the Gaussian distribution of the k-th gait in the t-th round. it , σ) represents the cumulative distribution function of the Gaussian distribution formed by the gait data from the walking experiment, and σ is the standard deviation of the noise value of the Gaussian distribution.

[0078] 4) Continuously sample gaits from the gait database and use pruning methods to eliminate gaits that clearly cannot meet user expectations, thereby gradually narrowing the exploration scope and accelerating the convergence process. Based on the current utility of each gait, sample gaits from the database to obtain a sampled gait set, and then select the gait with the highest utility from this set; the specific formula is as follows:

[0079]

[0080] Among them, g t+1 P(f) represents the gait parameters selected in round t+1, argmax represents the maximum value operation, and P(f) represents the gait parameters selected in round t+1. t |D t GMM t ) represents the utility of the gait cluster in round t.

[0081] 5) Based on the gait with the highest current utility, the user wears the lower limb rehabilitation exoskeleton and conducts a walking experiment to obtain the exoskeleton-gait fit under the current gait. Repeat steps 2)-4) until the preset number of rounds is reached, and take the gait parameters corresponding to the final gait as the optimal gait parameters.

[0082] In this embodiment, the sampled gait set is shown in Table 2:

[0083] Table 2 is a sample gait set table.

[0084] Serial Number Step length Step High stride width Gait cycle Gait utility 1 1.29 0.08 0.19 1.85 75.08 2 1.37 0.12 0.20 1.08 79.56 3 1.30 0.09 0.17 1.51 74.91 4 1.32 0.10 0.20 1.66 77.60 5 1.32 0.09 0.17 1.31 75.43 6 1.32 0.09 0.18 1.29 75.06 7 1.30 0.11 0.20 1.83 77.13 8 1.50 0.13 0.19 0.79 40.99

[0085] Based on the calculation results, the second set of gait is more in line with the current user. Therefore, the stride length of 1.37m, stride height of 0.12m, stride width of 0.20m, and gait cycle of 1.08s are used as the gait parameters for the next round.

[0086] The above specific embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for optimizing exoskeleton gait parameters for lower limb rehabilitation training, characterized in that, Includes the following steps: 1) Set the initial gait of the lower limb rehabilitation exoskeleton. After the user wears the lower limb rehabilitation exoskeleton and conducts a walking experiment, obtain the exoskeleton-gait fit degree under the current gait. 2) Based on the exoskeleton-gait fit of the gait in the current walking experiment, use a Gaussian mixture model to cluster the gait in the gait database to obtain gait clusters; 3) Calculate the utility of gait clusters and update the utility of the corresponding gait in the gait library; In step 3), gait clusters are used as the arms of the multi-armed gambling machine algorithm, and the utility corresponding to the gait clusters is calculated using the multi-armed gambling machine algorithm. In the multi-armed gambling machine algorithm described in 3), the utility calculation formula for gait clustering is as follows: in, The utility of gait clusters is represented by , D represents the set of gaits from the walking experiments, and GMM represents the Gaussian mixture model of gait. The number of clusters in the Gaussian mixture model. Let represent the probability of the k-th gait in the t-th round being a Gaussian distribution. Let represent the Gaussian mixture model of the gait in round t. This represents the Gaussian distribution of the k-th gait in the t-th round. Let represent the covariance matrix of the Gaussian distribution of the k-th gait in the t-th round. The cumulative distribution function represents the Gaussian distribution formed by the gait data from the walking experiment. is the standard deviation of the Gaussian-distributed noise values; 4) Sample the gaits in the gait library based on the current utility of the gait, obtain the sampled gait set, and then select the gait with the highest utility from the gait set; 5) Based on the gait with the highest current utility, after the user wears the lower limb rehabilitation exoskeleton and conducts a walking experiment, the exoskeleton-gait fit under the current gait is obtained. Repeat 2)-4) until the preset number of rounds is reached, and the gait parameters corresponding to the final gait are taken as the optimal gait parameters.

2. The method for optimizing exoskeleton gait parameters for lower limb rehabilitation training according to claim 1, characterized in that, Specifically, 2) refers to: 2.1) The Bayesian information criterion operator is used to calculate the Bayesian information criterion operator before and after doubling the number of clusters in the Gaussian mixture model. The difference between the two is used to obtain the operator difference. The final number of clusters in the Gaussian mixture model is obtained by adjusting the operator difference. 2.2) Based on the number of clusters in the current Gaussian mixture model and the exoskeleton-gait fit of the gait in the current walking experiment, the Gaussian mixture model is used to cluster the gaits in the walking experiment in the gait library to obtain the initial gait clusters; 2.3) Based on the minimum coverage area, adjust the coverage area of ​​the current gait cluster to obtain the updated gait cluster and use it as the final gait cluster.

3. The method for optimizing exoskeleton gait parameters for lower limb rehabilitation training according to claim 2, characterized in that, In section 2.1), the final number of clusters in the Gaussian mixture model is obtained by adjusting the operator differences, specifically as follows: The operator difference is obtained by subtracting the doubled Bayesian information criterion operator from the original Bayesian information criterion operator. When the operator difference is less than 0, the number of clusters in the current Gaussian mixture model is doubled to obtain the final number of clusters in the Gaussian mixture model; otherwise, the number of clusters in the current Gaussian mixture model remains unchanged.

4. The method for optimizing exoskeleton gait parameters for lower limb rehabilitation training according to claim 2, characterized in that, The formula for the Gaussian mixture model of gait is as follows: in, This represents a Gaussian mixture model of gait. The first term in the Gaussian mixture model of gait represents the second term. A Gaussian distributed cluster, Step size, For BBK, For step width, For gait period, For exoskeleton-gait fit, For the first The probability of a Gaussian distribution of gait. Indicates the first The probability density function of a Gaussian distribution of gait. This represents the number of clusters in the Gaussian mixture model.

5. The method for optimizing exoskeleton gait parameters for lower limb rehabilitation training according to claim 2, characterized in that, In section 2.3), for each initial gait cluster, the coverage area of ​​each initial gait cluster is calculated using the following formula: in, This represents the coverage area of ​​each initial gait cluster. The standard deviation of the i-th dimension of the gait cluster is represented. The dimension representing the gait parameters; If the coverage area of ​​the current gait cluster is less than the minimum coverage area, then the current minimum coverage area is used as the coverage area of ​​the current gait cluster, and the current gait cluster is updated accordingly; otherwise, the coverage area of ​​the current gait cluster remains unchanged.

6. The method for optimizing exoskeleton gait parameters for lower limb rehabilitation training according to claim 5, characterized in that, The formula for calculating the minimum coverage area is as follows: in, For the smallest coverage area, The current minimum coverage area, Let be the convergence ratio. This represents the number of iterations.

7. The method for optimizing exoskeleton gait parameters for lower limb rehabilitation training according to claim 1, characterized in that, The gait parameters include stride length, stride width, stride height, and gait period.

Images