A Method and System for Estimating Human Joint Stiffness Based on Musculoskeletal Dynamics Model

Abstract

This invention provides a method and system for estimating human joint stiffness based on a musculoskeletal dynamics model, belonging to the field of human mechanical impedance measurement. The method includes: inputting preprocessed electromyographic (EMG) signals into a muscle activation dynamics model to calculate muscle activation signals; inputting human limb characteristic parameters into a musculoskeletal geometric model to obtain microscopic parameters of muscle structure; calculating muscle-tendon unit forces and muscle forces using a muscle contraction dynamics model based on the muscle activation signals and muscle structure microscopic parameters; obtaining the stiffness of muscle-tendon units using a muscle-tendon unit stiffness model based on muscle forces and muscle structure microscopic parameters; and calculating joint stiffness using a muscle-tendon-to-joint mapping model based on muscle-tendon unit forces, muscle structure microscopic parameters, and muscle-tendon unit stiffness. This invention eliminates the need for stiffness identification experiments and model training; it only requires establishing a musculoskeletal dynamics model and collecting EMG signals from the human body surface to accurately estimate human joint stiffness.

Description

Technical Field

[0001] This invention belongs to the field of mechanical impedance measurement, and more specifically, relates to a method and system for estimating the stiffness of human joints based on a musculoskeletal dynamics model. Background Technology

[0002] In existing studies on human stiffness, two main methods are used to obtain ankle joint stiffness: stiffness identification and stiffness estimation. Stiffness estimation largely relies on stiffness identification results. While ankle joint stiffness identification methods can obtain ankle joint stiffness characteristics under various postures, they typically require the development of specialized identification devices and the conduct of corresponding identification experiments. This inherently necessitates a certain hardware foundation for stiffness acquisition. Furthermore, due to limitations of the identification device, the subject's posture is usually relatively uniform, making it difficult to extend to other movement scenarios. Therefore, stiffness identification methods have significant limitations in practical engineering applications. In addition, perturbation-dependent human stiffness identification can currently only be performed offline, and the few online stiffness identification studies still suffer from significant identification delays in practical applications.

[0003] To quickly and conveniently obtain the stiffness of human joints, researchers have made numerous attempts. Among these, establishing a stiffness estimation model based on identified stiffness data and collected electromyographic signals to estimate the stiffness of human joints is the most common method used in current research. Various methods exist for establishing stiffness estimation models. With the development of information technology, data-driven models with predictive capabilities, built using black-box models and neural networks, are increasingly used in engineering applications. This data-driven method has the advantages of simple modeling and excellent estimation accuracy. However, the training of stiffness estimation models is based on stiffness identification, which has many limitations, leading to difficulties in implementation and limited application scenarios for stiffness estimation based on data-driven models. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for estimating human joint stiffness based on musculoskeletal dynamics models. This invention addresses the problem that existing musculoskeletal dynamics models, as a method for quickly and accurately obtaining human ankle joint stiffness, suffer from numerous sub-model components, leading to both a high modeling difficulty and an amplified estimation error due to the increased number of intermediate variables, resulting in low estimation accuracy.

[0005] This invention proposes a method and system for estimating human joint stiffness based on a musculoskeletal dynamics model. This invention can directly obtain joint stiffness through muscle geometric parameters and electromyographic signals, without requiring stiffness identification or stiffness estimation model training, thus overcoming the limitations of traditional stiffness identification or estimation methods. Furthermore, this method not only conforms to the principles of neural transmission and human anatomical structure, giving it advantages in the biological interpretability and universality of the model, but also boasts advantages such as high model estimation accuracy and simple data computation.

[0006] To achieve the above objectives, the human joint stiffness estimation method based on a musculoskeletal dynamics model provided by this invention includes the following steps:

[0007] D1: The acquired raw surface electromyography (EMG) signals are subjected to high-pass filtering to remove motion artifacts, followed by full-wave rectification and normalization to complete the preprocessing of the EMG signals.

[0008] D2: Input the preprocessed surface electromyography signal into the muscle activation kinetics model, and calculate the muscle activation signal through the muscle activation kinetics model;

[0009] D3: Input the measured human limb feature parameters into the musculoskeletal geometry model, and obtain the microscopic parameters of muscle structure through the musculoskeletal geometry model;

[0010] D4: Based on muscle activation signals and microscopic parameters of muscle structure, muscle force and muscle-tendon unit force under muscle activation are calculated using a muscle contraction dynamics model.

[0011] D5: The stiffness of the muscle tendon unit is calculated using the muscle force and microscopic parameters of the muscle structure under muscle activation through the muscle tendon unit stiffness model.

[0012] D6: Based on the force of the muscle-tendon unit under muscle activation, the microscopic parameters of the muscle structure, and the stiffness of the muscle-tendon unit, the joint stiffness is calculated through the mapping model from muscle-tendon to joint.

[0013] More preferably, the musculoskeletal geometric model is as follows:

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026]

[0027]

[0028]

[0029] (φ o ) sol =28.3π / 180,(φ o ) mg =9.9π / 180,(φ o ) lg =12π / 180,(φ o ) ta =9.6π / 180

[0030] Among them, l mtu l is the length of a muscle-tendon unit. shank Calf length A function relating to the initial ankle angle α0 and the angle change Δα, (φ o ) sol 、(φ o ) mg 、(φ o ) lg 、(φ o ) ta The optimal pennate angles are for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. The optimal fiber lengths for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles are respectively. o h is the height of the subject; The resting lengths of the tendon units of the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. This is the resting length of the tendon. These represent the musculoskeletal geometric model coefficients σ for the four muscles mentioned above. 1 / σ 2 .

[0031] More preferably, the muscle contraction dynamics model is as follows:

[0032]

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043] Among them, F mtu For muscle and tendon unit force; F m For muscle strength, Active tension generated by muscles φ represents the passive tension generated by the muscle, and φ is the angle between the muscle force and the tendon. This represents the force-length relationship during active muscle contraction. This represents the force-velocity relationship during active muscle contraction. The normalized muscle fiber length, its value is equal to the current muscle fiber length l. m Optimal muscle fiber length at muscle activation intensity 'a' The ratio, The current velocity v of the muscle fiber m With the maximum contraction speed of the muscle The ratio, v m The value is equal to the current muscle fiber length l m The derivative, The value is equal to 10l o ;l o The optimal muscle fiber length, φ, is the length at which the muscle generates maximum contractile force. o For the optimal feather angle of the muscle, l is the resting length of the tendon. mtu The length of a muscle-tendon unit; This represents the maximum muscle force generated during isometric muscle contraction. This represents the passive elastic force-length relationship of the muscle.

[0044] More preferably, the muscle activation kinetics model is as follows:

[0045] u j (t)=f a (e j (t))

[0046]

[0047]

[0048] Among them, u j (t) represents the neural activation signal of the j-th muscle; e j (t) represents the surface electromyography signal; f a For neural activation models; a j (t) is the activation signal of the j-th muscle; c1+c2 is the activation ratio constant, c2 is the inactivation ratio constant, and t deact t is the deactivation time constant; act Let be the muscle activation time constant; ζ is the muscle activation-deactivation proportionality coefficient, taken as ζ = 0.5, and t is the muscle activation time constant. act Take 15ms.

[0049] More preferably, the mapping model from muscle / tendon to joint is as follows:

[0050]

[0051]

[0052]

[0053] in,

[0054]

[0055]

[0056] in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle. Let j be the deformation of the j-th muscle. Let T be the torque arm of the j-th muscle; j K represents the torque exerted by the j-th muscle on the joint under study; j For the contribution of the j-th muscle to the stiffness of the joint, K of the ankle joint jThis includes the contributions of four muscles: soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior; K represents joint stiffness. Let be the stiffness of the muscle-tendon unit of the j-th muscle. Let α be the torque arm of the muscle-tendon unit of the j-th muscle, and α be the joint angle. Let be the force of the muscle-tendon unit of the j-th muscle.

[0057] On the other hand, the present invention provides a human joint stiffness estimation system based on a musculoskeletal dynamics model, comprising:

[0058] The electromyography (EMG) signal preprocessing module is used to perform high-pass filtering to remove motion artifacts from the acquired raw surface EMG signals, and then perform full-wave rectification and normalization to complete the EMG signal preprocessing process.

[0059] The muscle activation signal calculation module is used to input the preprocessed surface electromyography signal into the muscle activation kinetic model and calculate the muscle activation signal through the muscle activation kinetic model.

[0060] The module for acquiring microscopic parameters of muscle structure is used to input the measured human limb feature parameters into the musculoskeletal geometric model, and to obtain the microscopic parameters of muscle structure through the musculoskeletal geometric model.

[0061] The muscle force calculation module is used to calculate the muscle force and muscle-tendon unit force under muscle activation based on the muscle activation signal and the microscopic parameters of the muscle structure through a muscle contraction dynamics model.

[0062] The muscle and tendon unit stiffness calculation module is used to calculate the stiffness of the muscle and tendon unit based on the muscle force and microscopic parameters of the muscle structure under muscle activation, through the muscle and tendon unit stiffness model.

[0063] The joint stiffness calculation module calculates joint stiffness based on the force of muscle and tendon units under muscle activation, the microscopic parameters of muscle structure, and the stiffness of muscle and tendon units, through a mapping model from muscle and tendon to joint.

[0064] The musculoskeletal dynamics model includes a muscle activation dynamics model, a musculoskeletal geometric model, a muscle contraction dynamics model, a muscle-tendon unit stiffness model, and a muscle-tendon-to-joint mapping model.

[0065] More preferably, the musculoskeletal geometric model is as follows:

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] (φ o ) sol =28.3π / 180,(φ o ) mg =9.9π / 180,(φ o ) lg =12π / 180,(φ o ) ta =9.6π / 180

[0082] Among them, l mtu l is the length of a muscle-tendon unit. shank The length of the lower leg. A function relating to the initial ankle angle α0 and the angle change Δα, (φ o ) sol 、(φ o ) mg 、(φ o ) lg 、(φ o ) ta The optimal pennate angles are for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. The optimal fiber lengths for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles are respectively. o h represents the height of the subject; The resting length of the tendon units of the above four muscles are respectively. This is the resting length of the tendon. The coefficients σ represent the musculoskeletal geometric models of the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. 1 / σ 2 .

[0083] More preferably, the muscle contraction dynamics model is as follows:

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095] Among them, F mtu For muscle and tendon unit force; F m For muscle strength, Active tension generated by muscles φ represents the passive tension generated by the muscle, and φ is the angle between the muscle force and the tendon. This represents the force-length relationship during active muscle contraction. This represents the force-velocity relationship during active muscle contraction. The normalized muscle fiber length, its value is equal to the current muscle fiber length l. m Optimal muscle fiber length at muscle activation intensity 'a' The ratio, The current velocity v of the muscle fiber m With the maximum contraction speed of the muscle The ratio, v m The value is equal to the current muscle fiber length l m The derivative, The value is equal to 10l o ;l oThe optimal muscle fiber length, φ, is the length at which the muscle generates maximum contractile force. o For the optimal feather angle of the muscle, l is the resting length of the tendon. mtu The length of a muscle-tendon unit; This represents the maximum muscle force generated during isometric muscle contraction. This represents the passive elastic force-length relationship of the muscle.

[0096] More preferably, the muscle activation kinetics model is as follows:

[0097] u j (t)=f a (e j (t))

[0098]

[0099]

[0100] Among them, u j (t) represents the neural activation signal of the j-th muscle; e j (t) represents the surface electromyography signal; f a For neural activation models; a j (t) is the activation signal of the j-th muscle; c1+c2 is the activation ratio constant, c2 is the inactivation ratio constant, and t deact t is the deactivation time constant; act Let be the muscle activation time constant; ζ is the muscle activation-deactivation proportionality coefficient, taken as ζ = 0.5, and t is the muscle activation time constant. act Take 15ms.

[0101] More preferably, the mapping model from muscle / tendon to joint is as follows:

[0102]

[0103]

[0104]

[0105] in,

[0106]

[0107]

[0108] in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle. Let j be the deformation of the j-th muscle. Let T be the torque arm of the j-th muscle;j K represents the torque exerted by the j-th muscle on the joint under study; j For the contribution of the j-th muscle to joint stiffness, the K value of the ankle joint... j This includes the contributions of four muscles: soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior; K represents joint stiffness. Let be the stiffness of the muscle-tendon unit of the j-th muscle. Let α be the torque arm of the muscle-tendon unit of the j-th muscle, and α be the joint angle. Let be the force of the muscle-tendon unit of the j-th muscle.

[0109] In summary, compared with the prior art, the above-described technical solutions conceived by this invention have the following advantages:

[0110] Beneficial effects:

[0111] This invention proposes a method for estimating human joint stiffness based on a musculoskeletal dynamics model. This method can directly obtain joint stiffness through muscle geometric parameters and electromyographic signals, without the need for stiffness identification or stiffness estimation model training, thus overcoming the limitations of traditional stiffness identification or estimation methods. Traditional methods for obtaining muscle micro-parameters often use OpenSim and SIMM models similar to human anatomical models. However, the establishment, application, and data acquisition of these models are relatively complex and computationally intensive, making it difficult to meet the needs of efficient stiffness estimation. Based on existing research on the relationship between human muscle structural parameters and limb characteristics, this invention establishes a musculoskeletal geometric model that conforms to the physiological structure of the ankle joint, enabling accurate and efficient acquisition of muscle micro-parameters. Furthermore, five sub-models are established: a muscle activation dynamics model, a muscle contraction dynamics model, a muscle-tendon unit stiffness model, and a muscle-tendon-to-joint mapping model, constituting a musculoskeletal dynamics model. Ultimately, this achieves the effect of estimating joint stiffness simply by inputting limb parameters and electromyographic signals. Compared to stiffness estimation models based on identification data, this method not only conforms to the principles of neural transmission and the characteristics of human anatomical structure, but also has advantages in terms of the biological interpretability and universality of the model. Furthermore, it has many advantages such as high model estimation accuracy and simple data computation. Attached Figure Description

[0112] Figure 1 This is a schematic diagram of the musculoskeletal dynamics model framework provided in an embodiment of the present invention;

[0113] Figure 2 This is a schematic diagram of an experimental scenario for stiffness identification research provided in an embodiment of the present invention;

[0114] Figure 3 This is a comparison diagram of ankle joint stiffness estimated by the musculoskeletal dynamics model provided in this embodiment of the invention and the identified stiffness. Detailed Implementation

[0115] 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.

[0116] This embodiment is a case study on model accuracy verification based on a research project on ankle joint stiffness estimation in the human lower limb. In stiffness estimation research, to verify the accuracy of the estimation model, the estimation results are usually compared with the identified values ​​of the research subjects. Following this research approach, this invention uses the limb characteristic parameters and surface electromyography data of the subjects collected in the human ankle joint stiffness identification process as inputs to the estimation model, and uses the ankle joint stiffness data identified in the human ankle joint stiffness identification process as a reference standard to evaluate the estimation performance of the model. The experimental scenario of the subjects in the human ankle joint stiffness identification research is as follows: Figure 2 As shown, the subject sits on a table with their feet fixed to the tool end of a collaborative robotic arm containing six rotational joints (the tool end of the six-DOF collaborative robotic arm is positioned in front and below the subject so that the angle between the ankle joint and the lower leg is 90° at rest). The angle of the subject's ankle joint is changed by adjusting the rotation angle of the end joint. Simultaneously, reflective markers are affixed to the subject's toes, heel, ankle, and knee joints. An optical motion capture system collects the three-dimensional positions of these markers to obtain displacement information, which is then processed to obtain the ankle joint angle information. A six-dimensional force / torque sensor fixed to the tool end of the robotic arm collects torque information during ankle joint perturbation. Based on the collected ankle joint angle and torque information, the ankle joint stiffness value is identified using the least squares method.

[0117] like Figure 1 As shown, the present invention provides a method for estimating the stiffness of human joints based on a musculoskeletal dynamics model, which specifically includes the following steps:

[0118] S1: Measure human limb characteristics and surface electromyography (EMG) signals; more specifically, measure the subject's ankle angle, knee angle, lower leg length, and surface EMG signals of the tibialis anterior, medial gastrocnemius, lateral gastrocnemius, and soleus muscles.

[0119] S2: The acquired raw surface electromyography (EMG) signals are subjected to high-pass filtering to remove motion artifacts, and then full-wave rectification and normalization are performed to complete the preprocessing of the EMG signals; the preprocessed surface EMG signals are input into the muscle activation kinetics model, and the muscle activation signals are calculated through the muscle activation kinetics model.

[0120] The muscle activation kinetics model is described in detail below:

[0121] The function of the muscle activation kinetics model is to convert the raw electromyography (EMG) signal into a muscle activation signal. First, the raw EMG signal is high-pass filtered to remove motion artifacts, and then full-wave rectification and normalization are performed to complete the preprocessing of the EMG signal. Next, the neural activation model is used to process the normalized EMG signal to obtain the neural signal. Finally, the muscle activation signal is obtained based on the muscle activation kinetics model.

[0122] The nerve activation signal u of the j-th muscle j (t) and surface electromyography signal e j The relationship between (t) can be expressed as:

[0123] u j (t)=f a (e j (t)) (1)

[0124] Among them, f a This is a neural activation model that characterizes the neural activation signal u. j (t) and surface electromyography signal e j The relationship between (t) and f a It is a direct proportional function with a slope of 1;

[0125] The muscle activation kinetics model can be expressed as follows:

[0126]

[0127] Among them, a j (t) is the activation signal of the j-th muscle, c1+c2 is the activation ratio constant, c2 is the inactivation ratio constant, and c1, c2 and the muscle activation time constant t are related. act and deactivation time constant t deact The following relationship exists:

[0128]

[0129] Where ζ is the muscle activation / deactivation ratio coefficient, taken as ζ = 0.5, and the muscle activation time constant t act Take 15ms;

[0130] Thus far, based on formulas (1) to (3), the surface electromyography signal e is used... j The muscle activation signal a can be calculated from (t). j (t);

[0131] S3: Input the measured human limb feature parameters into the musculoskeletal geometry model, and obtain the microscopic parameters of the muscle structure through the musculoskeletal geometry model;

[0132] The musculoskeletal geometry model is described below:

[0133] Based on the functional relationships between the four muscles—soleus (SOL), medial gastrocnemius (MG), lateral gastrocnemius (LG), and tibialis anterior (TA)—and limb parameters, the corresponding muscle and tendon unit structural parameters are solved using formulas (4), (5), (6), and (7):

[0134]

[0135]

[0136]

[0137]

[0138]

[0139]

[0140]

[0141]

[0142]

[0143] (φ o ) sol =28.3π / 180,(φ o ) mg =9.9π / 180,(φ o ) lg =12π / 180,(φ o ) ta = 9.6π / 180;

[0144] Among them, l mtu l is the length of a muscle-tendon unit. shank Calf length A function relating to the initial ankle angle α0 and the angle change Δα, (φ o ) sol 、(φ o ) mg 、(φ o ) lg 、(φ o ) ta The optimal pennate angles are for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. The optimal fiber lengths for the four muscles mentioned above are respectively... o h is the height of the subject; The resting lengths of the tendon units of the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. This is the resting length of the tendon. These represent the musculoskeletal geometric model coefficients σ for the four muscles mentioned above. 1 / σ 2 .

[0145] Because different muscles have different shapes and positions, the contribution of the muscle-tendon unit to movement during ankle joint motion is affected by its torque arm. The joint angle also changes the magnitude of the muscle torque arm. Specifically, the length l of the muscle-tendon unit... mtu The angle α and torque arm r of the joint mtu The expression for the relationship between them is:

[0146]

[0147] S4: Based on the muscle activation signal obtained in S2 and the microscopic parameters of the muscle structure obtained in S3, the muscle force and muscle-tendon unit force under specific muscle activation are calculated using the muscle contraction dynamics model.

[0148] The muscle contraction dynamics model is described in detail below:

[0149] As a crucial component of musculoskeletal dynamics models, muscle contraction dynamics models function to calculate muscle contractile force based on muscle activation levels, utilizing microscopic parameters of muscle structure and motion characteristics. Based on muscle structure and movement mechanisms, the force F generated by the muscle-tendon unit can be obtained. mtu for:

[0150]

[0151] Among them, F m For muscle strength, Active tension generated by muscles The passive tension generated by the muscle, φ is the angle between the muscle force and the tendon; F mtu For muscle and tendon unit force;

[0152] The active tension generated by the muscle in formula (9) Compared with the current level of muscle activation (a) and normalized muscle fiber length The ratio of fiber speeds and the maximum muscle force generated during isometric muscle contraction. The following relationship exists:

[0153]

[0154] in, This represents the force-length relationship during active muscle contraction. This represents the force-velocity relationship during active muscle contraction. The normalized muscle fiber length, its value is equal to the current muscle fiber length l. m Optimal muscle fiber length at muscle activation intensity 'a' The ratio, The current velocity v of the muscle fiber m With the maximum contraction speed of the muscle The ratio, v m The value is equal to the derivative of the current muscle fiber length lm. The value is equal to 10l o ;

[0155] Force-length relationship of the active muscle contraction unit in formula (10) The specific expression is:

[0156]

[0157] The force-velocity relationship during active muscle contraction in formula (5) The specific expression is:

[0158]

[0159] Based on the rigid tendon assumption, the normalized expression for muscle fiber length, where the current muscle fiber length lm is:

[0160]

[0161] Among them, l o The optimal muscle fiber length, φ, is the length at which the muscle generates maximum contractile force. o For the optimal feather angle of the muscle, l is the resting length of the tendon. mtu The length of a muscle-tendon unit;

[0162] The optimal muscle fiber length for muscle activation intensity 'a' is the same as the optimal muscle fiber length 'l' for muscle to produce maximum contractile force. o The activation intensity 'a' of the muscle has the following relationship:

[0163]

[0164] Furthermore, the angle φ between the muscle force and the tendon in formula (9) can be solved by the following formula:

[0165]

[0166] Among them, l o The optimal muscle fiber length for the muscle to generate maximum contractile force, l mThis represents the current length of the muscle fiber;

[0167] Passive tension generated by muscles With normalized muscle fiber length The maximum muscle force generated during isometric contraction of muscles The following relationship exists:

[0168]

[0169] In formula (16) The passive elastic force-length relationship of muscles is defined as follows:

[0170]

[0171] The maximum contractile force corresponding to the soleus (SOL), medial gastrocnemius (MG), lateral gastrocnemius (LG), and tibialis anterior (TA) muscles. The formula is:

[0172]

[0173] S5: Based on the muscle force obtained in S4 and the microscopic parameters of the muscle structure obtained in S3, the stiffness of the muscle tendon unit is calculated using the muscle tendon unit stiffness model.

[0174] The function of the muscle-tendon unit stiffness model is to calculate the stiffness of the muscle-tendon unit based on muscle force and muscle structural parameters. In this invention, the muscle-tendon unit is considered to be composed of muscle fibers and tendons connected in series. Therefore, the stiffness of the muscle-tendon unit of the j-th muscle is... It is usually expressed as:

[0175]

[0176] in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle;

[0177] Muscle stiffness According to muscle strength F m and the optimal length of its muscle fibers The solution is as follows:

[0178]

[0179] Tendon stiffness K of four muscles t Then solve according to the following expression:

[0180]

[0181] in, and The resting lengths of the tendons of the four muscles are respectively. Solve according to formula (7);

[0182] S6: Based on the muscle structure micro-parameters obtained in S3, the muscle-tendon unit forces obtained in S4, and the muscle-tendon unit stiffness obtained in S5, the joint stiffness is calculated using the muscle-tendon-to-joint mapping model.

[0183] The mapping model from muscles and tendons to joints is described below:

[0184] The function of the muscle-tendon to joint mapping model is to obtain the relationship between joint stiffness and muscle-tendon unit stiffness, so as to facilitate the conversion of muscle-tendon unit stiffness obtained by solving formula (19) into joint stiffness. The modeling process of this mapping relationship is as follows:

[0185] Assume the torque T generated by the j-th muscle at the joint j for:

[0186]

[0187] in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle. Let be the deformation of the tendon unit of the j-th muscle. Let be the torque arm of the j-th muscle tendon unit;

[0188] By definition, the magnitude of joint stiffness K is equal to the derivative of joint torque T with respect to joint angle α, and the contribution K of each muscle or tendon unit at that joint to its stiffness. j The calculation expression is:

[0189]

[0190] The simplified mapping relationship from muscles and tendons to joints obtained from the above equation is as follows:

[0191]

[0192]

[0193] in, Let j be the stiffness of the tendon unit of the j-th muscle. Let α be the torque arm of the j-th muscle / tendon unit, and α be the joint angle. Let be the force of the j-th muscle tendon unit.

[0194] The estimated values ​​of ankle joint stiffness were calculated using the above model. A comprehensive comparison was made between the identified and estimated ankle joint stiffness values ​​at different joint angles under three different muscle activation levels (no tibialis anterior activation, 60% tibialis anterior activation, and 100% tibialis anterior activation). The data were plotted in a graph, and the results are shown below. Figure 3 As shown, the musculoskeletal dynamics model can accurately estimate the stiffness of the human ankle joint.

[0195] On the other hand, the present invention provides a human joint stiffness estimation system based on a musculoskeletal dynamics model, comprising:

[0196] The electromyography (EMG) signal preprocessing module is used to perform high-pass filtering to remove motion artifacts from the acquired raw surface EMG signals, and then perform full-wave rectification and normalization to complete the EMG signal preprocessing process.

[0197] The muscle activation signal calculation module is used to input the preprocessed surface electromyography signal into the muscle activation kinetic model and calculate the muscle activation signal through the muscle activation kinetic model.

[0198] The module for acquiring microscopic parameters of muscle structure is used to input the measured human limb feature parameters into the musculoskeletal geometric model, and to obtain the microscopic parameters of muscle structure through the musculoskeletal geometric model.

[0199] The muscle force calculation module is used to calculate the muscle force and muscle-tendon unit force under muscle activation based on the muscle activation signal and the microscopic parameters of the muscle structure through a muscle contraction dynamics model.

[0200] The muscle and tendon unit stiffness calculation module is used to calculate the stiffness of the muscle and tendon unit based on the muscle force and microscopic parameters of the muscle structure under muscle activation, through the muscle and tendon unit stiffness model.

[0201] The joint stiffness calculation module calculates joint stiffness based on the force of muscle and tendon units under muscle activation, the microscopic parameters of muscle structure, and the stiffness of muscle and tendon units, through a mapping model from muscle and tendon to joint.

[0202] The musculoskeletal dynamics model includes a muscle activation dynamics model, a musculoskeletal geometric model, a muscle contraction dynamics model, a muscle-tendon unit stiffness model, and a muscle-tendon-to-joint mapping model.

[0203] More preferably, the musculoskeletal geometric model is as follows:

[0204]

[0205]

[0206]

[0207]

[0208]

[0209]

[0210]

[0211]

[0212]

[0213]

[0214]

[0215]

[0216]

[0217]

[0218]

[0219] (φ o ) sol =28.3π / 180,(φ o ) mg =9.9π / 180,(φ o ) lg =12π / 180,(φ o ) ta =9.6π / 180 where, l mtu l is the length of a muscle-tendon unit. shank The length of the lower leg. The function relating to the initial angle α0 and angle change Δα of the ankle joint, (φ o ) sol 、(φ o ) mg 、(φ o ) lg 、(φ o ) ta The optimal pennate angles are for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. The optimal fiber lengths for the four muscles mentioned above are respectively... o h is the height of the subject; The resting lengths of the myofascitis units of the four muscles mentioned above. This is the resting length of the tendon. These represent the musculoskeletal geometric model coefficients σ for the four muscles mentioned above. 1 / σ 2 .

[0220] More preferably, the muscle contraction dynamics model is as follows:

[0221]

[0222]

[0223]

[0224]

[0225]

[0226]

[0227]

[0228]

[0229]

[0230]

[0231]

[0232] Among them, F mtu For muscle and tendon unit force; F m For muscle strength, Active tension generated by muscles φ represents the passive tension generated by the muscle, and φ is the angle between the muscle force and the tendon. This represents the force-length relationship during active muscle contraction. This represents the force-velocity relationship during active muscle contraction. The normalized muscle fiber length, its value is equal to the current muscle fiber length l. m Optimal muscle fiber length at muscle activation intensity 'a' The ratio, The current velocity v of the muscle fiber m With the maximum contraction speed of the muscle The ratio, v m The value is equal to the current muscle fiber length l m The derivative, The value is equal to 10l o ;l oThe optimal muscle fiber length, φ, is the length at which the muscle generates maximum contractile force. o For the optimal feather angle of the muscle, l is the resting length of the tendon. mtu The length of a muscle-tendon unit; This represents the maximum muscle force generated during isometric muscle contraction. This represents the passive elastic force-length relationship of the muscle.

[0233] More preferably, the muscle activation kinetics model is as follows:

[0234] u j (t)=f a (e j (t))

[0235]

[0236]

[0237] Among them, u j (t) represents the neural activation signal of the j-th muscle; e j (t) represents the surface electromyography signal; f a For neural activation models; a j (t) is the activation signal of the j-th muscle; c1+c2 is the activation ratio constant, c2 is the inactivation ratio constant, and t deact t is the deactivation time constant; act Let be the muscle activation time constant; ζ is the muscle activation-deactivation proportionality coefficient, taken as ζ = 0.5, and t is the muscle activation time constant. act Take 15ms.

[0238] More preferably, the mapping model from muscle / tendon to joint is as follows:

[0239]

[0240]

[0241]

[0242] in,

[0243]

[0244]

[0245] in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle. Let j be the deformation of the j-th muscle. Let T be the torque arm of the j-th muscle;j K represents the torque exerted by the j-th muscle on the joint under study; j For the contribution of the j-th muscle to the stiffness of the joint, K of the ankle joint j This includes the contributions of four muscles: soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior; K represents joint stiffness. Let be the stiffness of the muscle-tendon unit of the j-th muscle. Let α be the torque arm of the muscle-tendon unit of the j-th muscle, and α be the joint angle. Let be the force of the muscle-tendon unit of the j-th muscle.

[0246] In summary, compared with the prior art, the present invention has the following advantages:

[0247] This invention proposes a method for estimating human joint stiffness based on a musculoskeletal dynamics model. This method can directly obtain joint stiffness through muscle geometric parameters and electromyographic signals, without the need for stiffness identification or stiffness estimation model training, thus overcoming the limitations of traditional stiffness identification or estimation methods. Traditional methods for obtaining muscle micro-parameters often use OpenSim and SIMM models similar to human anatomical models. However, the establishment, application, and data acquisition of these models are relatively complex and computationally intensive, making it difficult to meet the needs of efficient stiffness estimation. Based on existing research on the relationship between human muscle structural parameters and limb characteristics, this invention establishes a musculoskeletal geometric model that conforms to the physiological structure of the ankle joint, enabling accurate and efficient acquisition of muscle micro-parameters. Furthermore, five sub-models are established: a muscle activation dynamics model, a muscle contraction dynamics model, a muscle-tendon unit stiffness model, and a muscle-tendon-to-joint mapping model, constituting a musculoskeletal dynamics model. Ultimately, this achieves the effect of estimating joint stiffness simply by inputting limb parameters and electromyographic signals. Compared to stiffness estimation models based on identification data, this method not only conforms to the principles of neural transmission and the characteristics of human anatomical structure, but also has advantages in terms of the biological interpretability and universality of the model. Furthermore, it has many advantages such as high model estimation accuracy and simple data computation.

[0248] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for estimating the stiffness of human joints based on a musculoskeletal dynamics model, characterized in that, Includes the following steps: D1: The acquired raw surface electromyography (EMG) signals are subjected to high-pass filtering to remove motion artifacts, followed by full-wave rectification and normalization to complete the preprocessing of the EMG signals. D2: Input the preprocessed surface electromyography signal into the muscle activation kinetics model, and calculate the muscle activation signal through the muscle activation kinetics model; D3: Input the measured human limb feature parameters into the musculoskeletal geometry model, and obtain the microscopic parameters of muscle structure through the musculoskeletal geometry model; D4: Based on muscle activation signals and microscopic parameters of muscle structure, muscle force and muscle-tendon unit force under muscle activation are calculated using a muscle contraction dynamics model. D5: The stiffness of the muscle tendon unit is calculated using the muscle force and microscopic parameters of the muscle structure under muscle activation through the muscle tendon unit stiffness model. D6: Based on the force of the muscle-tendon unit under muscle activation, the microscopic parameters of the muscle structure, and the stiffness of the muscle-tendon unit, the joint stiffness is calculated through the mapping model from muscle-tendon to joint. The musculoskeletal dynamics model includes a muscle activation dynamics model, a musculoskeletal geometric model, a muscle contraction dynamics model, a muscle-tendon unit stiffness model, and a muscle-tendon-to-joint mapping model.

2. The method for estimating the stiffness of human joints according to claim 1, characterized in that, The musculoskeletal geometric model is as follows: (f o ) sol =28.3π / 180,(φ o ) mg =9.9π / 180,(φ o ) lg =12π / 180, (f o ) ta =9.6π / 180 Among them, l mtu The length of a muscle-tendon unit, l shank Calf length, A function relating to the initial ankle angle α0 and the angle change Δα, (φ o ) sol 、(φ o ) mg 、(φ o ) lg 、(φ o ) ta The optimal pennate angles are for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. The optimal fiber lengths for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles are respectively. o h is the height of the subject; The resting lengths of the tendon units of the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. This is the resting length of the tendon. The musculoskeletal geometric model coefficients σ represent the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. 1 / σ 2 .

3. The method for estimating the stiffness of human joints according to claim 1 or 2, characterized in that, The muscle contraction dynamics model is as follows: A P =0.129,k pe =4.525 Among them, F mtu For muscle and tendon unit force; F m For muscle strength, Active tension generated by muscles φ represents the passive tension generated by the muscle, and φ is the angle between the muscle force and the tendon. This represents the force-length relationship during active muscle contraction. This represents the force-velocity relationship during active muscle contraction. The normalized muscle fiber length, its value is equal to the current muscle fiber length l. m Optimal muscle fiber length at muscle activation intensity 'a' The ratio, The current velocity v of the muscle fiber m With the maximum contraction speed of the muscle The ratio, v m The value is equal to the current muscle fiber length l m The derivative of The value is equal to 10l o ;l o The optimal muscle fiber length, φ, is the length at which the muscle generates maximum contractile force. o For the optimal feather angle of the muscle, l is the resting length of the tendon. mtu The length of a muscle-tendon unit; This represents the maximum muscle force generated during isometric muscle contraction. This represents the passive elastic force-length relationship of the muscle.

4. The method for estimating the stiffness of human joints according to claim 3, characterized in that, The muscle activation kinetic model is as follows: u j (t)=f a (e j (t)) Among them, u j (t) represents the neural activation signal of the j-th muscle; e j (t) represents the surface electromyography signal; f a For neural activation models; a j (t) is the activation signal of the j-th muscle; c1+c2 is the activation ratio constant, c2 is the inactivation ratio constant, and t deact t is the deactivation time constant; act Let be the muscle activation time constant; ζ is the muscle activation-deactivation proportionality coefficient, taken as ζ = 0.5, and t is the muscle activation time constant. act Take 15ms.

5. The method for estimating the stiffness of human joints according to claim 4, characterized in that, The mapping model from muscles and tendons to joints is as follows: in, in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle. Let j be the deformation of the j-th muscle. Let T be the torque arm of the j-th muscle; j K represents the torque exerted by the j-th muscle on the joint under study; j For the contribution of the j-th muscle to joint stiffness, the K value of the ankle joint... j This includes contributions from the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles; K represents joint stiffness. Let be the stiffness of the muscle-tendon unit of the j-th muscle. Let α be the torque arm of the muscle-tendon unit of the j-th muscle, and α be the joint angle. Let be the force of the muscle-tendon unit of the j-th muscle.

6. A human joint stiffness estimation system based on a musculoskeletal dynamics model, characterized in that, include: The electromyography (EMG) signal preprocessing module is used to perform high-pass filtering to remove motion artifacts from the acquired raw surface EMG signals, and then perform full-wave rectification and normalization to complete the EMG signal preprocessing process. The muscle activation signal calculation module is used to input the preprocessed surface electromyography signal into the muscle activation kinetic model and calculate the muscle activation signal through the muscle activation kinetic model. The module for acquiring microscopic parameters of muscle structure is used to input the measured human limb feature parameters into the musculoskeletal geometric model, and to obtain the microscopic parameters of muscle structure through the musculoskeletal geometric model. The muscle force calculation module is used to calculate the muscle force and muscle-tendon unit force under muscle activation based on the muscle activation signal and the microscopic parameters of the muscle structure through a muscle contraction dynamics model. The muscle and tendon unit stiffness calculation module is used to calculate the stiffness of the muscle and tendon unit based on the muscle force and microscopic parameters of the muscle structure under muscle activation, through the muscle and tendon unit stiffness model. The joint stiffness calculation module calculates joint stiffness based on the force of muscle and tendon units under muscle activation, the microscopic parameters of muscle structure, and the stiffness of muscle and tendon units, through a mapping model from muscle and tendon to joint. The musculoskeletal dynamics model includes a muscle activation dynamics model, a musculoskeletal geometric model, a muscle contraction dynamics model, a muscle-tendon unit stiffness model, and a muscle-tendon-to-joint mapping model.

7. The human joint stiffness estimation system according to claim 6, characterized in that, The musculoskeletal geometric model is as follows: (f o ) sol =28.3π / 180,(φ o ) mg =9.9π / 180,(φ o ) lg =12π / 180, (f o ) ta =9.6π / 180 Among them, l mtu The length of a muscle-tendon unit, l shank Calf length, A function relating to the initial ankle angle α0 and the angle change Δα, (φ o ) sol 、(φ o ) mg 、(φ o ) lg 、(φ o ) ta The optimal pennate angles are for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. The optimal fiber lengths for the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles are respectively. o h represents the height of the subject; The resting lengths of the tendon units of the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. This is the resting length of the tendon. The musculoskeletal geometric model coefficients σ represent the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles, respectively. 1 / σ 2 .

8. The human joint stiffness estimation system according to claim 6 or 7, characterized in that, The muscle contraction dynamics model is as follows: μ=0.15 A P =0.129,k pe =4.525 Among them, F mtu For muscle and tendon unit force; F m For muscle strength, Active tension generated by muscles φ represents the passive tension generated by the muscle, and φ is the angle between the muscle force and the tendon. This represents the force-length relationship during active muscle contraction. This represents the force-velocity relationship during active muscle contraction. The normalized muscle fiber length, its value is equal to the current muscle fiber length l. m Optimal muscle fiber length at muscle activation intensity 'a' The ratio, The current velocity v of the muscle fiber m With the maximum contraction speed of the muscle The ratio, v m The value is equal to the current muscle fiber length l m The derivative of The value is equal to 10l o ;l o The optimal muscle fiber length, φ, is the length at which the muscle generates maximum contractile force. o For the optimal feather angle of the muscle, l is the resting length of the tendon. mtu The length of a muscle-tendon unit; This represents the maximum muscle force generated during isometric muscle contraction. This represents the passive elastic force-length relationship of the muscle.

9. The human joint stiffness estimation system according to claim 8, characterized in that, The muscle activation kinetic model is as follows: u j (t)=f a (e j (t)) Among them, u j (t) represents the neural activation signal of the j-th muscle; e j (t) represents the surface electromyography signal; f a For neural activation models; a j (t) is the activation signal of the j-th muscle; c1+c2 is the activation ratio constant, c2 is the inactivation ratio constant, and t deact t is the deactivation time constant; act Let be the muscle activation time constant; ζ is the muscle activation-deactivation proportionality coefficient, taken as ζ = 0.5, and t is the muscle activation time constant. act Take 15ms.

10. The human joint stiffness estimation system according to claim 9, characterized in that, The mapping model from muscles and tendons to joints is as follows: in, in, Let be the muscle stiffness of the j-th muscle. Let be the tendon stiffness of the j-th muscle. Let j be the deformation of the j-th muscle. Let T be the torque arm of the j-th muscle; j K represents the torque exerted by the j-th muscle on the joint under study; j For the contribution of the j-th muscle to joint stiffness, the K value of the ankle joint... j This includes contributions from the soleus, medial gastrocnemius, lateral gastrocnemius, and tibialis anterior muscles; K represents joint stiffness. Let be the stiffness of the muscle-tendon unit of the j-th muscle. Let α be the torque arm of the muscle-tendon unit of the j-th muscle, and α be the joint angle. Let be the force of the muscle-tendon unit of the j-th muscle.

Images