CN112294599B - Method, system and device for constructing training trajectory generation model based on human parameters - Google Patents

Abstract

The invention belongs to the field of computers, and in particular relates to a method, system and device for constructing a training trajectory generation model based on human parameters, and aims to solve the problem of a single gait training trajectory of an existing lower limb rehabilitation robot. The method of the invention includes constructing a training sample set based on the input feature set and the Fourier coefficient set; based on the training sample set, training a plurality of classification regression models for preset joints respectively, and selecting the regression model with the smallest prediction error as the training sample set. The angle generation model of the corresponding joint; the obtained angle models of multiple joints are combined to obtain the training trajectory generation model of the human body part including the preset joints. The training trajectory generation model constructed by the method of the present invention can generate the differentiated training trajectory based on the user's specific human body parameters.

Description

Method, system and device for constructing training trajectory generation model based on human parameters

technical field

The invention belongs to the field of computers, and in particular relates to a method, system and device for constructing a training trajectory generation model based on human body parameters.

Background technique

At present, thanks to the development of modern medicine, the fatality rate of stroke, traumatic brain injury, spinal cord injury and other diseases has been greatly reduced; however, after acute treatment, most patients have sequelae such as lower extremity motor dysfunction and require long-term rehabilitation Training, which is mainly based on gait rehabilitation training. In traditional gait rehabilitation training, multiple rehabilitation therapists are often required to complete it together. Under the current shortage of domestic rehabilitation therapist resources, lower limb rehabilitation robots are often used in patients' rehabilitation training to assist or actively drive patients to perform gait training. The gait trajectory adopted by the lower limb rehabilitation robot is mainly the angular trajectory of the three joints of the hip, knee and ankle in the sagittal plane of the human body. The gait trajectory data of normal people are collected, and then used on the robot as a template for gait rehabilitation training of patients. However, studies have shown that there are individual differences in gait trajectories, with strong correlations with human characteristics such as gender, age, and height. According to the characteristics of the patient, providing a gait trajectory training more in line with the patient through the lower limb rehabilitation robot will promote the improvement of the patient's rehabilitation effect.

The invention proposes a method for generating a personalized gait trajectory based on human body parameters. The three-joint angle curves of hip, knee and ankle are fitted by Fourier series, and the fitted Fourier coefficient is used to replace the gait trajectory, and It is applied to the lower limb rehabilitation robot; based on 14 human parameters, the relationship between human parameters and Fourier coefficients is modeled by machine learning, and then a personalized gait trajectory generation model based on human parameters is constructed.

SUMMARY OF THE INVENTION

In order to solve the above problems in the prior art, that is, in order to solve the problem of a single gait training trajectory of the existing lower limb rehabilitation robot, the first aspect of the present invention proposes a method for building a training trajectory generation model based on human parameters, including the following steps :

Build a training sample set based on the input feature set and Fourier coefficient set;

Based on the training sample set, the preset joints are respectively trained for multiple categories of regression models, and the regression model with the smallest prediction error is selected as the angle generation model of the corresponding joint;

Combining the obtained angle models of a plurality of joints to obtain a training trajectory generation model of the human body part including the preset joints;

in,

The input feature set includes human body feature category parameters of a plurality of human bodies;

The set of Fourier coefficients is the coefficient term in the joint angle function obtained by fitting the joint angle function based on the test data of the preset joint;

The preset joints are the joints included in the human body part of the trajectory to be generated.

In some preferred embodiments, the input feature set, the acquisition method is:

Based on the initial sample set, the maximum correlation minimum redundancy method is adopted to obtain the input feature set; each sample in the initial sample set includes an individual parameter corresponding to a preset human body feature category.

In some preferred embodiments, "using the maximum correlation minimum redundancy method to select the input feature set", the method is:

Step S110, for the set of Fourier coefficients, based on preset human body characteristics, obtain a feature ranking corresponding to each Fourier coefficient through mutual information calculation;

Step S120, rank the features corresponding to each Fourier coefficient, and obtain the final feature ranking by calculating the mean value of the serial numbers;

Step S130, based on the final feature sorting, sequentially adding preset human features as a candidate input feature set, respectively modeling with the Fourier coefficient set, obtaining an intermediate model, and selecting the corresponding intermediate model with the smallest mean error. The set of alternative input features for is the selected set of input features.

In some preferred embodiments, "for the set of Fourier coefficients, based on preset human body characteristics, the feature ranking corresponding to each Fourier coefficient is obtained through mutual information calculation", and the method is:

Step S111, select a Fourier coefficient from the Fourier coefficient set;

Step S112, for a single Fourier coefficient, calculate its mutual information with each preset human body feature in turn, put the preset human body feature corresponding to the maximum mutual information into feature set D ₁ , and put the rest into feature set D ₂ ;

Step S113, for each feature in the feature set D2, calculate the average value _of the mutual information with each feature in the feature set D1 _;

Step S114, select the feature in the feature set D2 corresponding to the maximum mutual information average value in step S113, and move it into the original feature sequence _of the feature set D1 _;

Step _S115 , perform steps S112 to S114, until the feature set D2 is empty, generate the feature ranking corresponding to the corresponding Fourier coefficient;

Step S116, select one of the remaining Fourier coefficients in the set of Fourier coefficients, and perform step S112 until all the Fourier coefficients in the set of Fourier coefficients have obtained corresponding feature rankings.

In some preferred embodiments, the part of the human body containing the preset joints is the lower limbs of the human body; the preset joints include hip joints, knee joints, and ankle joints.

In some preferred embodiments, the training trajectory generation model further includes a center of gravity calculation module corresponding to the joint angle;

The center of gravity calculation module is configured to calculate the center of gravity based on the angle of the hip joint, the angle of the knee joint, and the angle of the ankle joint, using the geometric relationship between the limb and the joint.

In some preferred embodiments, based on the test data of the preset joints, the joint angle function f(t) obtained by fitting the Fourier series method is

为角频率，T为步态周期，a₀,a_i,b_i(i＝1,...,n)为系数项。where n is the order of fitting,

is the angular frequency, T is the gait period, and a ₀ , a _i , b _i (i=1,...,n) are the coefficient terms.

In some preferred embodiments, the plurality of class regression models include support vector machine regression models and random forest regression models.

In a second aspect of the present invention, a training trajectory generation model building system based on human parameters is proposed, including a first unit, a second unit, and a third unit;

The first unit is configured to construct a training sample set based on the input feature set and the Fourier coefficient set;

The second unit is configured to, based on the training sample set, respectively perform training of multiple classification regression models on the preset joints, and select the regression model with the smallest prediction error as the angle generation model of the corresponding joint;

The third unit is configured to combine the obtained angle models of a plurality of joints to obtain a training trajectory generation model of a human body part including preset joints;

in,

The input feature set includes human body feature category parameters of a plurality of human bodies;

The set of Fourier coefficients is the coefficient term in the joint angle function obtained by fitting the joint angle function based on the test data of the preset joint;

The preset joints are the joints included in the human body part of the trajectory to be generated.

In a third aspect of the present invention, a processing device is provided, including a processor and a storage device; the processor is adapted to execute various programs; the storage device is adapted to store multiple programs; the programs are adapted to be loaded by the processor And execute it to realize the above-mentioned method for building a model for training trajectory generation based on human parameters.

Beneficial effects of the present invention:

The training trajectory generation model constructed based on the method of the present invention can generate differentiated training trajectories based on the user's specific human body parameters, the generated training trajectory is more suitable for the user's human body condition, can improve the user's training effect, and further Improve the user's motor function.

Description of drawings

Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

1 is a schematic flowchart of a method for constructing a training trajectory generation model based on human parameters according to an embodiment of the present invention;

Fig. 2 is a schematic diagram of the angle of lower limb joints in an embodiment of the present invention;

3 is a schematic diagram of a simplified model of a lower limb according to an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not All examples. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the related invention, but not to limit the invention. In addition, it should be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.

It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict.

A method for constructing a training trajectory generation model based on human parameters of the present invention, as shown in Figure 1, includes the following steps:

Build a training sample set based on the input feature set and Fourier coefficient set;

Based on the training sample set, the preset joints are respectively trained for multiple categories of regression models, and the regression model with the smallest prediction error is selected as the angle generation model of the corresponding joint;

Combining the obtained angle models of a plurality of joints to obtain a training trajectory generation model of the human body part including the preset joints;

in,

The input feature set includes human body feature category parameters of a plurality of human bodies;

The set of Fourier coefficients is the coefficient term in the joint angle function obtained by fitting the joint angle function based on the test data of the preset joint;

The preset joints are the joints included in the human body part of the trajectory to be generated.

In order to describe the present invention more clearly, each part of an embodiment of the present invention will be described in detail below with reference to the accompanying drawings.

The method for constructing a training trajectory generation model based on human body parameters according to an embodiment of the present invention takes the lower limbs of the human body as an example for description, and of course, it can also be applied to the modeling of training trajectory of other human body parts, such as upper limbs.

Before the description of the specific steps in this embodiment, the collection and fitting of the gait trajectory of the lower limbs of the human body will be described first.

Figure 2 shows the schematic diagram of the lower limbs of the human body, respectively defining the hip joint angle, knee joint angle and ankle joint angle in the gait trajectory. The hip joint angle is the angle between the center of gravity curve (l ₁ ) and the line connecting the center of the hip and knee joint (l ₂ ), namely θ _hip in Figure 2; the knee joint angle is the angle between l ₂ and the line connecting the center of the knee-ankle joint (l ₃ ). The included angle is θ _knee in Figure 2; the ankle joint angle θ _ankle is the included angle between the vertical line (l ₄ ) of l ₃ at the center of the ankle joint and the parallel line (l ₅ ) of the foot surface.

Use the joint angle data measuring instrument to collect the angle data of the three joints when a normal person walks at different speeds (v ₁ ... _vi ). For the collected angle data, firstly find the missing values in the data, and fill in the missing values with the average value of the 5 data points before and after; then use the moving average method to filter and smooth the angle curve. Since the acquisition process involves data of multiple gait cycles, it is necessary to divide the data into cycles and mark each cycle. The starting point of the gait cycle is the moment when one heel just touches the ground.

The processed three-joint angle curve is composed of hundreds of discrete sampling points. If the discrete gait trajectory data points are used directly, there will be two problems: on the one hand, it will increase the difficulty of modeling and predict hundreds of discrete points. It will lead to the high complexity of the trajectory generation model; on the other hand, it is not conducive to the motor control at the joints of the lower limb rehabilitation robot.

In the present invention, using Fourier series method (1) to fit joint angle data, a function f(t) related to time t can be obtained, as shown in formula (1):

为角频率，T为步态周期，a₀,a_i,b_i(i＝1,...,n)为系数项。where n is the order of fitting,

is the angular frequency, T is the gait period, and a ₀ , a _i , b _i (i=1,...,n) are the coefficient terms.

The choice of order n is determined according to the absolute error value between the fitted curve and the original curve. Through experimental testing, when the fitting order is 5, the error value decreases gradually, and the fitting errors of the three joints are all Within 0.3 degrees, considering the error and the complexity of the fitting model, n=5 is finally selected as the final fitting order. Therefore, after fitting each joint trajectory, 12 Fourier coefficients (ω, a ₀ , a _i , b _i (i=1,...,5)) can be obtained, and these coefficients are used as the surrogate for the trajectory value, participating in the modeling of personalized gait trajectory. Furthermore, as long as 12 parameters are generated according to human characteristics, the joint angle trajectory can be reconstructed. In the joint motor control of the lower limb rehabilitation robot, the velocity curve and acceleration curve of the trajectory are obtained by derivation of the reconstructed joint trajectory equation, so a variety of motor control methods (position control and speed control) can be selected.

The method embodiment of the present invention is a preferred embodiment of the method of the present invention, comprising the following steps:

Step S100, based on the initial sample set, using the maximum correlation minimum redundancy method to obtain the input feature set.

Each sample in the initial sample set includes a parameter of an individual corresponding to a preset human body feature category. For this embodiment, by analyzing the anatomy of the lower limbs of the human body and investigating related data, 14 human body characteristics (preset human body characteristics) can be selected to model the gait trajectory, and the characteristics mainly include: gender, age, height, weight , ilium width, trochanter width on both sides of the hip, anterior superior ilium spine width, thigh length, calf length, knee diameter, ankle height, ankle width, foot length and foot width. The human body data is collected based on the preset human body characteristics, and a set of data corresponding to each human body is used as a sample to construct a first sample set.

Based on the method of the present invention, the input feature set can be the above-mentioned 14 human body features, but the input features are too many, which brings the complexity of the system and the tediousness of user data collection. The present invention reconsidered from two aspects: one is that there may be a large correlation between the 14 features, such as the width of the ilium and the width of the anterior superior spine of the ilium; the other is to simplify the feature set and reduce the measurement feature time. In the embodiment of the present invention, the maximum correlation minimum redundancy criterion (mRMR) is used to optimize the feature set, and the redundancy between the features and the correlation between the features and the Fourier coefficients are represented by mutual information, and the mutual information I ( The calculation method of X, Y) is shown in formula (2),

Among them, X and Y are two feature attributes, x and y are the feature attribute values of the sample, p(x) and p(y) are the marginal probability distribution, and p(x, y) is the joint probability distribution.

Using the maximum correlation minimum redundancy method to select the input feature set can be achieved by the following steps:

Step S110, for the set of Fourier coefficients, based on preset human body characteristics, obtain a feature ranking corresponding to each Fourier coefficient through mutual information calculation.

The step further includes:

Step S111, select one Fourier coefficient from the set of Fourier coefficients.

The coefficient items in the Fourier coefficient set can be sorted, and then the operations of the following steps are performed in sequence, for example, the coefficient item with the first ranking is selected for the first time; in step S116, the selection of the subsequent coefficient items is performed in turn. .

Step S112, for a single Fourier coefficient, calculate its mutual information with each preset human body feature in turn, put the preset human body feature corresponding to the maximum mutual information into feature set D ₁ , and the rest into feature set D ₂ .

The preset human body features put into the feature set D ₂ may be arranged in descending order according to the size of mutual information, or may be randomly arranged out of order.

Step _S113 , for each feature in the feature set D2, calculate the average value _of the mutual information with each feature in the feature set D1.

In this step, a feature can be taken out from D ₂ in turn, and then for the selected feature, the mutual information between it and each feature in the feature set D ₁ is calculated respectively, and then the mean value is calculated to obtain the score of the feature score _i , Its calculation formula is shown in formula (3),

Among them, i=1,...,n ₁ (n ₁ is the number of features in D ₁ ), n ₂ is the number of features in D ₂ , F represents the feature, and Y represents the Fourier coefficient.

_In step S114, the feature in the feature set D2 corresponding to the maximum mutual information average value in step S113 is selected and moved into the original feature sequence _of the feature set D1.

According to the score of each feature in the feature set D ₂ calculated in step S113, select the feature with the largest score, insert it into the back of the original feature sequence of the feature set D ₁ , and update the feature set D ₁ . ₂ , delete the feature, and update the feature set D ₂ .

Step _S115 : Steps S112 to S114 are executed until the feature set D2 is empty, and the feature ranking corresponding to the corresponding Fourier coefficient is generated.

Step S116, select one of the remaining Fourier coefficients in the set of Fourier coefficients, and perform step S112 until all the Fourier coefficients in the set of Fourier coefficients have obtained corresponding feature rankings.

Execute to step, each Fourier coefficient will correspond to a feature ranking. In this embodiment, 12 Fourier coefficients correspond to 12 feature rankings.

Step S120: Rank the features corresponding to each Fourier coefficient, and obtain the final feature ranking by calculating the mean value of the serial numbers.

In the 12 feature sorting, any feature corresponds to 12 serial numbers, and the average value of the 12 serial numbers can be calculated to obtain its serial number in the final feature sorting. If the calculated values of the mean values of the two features are the same, according to the preset sorting of the coefficient items of the Fourier coefficient set, select and sort the two features in the feature sorting corresponding to the first coefficient item (or other specified coefficient items). The order of the two features in the final feature sorting is determined before and after the sorting.

Step S130, based on the final feature sorting, sequentially adding preset human features as a candidate input feature set, respectively modeling with the Fourier coefficient set, obtaining an intermediate model, and selecting the corresponding intermediate model with the smallest mean error. The set of alternative input features for is the selected set of input features.

Modeling is performed based on the set of candidate input features and the set of Fourier coefficients, and the method is preferably performed by the following methods of steps S200 and S300.

The selection method of the alternative input feature set is: from the final feature ranking, the first time the first ranked feature is modeled with the Fourier coefficient set, and the second time the second ranked feature is added, and then the Fourier coefficient set is added. The coefficient set is modeled, and so on, and each modeling adds a feature in turn to construct an alternative input feature set.

For models constructed separately from different candidate input feature sets, this embodiment uses 5-fold cross-validation to calculate the mean error of the corresponding model, and then selects the candidate input feature set corresponding to the intermediate model with the smallest mean error as the selected input feature set.

After testing, when the candidate input feature set in this implementation includes the first six features, the error of joint angle reconstruction is the smallest. Therefore, this embodiment selects the first six features of the final feature ranking as the input feature set.

Step S200, the construction of the training trajectory generation model.

Based on the input feature set and the Fourier coefficient set, a training sample set is constructed; based on the training sample set, the preset joints are respectively trained for multiple categories of regression models, and the regression model with the smallest prediction error is selected as the angle of the corresponding joint generating a model; combining the obtained angle models of a plurality of joints to obtain a training trajectory generation model of human body parts including preset joints; the input feature set includes human body feature category parameters of a plurality of human bodies; the Fourier coefficients The set is the coefficient term in the joint angle function obtained by fitting through the Fourier series method based on the test data of the preset joint; the preset joint is the joint included in the human body part of the trajectory to be generated. In this embodiment, the part of the human body including the preset joints is the lower limbs of the human body; the preset joints include hip joints, knee joints, and ankle joints.

In this embodiment, two regression modeling methods are adopted, namely support vector machine regression model and random forest regression model. The input of both models is the optimized feature subset (6 features in the input feature set), and the output is 12 Fourier coefficients (ω, a ₀ , a _i , b _i (i=1,... ,5)).

Based on the input feature set and the Fourier coefficient set, a training sample set and a test sample set are constructed. Through the training sample set, two regression models are used to train three joints at the same time, and each joint gets two trained models, and then the prediction error is calculated using the test sample set. Among the models of , respectively, a model with the smallest prediction error is selected as the angle generation model of the corresponding joint, and the final training trajectory generation model is constructed based on the obtained angle generation models of the three joints. For this embodiment, it can be more appropriately named as step state trajectory generation model.

Based on the obtained gait trajectory generation model, when there is a new patient, only 6 human characteristics need to be measured, and the corresponding Fourier coefficients are generated by establishing a personalized gait trajectory generation model, and then the gait is reconstructed. trajectory.

Step S300, dynamic center of gravity adjustment.

In this embodiment, a center of gravity calculation module corresponding to the joint angle is also added to the gait trajectory generation model. The center of gravity calculation module is configured to calculate the center of gravity based on the angle of the hip joint, the angle of the knee joint, and the angle of the ankle joint, using the geometric relationship between the limb and the joint.

When the human body is walking, the center of gravity will also change with the change of the joint angle. In the process of generating personalized gait trajectory, it is very important to provide the change curve of the center of gravity corresponding to the change of joint angle. Especially in the patient's gait rehabilitation training, the change of the center of gravity affects the perception of gait. In the present invention, a simplified human lower limb model is constructed, and the trajectory curve of the center of gravity is calculated based on the existing three joint angles of the hip, knee and ankle, using the geometric relationship between the limbs and the joints.

The simplified model of human lower limbs is shown in Figure 3, in which the thigh and calf are simplified as two links, and the foot is simplified as a triangle (the inner angle near the ankle is θ _foot ). Taking the lowest point of the heel as the contact point with the ground, the height h of the center of gravity consists of three parts: h ₁ , h ₂ , and h ₃ , as shown in equations (4)-(7),

h=h ₁ +h ₂ +h ₃ (4)

h ₁ =l _thigh ·sin(θ _hip ) (5)

h ₂ =l _calf ·sin(θ _hip -θ _knee ) (6)

Among them, l _thigh is the length of the thigh, l _calf is the length of the calf, l _calf is the length of the heel, θ _hip is the hip joint angle, θ _knee is the knee joint angle, θ _ankle is the ankle joint angle, and θ _foot is the foot after simplifying the foot into a triangle. The inner corner of the triangle near the ankle.

By substituting the angle data of the three joints of the hip, knee, and ankle in a gait cycle into formula (4), the periodic curve of the center of gravity can be calculated.

A training trajectory generation model building system based on human parameters according to the second embodiment of the present invention includes a first unit, a second unit, and a third unit;

The first unit is configured to construct a training sample set based on the input feature set and the Fourier coefficient set;

The second unit is configured to, based on the training sample set, respectively perform training of multiple classification regression models on the preset joints, and select the regression model with the smallest prediction error as the angle generation model of the corresponding joint;

The third unit is configured to combine the obtained angle models of a plurality of joints to obtain a training trajectory generation model of a human body part including preset joints;

in,

The input feature set includes human body feature category parameters of a plurality of human bodies;

The set of Fourier coefficients is the coefficient term in the joint angle function obtained by fitting the joint angle function based on the test data of the preset joint;

The preset joints are the joints included in the human body part of the trajectory to be generated.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, for the specific working process and related description of the system described above, reference may be made to the corresponding process in the foregoing method embodiments, which will not be repeated here.

It should be noted that the system for building a model for training trajectory generation based on human body parameters provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functions as required. module, that is, the modules or steps in the embodiments of the present invention are decomposed or combined. For example, the modules in the above embodiments can be combined into one module, or can be further split into multiple sub-modules to complete all or part of the above description. Function. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing each module or step, and should not be regarded as an improper limitation of the present invention.

A storage device according to a third embodiment of the present invention stores a plurality of programs, and the programs are adapted to be loaded and executed by a processor to implement the above-mentioned method for constructing a training trajectory generation model based on human parameters.

A processing device according to a fourth embodiment of the present invention includes a processor and a storage device; the processor is adapted to execute various programs; the storage device is adapted to store multiple programs; the programs are adapted to be loaded and executed by the processor In order to realize the above-mentioned construction method of training trajectory generation model based on human parameters.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process and relevant description of the storage device and processing device described above can refer to the corresponding process in the foregoing method embodiments, which is not repeated here. Repeat.

In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the method illustrated in the flowchart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication portion, and/or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), the above-mentioned functions defined in the method of the present application are performed. It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In this application, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for performing the operations of the present application may be written in one or more programming languages, including object-oriented programming languages—such as Java, Smalltalk, C++, but also conventional Procedural programming language - such as the "C" language or similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (eg, using an Internet service provider through Internet connection).

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more logical functions for implementing the specified functions executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can be implemented in dedicated hardware-based systems that perform the specified functions or operations , or can be implemented in a combination of dedicated hardware and computer instructions.

The terms "first," "second," etc. are used to distinguish between similar objects, and are not used to describe or indicate a particular order or sequence.

The term "comprising" or any other similar term is intended to encompass a non-exclusive inclusion such that a process, method, article or device/means comprising a list of elements includes not only those elements but also other elements not expressly listed, or Also included are elements inherent to these processes, methods, articles or devices/devices.

So far, the technical solutions of the present invention have been described with reference to the preferred embodiments shown in the accompanying drawings, however, those skilled in the art can easily understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims (7)

1. a kind of training track generation model building method based on human body parameters, is characterized in that, comprises the following steps: Build a training sample set based on the input feature set and Fourier coefficient set; The input feature set, its acquisition method is: Based on the initial sample set, the maximum correlation minimum redundancy method is used to obtain the input feature set; the specific method is: Step S110, for the set of Fourier coefficients, based on preset human body characteristics, obtain a feature ranking corresponding to each Fourier coefficient through mutual information calculation; the preset human body characteristics include: gender, age, height, weight, Ilium width, trochanter width on both sides of hip, width of anterior superior ilium spine, thigh length, calf length, knee diameter, ankle height, ankle width, foot length and foot width; specifically: Step S111, select a Fourier coefficient from the Fourier coefficient set; Step S112, for a single Fourier coefficient, calculate its mutual information with each preset human body feature in turn, put the preset human body feature corresponding to the maximum mutual information into feature set D ₁ , and put the rest into feature set D ₂ ; Step S113, for each feature in the feature set D2, calculate the average value _of the mutual information with each feature in the feature set D1 _; Step S114, select the feature in the feature set D2 corresponding to the maximum mutual information average value in step S113, and move it into the original feature sequence _of the feature set D1 _; Step _S115 , perform steps S112 to S114, until the feature set D2 is empty, generate the feature ranking corresponding to the corresponding Fourier coefficient; Step S116, select one from the remaining Fourier coefficients of the Fourier coefficient set, and perform step S112, until all the Fourier coefficients in the Fourier coefficient set have obtained corresponding feature sorting; Step S120, rank the features corresponding to each Fourier coefficient, and obtain the final feature ranking by calculating the mean value of the serial numbers; Step S130, based on the final feature sorting, sequentially adding preset human features as a candidate input feature set, respectively modeling with the Fourier coefficient set, obtaining an intermediate model, and selecting the corresponding intermediate model with the smallest mean error. The candidate input feature set of , as the selected input feature set; Each sample in the initial sample set includes a parameter corresponding to a preset human body feature category of an individual; Based on the training sample set, the preset joints are respectively trained for multiple categories of regression models, and the regression model with the smallest prediction error is selected as the angle generation model of the corresponding joint; Combining the obtained angle generation models of multiple joints to obtain a training trajectory generation model of human body parts including preset joints; in, The input feature set includes preset human body features of a plurality of human bodies; The set of Fourier coefficients is the coefficient term in the joint angle function obtained by fitting the joint angle function based on the test data of the preset joint; The preset joints are the joints included in the human body part of the trajectory to be generated. 2 . The method for constructing a training trajectory generation model based on human parameters according to claim 1 , wherein the human body part comprising preset joints is a human lower limb; and the preset joints include hip joints, knee joints, ankle joints, and ankle joints. joint. 3. The method for constructing a training trajectory generation model based on human body parameters according to claim 2, wherein the training trajectory generation model further comprises a center of gravity calculation module corresponding to the joint angle; The center of gravity calculation module is configured to calculate the center of gravity based on the angle of the hip joint, the angle of the knee joint, and the angle of the ankle joint, using the geometric relationship between the limb and the joint. 4. The method for building a model based on a training trajectory based on human parameters according to claim 2, wherein, based on the test data of the preset joints, the joint angle function f(t) obtained by Fourier series fitting is obtained for

where n is the order of fitting,

is the angular frequency, T is the gait period, and a ₀ , a _i , b _i (i=1,...,n) are the coefficient terms. 5 . The method for constructing a training trajectory generation model based on human parameters according to claim 1 , wherein the multiple category regression models include a support vector machine regression model and a random forest regression model. 6 . 6. A training trajectory generation model building system based on human parameters, characterized in that it comprises a first unit, a second unit, and a third unit; The first unit is configured to construct a training sample set based on the input feature set and the Fourier coefficient set; The input feature set, its acquisition method is: Based on the initial sample set, the maximum correlation minimum redundancy method is used to obtain the input feature set; the specific method is: Step S110, for the set of Fourier coefficients, based on preset human body characteristics, obtain a feature ranking corresponding to each Fourier coefficient through mutual information calculation; the preset human body characteristics include: gender, age, height, weight, Width of ilium, width of trochanter on both sides of hip, width of anterior superior spine of ilium, thigh length, calf length, knee diameter, ankle height, ankle width, foot length and foot width; specifically: Step S111, select a Fourier coefficient from the Fourier coefficient set; Step S112, for a single Fourier coefficient, calculate its mutual information with each preset human body feature in turn, put the preset human body feature corresponding to the maximum mutual information into feature set D ₁ , and put the rest into feature set D ₂ ; Step S113, for each feature in the feature set D2, calculate the average value _of the mutual information with each feature in the feature set D1 _; Step S114, select the feature in the feature set D2 corresponding to the maximum mutual information average value in step S113, and move it into the original feature sequence _of the feature set D1 _; Step _S115 , perform steps S112 to S114, until the feature set D2 is empty, generate the feature ranking corresponding to the corresponding Fourier coefficient; Step S116, select one from the remaining Fourier coefficients in the Fourier coefficient set, and perform step S112, until all the Fourier coefficients in the Fourier coefficient set have obtained corresponding feature sorting; Step S120, rank the features corresponding to each Fourier coefficient, and obtain the final feature ranking by calculating the mean value of the serial numbers; Step S130, based on the final feature sorting, sequentially adding preset human body features as a candidate input feature set, respectively modeling with the Fourier coefficient set, obtaining an intermediate model, and selecting the corresponding intermediate model with the smallest mean error. The candidate input feature set of , as the selected input feature set; Each sample in the initial sample set includes a parameter corresponding to a preset human body feature category of an individual; The second unit is configured to, based on the training sample set, respectively perform training of multiple classification regression models on the preset joints, and select the regression model with the smallest prediction error as the angle generation model of the corresponding joint; The third unit is configured to combine the obtained angle generation models of a plurality of joints to obtain a training trajectory generation model of human body parts including preset joints; in, The input feature set includes preset human body features of a plurality of human bodies; The set of Fourier coefficients is the coefficient term in the joint angle function obtained by fitting the joint angle function based on the test data of the preset joint; The preset joints are the joints included in the human body part of the trajectory to be generated. 7. A processing device, comprising a processor and a storage device; the processor is adapted to execute various programs; the storage device is adapted to store a plurality of programs; characterized in that the programs are adapted to be loaded and executed by the processor to The method for constructing a training trajectory generation model based on human parameters according to any one of claims 1 to 5 is implemented.

Images