Astronaut digital and intelligent twin training method and system based on multi-dimensional data fusion

Abstract

The invention discloses an astronaut digital and intelligent twin training method and system based on multi-dimensional data fusion, and the method comprises the steps: carrying out the data collection of an analysis sample through a miniaturized multi-modal sensor network, and obtaining multi-source heterogeneous data; carrying out abnormal value and missing value processing through a three-stage outlier detection unit to obtain a standardized input data set; constructing an astronaut ergonomics standardization model based on the hierarchical structure and the metadata model; constructing an atomic action library through integration; the task load is evaluated and calculated through a VACP model; performing heaven and earth difference analysis on the atomic action to generate an ergonomics evaluation result; and according to an ergonomics evaluation result and time difference analysis of the atomic action library, dynamically adjusting a task sequence and resource configuration, and optimizing an astronaut training process. The problems that in existing astronaut training and task evaluation, data standardization is insufficient, man-machine work efficiency evaluation is inaccurate in a microgravity environment, and biomechanical analysis of a bone-muscle system is affected by man-servo coupling are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of astronaut training and mission assessment, and particularly to an astronaut digital twin training method and system based on multi-dimensional data fusion. Background Art

[0002] Current astronaut training technologies have integrated means such as physical simulation, virtual reality, and computer simulation to construct multi-modal training scenarios to improve operation proficiency and emergency response capabilities. With the digital and intelligent transformation, training systems based on digital twins have been gradually popularized, realizing the visualization of the training process and data-driven optimization through virtual-real mapping. At the same time, multi-dimensional data fusion technologies have begun to integrate physiological, behavioral, and environmental data to support the design of personalized training programs.

[0003] However, the existing technologies still have significant deficiencies: First, there is a lack of a deep association and dynamic fusion mechanism for multi-source heterogeneous data (such as physiological signals, operation instructions, equipment status), resulting in one-sided training evaluation; Second, most digital twin systems are statically modeled and are difficult to synchronize the changes of astronaut status and environmental variables in real time, with insufficient dynamic adaptability; Third, the training program relies on the iteration of historical experience and lacks the predictive deduction and risk warning capabilities based on real-time data; Fourth, the reuse rate of training data across tasks and scenarios is low, and the scalability and generalization performance of the system are limited, restricting the training efficiency and cost optimization.

[0004] Therefore, how to invent an astronaut digital twin training method to solve problems such as insufficient data standardization, inaccurate man-machine ergonomics evaluation in the microgravity environment, and the influence of human-garment coupling on the biomechanics analysis of the musculoskeletal system in existing astronaut training and mission assessment has become an urgent problem to be solved. Summary of the Invention

[0005] To achieve the above object, the present invention provides the following technical solutions: An astronaut digital twin training method based on multi-dimensional data fusion, including:

[0006] To achieve the above object, the present invention provides the following technical solutions: An astronaut digital twin training method based on multi-dimensional data fusion, including:

[0007] Divide the types of space missions and the characteristic groups of astronauts through a stratified sampling strategy to obtain the space mission types at the set levels and the characteristic groups of astronauts at the set levels; calculate the sample size through a statistical strategy; randomly select from the characteristic groups of astronauts at the set levels according to the sample size to obtain an analysis sample; collect data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data;

[0008] Process the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate the dimension difference and obtain consistent data; compress the consistent data through PCA dimensionality reduction technology to obtain compressed data; construct a Mandel-Cochran-Grubbs three-level outlier detection unit; process the compressed data for outliers and missing values through the three-level outlier detection unit to obtain a standardized input data set;

[0009] Based on the standardized input data set, construct a space ergonomics standardization model based on a hierarchical structure and a metadata model;

[0010] Based on the space ergonomics standardization model, disassemble the astronaut's operation trajectory into atomic actions according to the human terminal; integrate the atomic actions, attributes, interaction relationships, and logical hierarchical relationships to construct an atomic action library;

[0011] Based on the atomic action library, evaluate and calculate the task load through the VACP model to obtain a task load index; generate an ergonomics evaluation result through the analysis of the differences between space and ground of the atomic actions;

[0012] According to the ergonomics evaluation result and the time difference analysis of the atomic action library, dynamically adjust the task sequence and resource allocation to optimize the astronaut training process.

[0013] As an optimal solution of an astronaut digital twin training method based on multi-dimensional data fusion, in the process of constructing a space ergonomics standardization model based on a hierarchical structure and a metadata model, it includes: constructing a virtual human upper limb bone and muscle model on the OpenSim platform, and setting the body segment mass and centroid parameters of the upper limb joints in the virtual human upper limb bone and muscle model; representing the upper limb force-generating muscle groups through Hill-type muscle-tendon descriptions; defining the muscle attribute characteristics of the muscle-tendon through the Schutte dynamic muscle model; constructing an upper limb joint fatigue model based on the body segment mass, the centroid parameters, and the muscle attribute characteristics of the upper limb joints; simulating the muscle fatigue accumulation effect through the upper limb joint fatigue model;

[0014] The expression of the upper limb joint fatigue model is:

[0015] The expression of the upper limb joint fatigue model is:

[0016]

[0017] where Γ cem (t) is the maximum joint force application capacity of the joint in the state at time t; C is the integral constant; e is the natural exponential function; k is the coefficient; n is the co - contraction factor; t0 is the starting time of the movement; Γ jolin is the joint torque of the joint under external force load being examined; θ(t) is the function of the joint angle changing with time; M load is the external force load; du is the differential symbol; Γ MVC is the maximum joint torque of the joint. In the model, the co - contraction factor represents the main activity of the muscle. n is the introduced co - contraction factor, and the co - contraction factor represents the main activity of the muscle in each dynamic cycle, excluding the co - contraction area in the same cycle.

[0018] As an optimal solution of a digital twin training method for astronauts based on multi - dimensional data fusion, it further includes: constructing a mechanical model of the upper limb of the spacesuit according to the rigid - body geometric model, kinematic model, and joint damping torque hysteresis model; in the mechanical model of the upper limb of the spacesuit, representing the rotation of the joint through Euler angles; calculating the joint torque based on the weight of the human and the upper limb of the spacesuit, the damping torque of the spacesuit, and the external force; determining the numerical solution of the joint torque through a joint torque model based on Kane's equation according to the angular velocity equation;

[0019] The calculation formula for the joint torque is:

[0020]

[0021] where F r is the joint torque; are the weights of different segments of the upper limb of the human - special clothing respectively; are the three components of the velocity of the center point of the hand in the generalized coordinate system; are the three components of the external force of the hand operation decomposed in the direction respectively; are the three components of the angular velocity of the center point of the hand respectively; are the fixed - angular - velocity vector coefficients respectively;

[0022] The calculation formula for the angular velocity equation is:

[0023]

[0024] where are the three components of the velocity of the center point of the hand in the generalized coordinate system respectively; J1, J2, and J3 are the moments of inertia of the upper arm, forearm, and hand respectively; They are two components of the angular velocity of the hand center point respectively.

[0025] As an optimal solution of an astronaut digital twin training method based on multi-dimensional data fusion, the miniaturized multi-modal sensor network includes: a temperature sensor, a pressure sensor, a vibration sensor, a laser displacement sensor, and a distributed fiber Bragg grating sensor.

[0026] As an optimal solution of an astronaut digital twin training method based on multi-dimensional data fusion, in the process of processing the multi-source heterogeneous data through the dimensionless normalization strategy to eliminate the dimension difference, the expression of the dimensionless normalization strategy is:

[0027]

[0028] In the formula, x is the original data; x norm is the dimensionless data; min(x) and max(x) are the minimum and maximum values of the data respectively.

[0029] As an optimal solution of an astronaut digital twin training method based on multi-dimensional data fusion, in the process of compressing the consistency data through the PCA dimensionality reduction technology, the mathematical model of PCA dimensionality reduction is:

[0030] X = UΣV T

[0031] In the formula, X is the original data matrix; U and V are the left singular matrix and the right singular matrix respectively; Σ is the singular value matrix; T is the transpose matrix.

[0032] The present invention also provides an astronaut digital twin training system based on multi-dimensional data fusion. Based on the above-mentioned astronaut digital twin training method based on multi-dimensional data fusion, it includes:

[0033] A multi-source heterogeneous data acquisition module, which is used to divide the space mission types and astronaut characteristic groups through a stratified sampling strategy to obtain space mission types at a set level and astronaut characteristic groups at a set level; calculate the sample size through a statistical strategy; randomly screen from the astronaut characteristic groups at the set level according to the sample size to obtain an analysis sample; collect data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data;

[0034] A multi-source heterogeneous data processing module, which is used to process the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate the dimensional differences and obtain consistent data; compress the consistent data through PCA dimensionality reduction technology to obtain compressed data; construct a Mandel-Cochran-Grubs three-level outlier detection unit; and process the compressed data for outliers and missing values through the three-level outlier detection unit to obtain a standardized input data set.

[0035] An aerospace ergonomics standardization model construction module, which is used to construct an aerospace ergonomics standardization model based on the hierarchical structure and metadata model according to the standardized input data set.

[0036] An atomic action library construction module, which is used to disassemble the astronaut operation trajectory into atomic actions according to the human body terminal based on the aerospace ergonomics standardization model; and construct an atomic action library by integrating the atomic actions, attributes, interaction relationships, and logical hierarchical relationships.

[0037] A man-machine ergonomics evaluation module, which is used to evaluate and calculate the task load through the VACP model based on the atomic action library to obtain a task load index; and generate an ergonomics evaluation result by analyzing the differences between space and ground for the atomic actions.

[0038] A task process optimization module, which is used to dynamically adjust the task sequence and resource allocation according to the ergonomics evaluation result and the time difference analysis of the atomic action library to optimize the astronaut training process.

[0039] As a preferred solution of an astronaut digital twin training system based on multi-dimensional data fusion, it further includes: a virtual human upper limb bone and muscle model construction and application module, which is used to construct a virtual human upper limb bone and muscle model on the OpenSim platform and set the segment mass and centroid parameters of the upper limb joints in the virtual human upper limb bone and muscle model; represent the upper limb force-generating muscle groups through Hill-type muscle-tendon descriptions; define the muscle property characteristics of the muscle-tendon through the Schutte dynamic muscle model; construct an upper limb joint fatigue model based on the segment mass, the centroid parameters, and the muscle property characteristics of the upper limb joints; and simulate the muscle fatigue accumulation effect through the upper limb joint fatigue model.

[0040] The expression of the upper limb joint fatigue model is:

[0041]

[0042] In the formula, Γ cem (t) is the maximum joint force application ability of the joint at time t; C is an integral constant; e is the natural exponential function; k is a coefficient; n is a co-contraction factor; t0 is the start time of the movement; Γ jolinis the joint torque of the joint under investigation under external load; θ(t) is the function of the joint angle changing with time; M load is the external load; du is the differential symbol; Γ MVC is the maximum joint torque of the joint. As an optimal solution of an astronaut digital twin training system based on multi-dimensional data fusion, it further includes: a construction and processing module for the upper limb mechanical model of the spacesuit, which is used to construct the upper limb mechanical model of the spacesuit according to the rigid body geometric model, kinematic model and joint damping torque hysteresis model; in the upper limb mechanical model of the spacesuit, the joint is rotated and expressed by Euler angles; according to the weights of the human body and the upper limb of the spacesuit, the damping torque of the spacesuit and the external force, the joint torque is calculated; through the joint torque model based on Kane's equation and according to the angular velocity equation, the numerical solution of the joint torque is determined;

[0043] The calculation formula of the joint torque is:

[0044]

[0045] In the formula, F r is the joint torque; are the weights of different segments of the upper limb of the human-special clothing respectively; are the three components of the central point velocity of the hand in the generalized coordinate system; are the three components of the external force of the hand operation decomposed in the direction respectively; are the three components of the angular velocity of the central point of the hand respectively; are the constant angular velocity vector coefficients respectively;

[0046] The calculation formula of the angular velocity equation is:

[0047]

[0048] In the formula, are the three components of the central point velocity of the hand in the generalized coordinate system respectively; J1, J2, and J3 are the inertias of the upper arm, forearm, and hand respectively; are the two components of the angular velocity of the central point of the hand respectively.

[0049] As an optimal solution of an astronaut digital twin training system based on multi-dimensional data fusion, in the multi-source heterogeneous data acquisition module, the miniaturized multi-modal sensor network includes: a temperature sensor, a pressure sensor, a vibration sensor, a laser displacement sensor, and a distributed fiber optic grating sensor.

[0050] As an optimal solution for an astronaut digital twin training system based on multi-dimensional data fusion, in the multi-source heterogeneous data processing module, during the process of processing the multi-source heterogeneous data through the dimensionless normalization strategy to eliminate the dimension difference, the expression of the dimensionless normalization strategy is:

[0051]

[0052] In the formula, x is the original data; x norm is the dimensionless data; min(x) and max(x) are the minimum and maximum values of the data respectively.

[0053] As an optimal solution for an astronaut digital twin training system based on multi-dimensional data fusion, in the multi-source heterogeneous data processing module, during the process of compressing the consistent data through the PCA dimensionality reduction technique, the mathematical model of PCA dimensionality reduction is:

[0054] X = UΣV T

[0055] In the formula, X is the original data matrix; U and V are the left singular matrix and the right singular matrix respectively; Σ is the singular value matrix; T is the transpose matrix.

[0056] The present invention has the following advantages: The present invention divides the space mission types and astronaut characteristic groups through a stratified sampling strategy to obtain the space mission types at the set levels and the astronaut characteristic groups at the set levels; calculates the sample size through a statistical strategy; randomly selects from the astronaut characteristic groups at the set levels according to the sample size to obtain an analysis sample; collects data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data; processes the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate the dimension difference and obtain consistent data; compresses the consistent data through a PCA dimensionality reduction technique to obtain compressed data; constructs a Mandel-Cochran-Grubs three-level outlier detection unit; processes the compressed data for outliers and missing values through the three-level outlier detection unit to obtain a standardized input data set; constructs a space ergonomics standardization model based on the hierarchical structure and metadata model according to the standardized input data set; disassembles the astronaut operation trajectory into atomic actions according to the human terminal based on the space ergonomics standardization model; constructs an atomic action library by integrating the atomic actions, attributes, interaction relationships, and logical hierarchical relationships; evaluates and calculates the task load through a VACP model based on the atomic action library to obtain a task load index; generates an ergonomics evaluation result through the analysis of the heaven-earth differences of the atomic actions; dynamically adjusts the task sequence and resource allocation according to the ergonomics evaluation result and the time difference analysis of the atomic action library to optimize the astronaut training process. The present invention constructs a high-precision space ergonomics standardization model through multi-dimensional standardized data collection and fusion processing, effectively solving the problem of the short board of data standardization. The present invention uses the task analysis method to atomize the astronaut operation process and combines a multi-dimensional evaluation system to accurately optimize the task process, significantly improving the work efficiency and safety of astronauts. In addition, the present invention constructs a biomechanical simulation model of the upper limb bone and muscle system and a human-suit coupling model to accurately quantify the joint forces and muscle fatigue levels of astronauts in complex tasks, providing a scientific basis for the design of space equipment and the optimization of astronaut training programs. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, other implementation drawings can be obtained by extending according to the provided drawings without creative efforts.

[0058] The structures, proportions, sizes, etc. illustrated in this specification are only used to match the content disclosed in the specification for those skilled in this technology to understand and read, and are not used to limit the conditions for the implementation of the present invention. Therefore, they do not have substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size, without affecting the efficacy that the present invention can produce and the purpose that can be achieved, should still fall within the scope covered by the technical content disclosed in the present invention.

[0059] Figure 1 It is a schematic flowchart of a method for astronaut digital twin training based on multi-dimensional data fusion provided in Embodiment 1 of the present invention;

[0060] Figure 2 It is a schematic diagram of an ergonomics evaluation system in a method for astronaut digital twin training based on multi-dimensional data fusion provided in Embodiment 1 of the present invention;

[0061] Figure 3 It is a schematic architecture diagram of a system for astronaut digital twin training based on multi-dimensional data fusion provided in Embodiment 2 of the present invention. Specific implementation manners

[0062] The following specific embodiments illustrate the implementation manners of the present invention. Those skilled in this technology can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0063] Embodiment 1

[0064] See Figure 1 , Embodiment 1 of the present invention provides a method for astronaut digital twin training based on multi-dimensional data fusion, including the following steps:

[0065] S1. Divide the types of spaceflight tasks and the groups of astronaut characteristics through a stratified sampling strategy to obtain the types of spaceflight tasks at the set levels and the groups of astronaut characteristics at the set levels; calculate the sample size through a statistical strategy; randomly select from the groups of astronaut characteristics at the set levels according to the sample size to obtain an analysis sample; collect data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data;

[0066] S2. Process the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate dimensional differences and obtain consistent data; compress the consistent data through PCA dimensionality reduction technology to obtain compressed data; construct a Mandel-Cochran-Grubs three-level outlier detection unit; process outliers and missing values in the compressed data through the three-level outlier detection unit to obtain a standardized input data set;

[0067] S3. Based on the standardized input data set, construct an aerospace ergonomics standardization model based on a hierarchical structure and a metadata model;

[0068] S4. Based on the aerospace ergonomics standardization model, disassemble the astronaut operation trajectory into atomic actions according to the human terminal; integrate the atomic actions, attributes, interaction relationships, and logical hierarchical relationships to construct an atomic action library;

[0069] S5. Based on the atomic action library, evaluate and calculate the task load through the VACP model to obtain a task load index; generate an ergonomics evaluation result through the analysis of the differences between space and ground for the atomic actions;

[0070] S6. According to the ergonomics evaluation result and the time difference analysis of the atomic action library, dynamically adjust the task sequence and resource allocation to optimize the astronaut training process.

[0071] In this embodiment, in step S1, the aerospace mission types and astronaut characteristic groups are divided through a stratified sampling strategy to obtain aerospace mission types at set levels and astronaut characteristic groups at set levels; the sample size is calculated through a statistical strategy; random screening is performed from the astronaut characteristic groups at the set levels according to the sample size to obtain an analysis sample; multi-source heterogeneous data is collected from the analysis sample through a miniaturized multi-modal sensor network;

[0072] Specifically, stratify according to aerospace mission types and individual astronaut characteristics through a stratified sampling strategy to obtain aerospace mission types at set levels and astronaut characteristic groups at set levels, ensuring the comprehensiveness and representativeness of the collected data. Calculate and adjust the sample size through statistical methods according to the variation degree of astronaut physiological indicators and the actual task arrangement, ensuring the accuracy of the data and the economy of the research. Use a random number sequence generated by a computer to number and screen the astronauts to ensure the objectivity of the sample. Deploy a miniaturized and integrated sensor array at the spacecraft end, and collect multi-source heterogeneous data from the analysis sample through a miniaturized multi-modal sensor network.

[0073] Among them, the miniaturized multi-modal sensor network includes physical quantity sensors such as temperature, pressure, and vibration, as well as new devices such as laser displacement sensors and distributed fiber Bragg grating sensors, providing key data for evaluating the safety and health status of astronauts during operations.

[0074] In this embodiment, in step S2, the multi-source heterogeneous data is processed by a dimensionless normalization strategy to eliminate the dimension difference and obtain consistent data; the consistent data is compressed by PCA dimensionality reduction technology to obtain compressed data; a Mandel-Cochran-Grubs three-level outlier detection unit is constructed; and the compressed data is processed for outliers and missing values by the three-level outlier detection unit to obtain a standardized input data set.

[0075] Specifically, the multi-source heterogeneous data is processed by dimensionless processing to eliminate the dimension difference between different data sources, improve the consistency of the data, and obtain consistent data. The consistent data is compressed by PCA dimensionality reduction technology so that the consistent data is compressed to less than 40% of the original dimension and 95% of the key information is retained, obtaining compressed data and improving the subsequent modeling efficiency. A Mandel-Cochran-Grubs three-level outlier screening system is constructed to identify and process outliers and missing values to ensure data quality. Based on the anthropometric data of astronauts, a multivariate design theory framework is constructed to achieve full-process standardized control from data input to result output.

[0076] Among them, dimensionless processing is to convert data with different dimensions into a dimensionless form for unified processing. The expression for dimensionless normalization processing is:

[0077]

[0078] In the formula, x is the original data; x norm is the dimensionless data; min(x) and max(x) are the minimum and maximum values of the data respectively.

[0079] In this embodiment, during the process of compressing the consistent data by the PCA dimensionality reduction technology, the mathematical model of PCA dimensionality reduction is:

[0080] X = UΣV T

[0081] In the formula, X is the original data matrix; U and V are the left singular matrix and the right singular matrix respectively; Σ is the singular value matrix; T is the transpose matrix. By selecting the first k principal components, the data can be reduced to the k-dimensional space.

[0082] In this embodiment, outlier detection is an important method for identifying outliers in data. The present invention uses three statistical methods of Mandel, Cochran, and Grubbs for outlier detection. For example, the formula for Grubbs test is:

[0083]

[0084] where G is the test statistic; x max is the maximum value in the data; is the mean value of the data; s is the standard deviation of the data.

[0085] In this embodiment, in step S3, according to the standardized input data set, an aerospace ergonomics standardization model is constructed based on the hierarchical structure and the metadata model;

[0086] Specifically, the aerospace ergonomics standardization model adopts technologies such as hierarchical structure design, metadata model definition, and normalization processing to realize the full-process standardization of model description, structure design, metadata management, sharing mechanism, and version control. Specific specifications are proposed from two dimensions of functional design and user support, emphasizing interface clarity, operation consistency, multi-functionality, and customizability. At the same time, the system is required to have a modular architecture to facilitate the expansion of new functions. A method for error source analysis and error control based on the detection and calibration of digital standard humans in aerospace is established to ensure the effectiveness of aerospace digital standard humans for value transfer and traceability.

[0087] In this embodiment, in step S4, based on the aerospace ergonomics standardization model, the astronaut operation trajectory is disassembled into atomic actions according to the human terminal; by integrating the atomic actions, attributes, interaction relationships, and logical hierarchical relationships, an atomic action library is constructed;

[0088] Specifically, it is difficult for the action data collected on the ground to reflect the characteristics of aerospace tasks in a microgravity environment. Fine and accurate simulation action disassembly has become the cornerstone for the simulation results to accurately fit the reality. To solve this problem, the present invention disassembles the operation movement trajectory of the astronaut into atomic actions. Atomic actions are the most basic movement units of the human terminal (fingers, palms, forearms, upper arms, torso) or tools that make up the maintenance task. By integrating atomic actions, attributes, interaction relationships, and logical hierarchical relationships, an atomic action library can be formed. The atomic action library should be expandable according to specific tasks, and should be classified hierarchically for easy searching and calling. Each atomic action should contain at least attribute parameters such as time distribution in addition to the basic description.

[0089] The atomic action library has initial atomic action content covering typical operation tasks and supports add / delete / edit functions with management permission constraints. The attribute parameters of each atomic action can be edited and modified. The atomic action library has import and export interfaces, and the files are presented in the form of.xlsx lists. The interface parameters at least include: atomic action name, classification, action description, and time distribution parameters. It supports marking and processing video data: when the number of samples < 3, mean benchmark description statistics are used; when the number of samples ≥ 3, normal test and median test are performed to obtain measured values and confidence intervals.

[0090] Enter the statistical results of atomic action time measurements in ground and on-orbit scenarios into the database as data support for the comprehensive load analysis module. Guide the operator through the interface, and through comparative analysis of ground and space video data, for the same type of atomic action, analyze the time distribution differences between manual operations and tool positioning from dimensions such as interaction path, guiding factors, target state, terminal state, and constraint conditions, support recording the time difference value and storing it in the database, providing data support for comprehensive load analysis. The statistics and analysis of atomic actions are shown in Table 1:

[0091]

[0092] Table 1 Statistics and Analysis of Atomic Actions

[0093] In this embodiment, in step S5, based on the atomic action library, the task load is evaluated and calculated through the VACP model to obtain the task load index; through the analysis of the ground-space differences of the atomic actions, an ergonomic evaluation result is generated;

[0094] Specifically, the basic idea of the VACP evaluation model is to divide the user's brain information processing resources into vision, hearing, cognition, and psychomotor, and evaluate the load level experienced by the operator from these four different components of the information channel. The VACP scoring table is shown in Table 2:

[0095]

[0096] Table 2 VACP Scoring Table

[0097] In this embodiment, the ergonomic evaluation system architecture is as Figure 2As shown in the figure, based on relevant national standards, aerospace standards, and ergonomic requirements, and according to the characteristics of aerospace equipment, the ergonomic evaluation module supports 4 major categories and 12 sub-categories, and adopts detailed and scientific models and data for the ergonomic analysis and evaluation of aerospace equipment. The 4 major categories include geometric ergonomic evaluation, biomechanical evaluation, load metabolism evaluation, and kinematic evaluation; among them, the geometric ergonomic evaluation includes 3 sub-categories: reachability, visibility, and center of gravity calculation, the biomechanical evaluation includes 4 sub-categories: handling force analysis, lower back force analysis, rapid upper limb analysis, and posture comfort analysis, the load metabolism evaluation includes 3 sub-categories: task load analysis, fatigue recovery analysis, and metabolism analysis, and the kinematic evaluation includes 2 sub-categories: operating space analysis and equipment operation limit analysis.

[0098] In this embodiment, in step S6, according to the ergonomic evaluation results and the time difference analysis of the atomic action library, the task sequence and resource allocation are dynamically adjusted to optimize the astronaut training process.

[0099] Among them, the time difference analysis of the atomic action library includes the time difference between space operations and ground operations; specifically, the formula for the time difference between manual operations and tool positioning is:

[0100] T space =T ground ·(1 + β·||▽g||)

[0101] In the formula, T space represents the time of space operation; T ground represents the time of ground operation; β represents the coefficient of operation efficiency; ||▽g|| represents the norm of the gradient, describing the magnitude of the tool positioning change rate.

[0102] This embodiment further includes: constructing a virtual human upper limb bone and muscle model on the OpenSim platform, and setting the segment mass and centroid parameters of the upper limb joints in the virtual human upper limb bone and muscle model; representing the upper limb force-generating muscle groups through Hill-type muscle-tendon descriptions; defining the muscle property characteristics of the muscle-tendon through the Schutte dynamic muscle model; constructing an upper limb joint fatigue model based on the segment mass, centroid parameters, and muscle property characteristics of the upper limb joints; simulating the muscle fatigue accumulation effect through the upper limb joint fatigue model;

[0103] Specifically, on the basis of constructing a bone and muscle model on the OpenSim platform, a virtual human upper limb bone and muscle model matching the astronaut is established. The settings of the upper limb joints and muscles in the model are as follows:

[0104] The main joints of the human upper limb mainly include the scapula, carpal bones, clavicle, humerus, ulna, and radius. With the help of the research results of anatomy, the present invention adds the inertial moment parameter attributes to the hand, radius, ulna, humerus, scapula, and carpal bone segments in the model. The mass and centroid parameters of the upper limb body segments of the model are shown in Table 3:

[0105]

[0106] Table 3 Mass and centroid parameters of the upper limb body segments The inertial moment parameters of the upper limb segments are shown in Table 4:

[0107]

[0108] Table 4 Inertial moment parameters of the upper limb segments

[0109] In this embodiment, eight muscle groups representing the main force - generating muscles of the upper limb are described by the Hill - type muscle - tendon model. The dynamic muscle model described by Schutte is used to define the muscle property characteristics for each muscle - tendon. The fiber length, muscle peak force, and pennation angle parameters of the muscle groups refer to the research results of Holzbaur et al., and their muscle parameter definitions are shown in Table 5:

[0110] Muscle name Abbreviation Fiber length Peak force value Pennation angle Anterior deltoid DELT1 9.8 1218.9 22 Middle deltoid DELT2 10.8 1103.5 15 Posterior deltoid DELT3 13.7 201.6 18 Supraspinatus SUPRA 6.8 499.2 7 Pectoralis major PMAJ1 14.4 444.3 17 Latissimus dorsi LAT1 25.4 290.5 25 Long head of triceps brachii TRIlong 13.4 771.8 12

[0111] Table 5 Upper limb muscle parameters

[0112] In this embodiment, based on the body segment mass, centroid parameters, and muscle property characteristics of the upper limb joints, an upper limb joint fatigue model is constructed; the cumulative effect of muscle fatigue is simulated through the upper limb joint fatigue model;

[0113] Specifically, during the continuous operation of the astronaut, the load accumulates continuously, thus causing fatigue. To explore the fatigue mechanism affecting the movement of the upper limb joints, starting from the perspective of the decline in joint force - applying ability, a joint fatigue model is established under the maximum joint torque and relative joint torque of the joint, and its expression is:

[0114]

[0115] Γ joint =Γ i +Γ ext i = 1,...,n

[0116] In the formula, Γ cem (t) is the maximum joint force - applying ability of the joint at time t; k is the fatigue factor; Γ joint (t) is the external force load received by the joint; Γ MVC is the maximum joint torque of the joint; Γ i is the joint torque exerted by the body limb weight on the joint; Γ extis the external force load; Γ i is the load of the limb itself;

[0117] The joint moment exerted on the joint by the weight of the body limb is composed of variables such as joint angle, joint angular velocity, and joint angular acceleration, and the expression is:

[0118]

[0119] In the formula, d is the differential symbol; t is a certain moment; L is the Lagrangian function; θ i is the joint angle; is the joint angular velocity.

[0120] The external force load on the joint can be expressed as:

[0121]

[0122] In the formula, θ is the joint angle; is the joint angular velocity; is the joint angular acceleration; M load is the external force load.

[0123] After expanding the model parameters, its expression is:

[0124]

[0125] In the formula, Γ MVC(θ(u)) is the function of the maximum joint moment changing with the joint angle;

[0126] Solving the solution can obtain the expression of the force application ability changing with time:

[0127]

[0128] In the formula, C is the integration constant; e is the natural exponential function;

[0129] After introducing the co-contraction factor, the expression of the dynamic fatigue model is:

[0130]

[0131] In the formula, Γ cem (t) is the maximum joint force application ability of the joint at the state of time t; C is the integration constant; e is the natural exponential function; k is the coefficient; n is the co-contraction factor; t0 is the start time of the movement; Γ jolin is the joint moment of the joint under the external force load being investigated; θ(t) is the function of the joint angle changing with time; M load is the external force load; du is the differential symbol; Γ MVCis the maximum joint torque of the joint. In this embodiment, it further includes: constructing a mechanical model of the upper limb of the spacesuit according to the rigid body geometric model, kinematic model, and joint damping torque hysteresis model; in the mechanical model of the upper limb of the spacesuit, the joint is rotated and expressed by Euler angles; calculating the joint torque according to the weight of the human and the upper limb of the spacesuit, the damping torque of the spacesuit, and the external force; determining the numerical solution of the joint torque through the joint torque model based on Kane's equation according to the angular velocity equation;

[0132] Specifically, the mechanical model of the upper limb of the spacesuit is composed of a rigid body geometric model, a kinematic model, and a joint damping torque hysteresis model. According to the actual structure of the spacesuit, it is assumed that each segment of the upper limb of the spacesuit is a hollow cylinder, and it is assumed that the center of mass is located at the center of the hollow cylinder. The formula for calculating the inertia of this segment of the upper limb of the spacesuit is:

[0133]

[0134] In the formula, I″ xx is the inertia of each segment of the upper limb in the x-axis direction; I″ yy is the inertia of each segment of the upper limb in the y-axis direction; I″ zz is the inertia of each segment of the upper limb in the z-axis direction; M is the segment mass; R1 and R2 are the inner radius and outer radius respectively; L is the length.

[0135] Among them, the geometric and physical parameters of the spacesuit are shown in Table 6:

[0136] Trunk segment Length (m) Center of mass (m) Weight (kg) Moment of inertia (kg·m2) Upper arm <![CDATA[l1]]> <![CDATA[c1]]> <![CDATA[m1]]> <![CDATA[J1]]> Forearm <![CDATA[l2]]> <![CDATA[c2]]> <![CDATA[m2]]> <![CDATA[J2]]> Hand <![CDATA[l3]]> <![CDATA[c3]]> <![CDATA[m3]]> <![CDATA[J3]]>

[0137] Table 6 Geometric and Physical Parameters of the Spacesuit

[0138] In this embodiment, the kinematic model of the upper limb of the spacesuit refers to the kinematic modeling idea of the human upper limb. The kinematic model of the upper limb of the spacesuit has three joints: the shoulder, elbow, and wrist, and Euler angles are used to represent the joint rotation. From the perspective of biomechanics, the upper limb (excluding the hand joints) has six degrees of freedom. According to the coordinate rotation of Euler angles, the direction rotation matrices of the shoulder, elbow, and wrist under their respective degrees of freedom are expressed as:

[0139]

[0140] In the formula, A(α) is the rotation matrix of the shoulder, elbow, and wrist under the α degree of freedom; A(β) is the rotation matrix of the shoulder, elbow, and wrist under the β degree of freedom; A(γ) is the rotation matrix of the shoulder, elbow, and wrist under the γ degree of freedom.

[0141] According to the characteristics of the human-suit system, in the dynamic model of the human-suit system, the joint torque borne by the human body under the human-suit system is composed of the weight of the human upper limb of the spacesuit, the damping torque of the spacesuit, and the external force. The formula for calculating the joint torque is:

[0142]

[0143] In the formula, F r is the joint torque; are the weights of different segments of the upper limb of the human-special clothing respectively; are the three components of the velocity of the center point of the hand in the generalized coordinate system; are the three components of the external force of the hand operation decomposed in the direction respectively; are the three components of the angular velocity of the center point of the hand respectively; are the constant angular velocity vector coefficients respectively;

[0144] For the final joint torque model based on Kane's equation, the numerical solution of the joint torque can be determined according to the angular velocity equation (shown in the formula).

[0145] The calculation formula of the angular velocity equation is:

[0146]

[0147] In the formula, are the three components of the velocity of the center point of the hand in the generalized coordinate system respectively; J1, J2, and J3 are the inertias of the upper arm, forearm, and hand respectively; are the two components of the angular velocity of the center point of the hand respectively.

[0148] In summary, the present invention divides space mission types and astronaut characteristic groups through a stratified sampling strategy to obtain space mission types at set levels and astronaut characteristic groups at set levels; calculates the sample size through a statistical strategy; randomly selects from the astronaut characteristic groups at the set levels according to the sample size to obtain an analysis sample; collects data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data; processes the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate dimension differences and obtain consistent data; compresses the consistent data through PCA dimensionality reduction technology to obtain compressed data; constructs a Mandel-Cochran-Grubs three-level outlier detection unit; processes outliers and missing values of the compressed data through the three-level outlier detection unit to obtain a standardized input data set; constructs a space ergonomics standardization model based on a hierarchical structure and a metadata model according to the standardized input data set; disassembles the astronaut operation trajectory into atomic actions according to the human terminal based on the space ergonomics standardization model; integrates the atomic actions, attributes, interaction relationships, and logical hierarchical relationships to construct an atomic action library; evaluates and calculates the task load through a VACP model based on the atomic action library to obtain a task load index; generates an ergonomics evaluation result through a space-earth difference analysis of the atomic actions; dynamically adjusts the task sequence and resource allocation according to the ergonomics evaluation result and the time difference analysis of the atomic action library to optimize the astronaut training process. The present invention constructs a high-precision space ergonomics standardization model through multi-dimensional standardized data collection and fusion processing, effectively solving the problem of the short board in data standardization. The present invention uses a task analysis method to atomize the astronaut operation process, combines a multi-dimensional evaluation system, and precisely optimizes the task process, significantly improving the work efficiency and safety of astronauts. In addition, the present invention constructs a biomechanical simulation model of the upper limb bone and muscle system and a human-suit coupling model, precisely quantifying the joint forces and muscle fatigue levels of astronauts in complex tasks, providing a scientific basis for space equipment design and optimization of astronaut training programs.

[0149] It should be noted that the method of the embodiment of the present disclosure can be executed by a single device, such as a computer or a server. The method of this embodiment can also be applied to a distributed scenario, and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only execute one or more steps of the method of the embodiment of the present disclosure, and these multiple devices will interact with each other to complete the described method.

[0150] It should be noted that some embodiments of the present disclosure have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the above embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0151] Embodiment 2

[0152] Refer to Figure 3 , Embodiment 2 of the present invention further provides an astronaut digital twin training system based on multi-dimensional data fusion, including:

[0153] A multi-source heterogeneous data acquisition module 001, configured to divide space mission types and astronaut characteristic groups through a stratified sampling strategy to obtain space mission types at a set level and astronaut characteristic groups at a set level; calculate the sample size through a statistical strategy; randomly select from the astronaut characteristic groups at the set level according to the sample size to obtain an analysis sample; collect data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data;

[0154] A multi-source heterogeneous data processing module 002, configured to process the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate dimension differences and obtain consistent data; compress the consistent data through PCA dimensionality reduction technology to obtain compressed data; construct a Mandel-Cochran-Grubs three-level outlier detection unit; process outliers and missing values of the compressed data through the three-level outlier detection unit to obtain a standardized input data set;

[0155] An aerospace ergonomics standardization model construction module 003, configured to construct an aerospace ergonomics standardization model based on the standardized input data set based on a hierarchical structure and a metadata model;

[0156] An atomic action library construction module 004, configured to decompose the astronaut operation trajectory into atomic actions according to the human body terminal based on the aerospace ergonomics standardization model; integrate the atomic actions, attributes, interaction relationships, and logical hierarchical relationships to construct an atomic action library;

[0157] A human-machine ergonomics evaluation module 005, configured to evaluate and calculate the task load through the VACP model based on the atomic action library to obtain a task load index; generate an ergonomics evaluation result through an analysis of the differences between space and earth for the atomic actions;

[0158] The task process optimization module 006 is used to dynamically adjust the task sequence and resource allocation according to the ergonomics evaluation results and the time difference analysis of the atomic action library, so as to optimize the astronaut training process.

[0159] In this embodiment, it further includes: a virtual human upper limb bone and muscle model construction and application module, which is used to construct a virtual human upper limb bone and muscle model on the OpenSim platform, and set the segment mass and centroid parameters of the upper limb joints in the virtual human upper limb bone and muscle model; represent the upper limb force-generating muscle groups through the Hill-type muscle-tendon description; define the muscle property characteristics of the muscle-tendon through the Schutte dynamic muscle model; construct an upper limb joint fatigue model based on the segment mass, the centroid parameters and the muscle property characteristics of the upper limb joints; simulate the muscle fatigue cumulative effect through the upper limb joint fatigue model;

[0160] The expression of the upper limb joint fatigue model is:

[0161]

[0162] In the formula, Γ cem (t) is the maximum joint force application ability of the joint at time t; C is the integral constant; e is the natural exponential function; k is the coefficient; n is the co-contraction factor; t0 is the start time of the movement; Γ jolin is the joint torque of the joint under the external force load being investigated; θ(t) is the function of the joint angle changing with time; M load is the external force load; du is the differential symbol; Γ MVC is the maximum joint torque of the joint. In this embodiment, it further includes: a spacesuit upper limb mechanical model construction and processing module, which is used to construct a spacesuit upper limb mechanical model according to the rigid body geometric model, the kinematic model and the joint damping torque hysteresis model; in the spacesuit upper limb mechanical model, the joint is rotated and expressed through Euler angles; the joint torque is calculated according to the human and spacesuit upper limb limb weights, the spacesuit damping torque and the external force; the numerical solution of the joint torque is determined through the joint torque model based on Kane's equation according to the angular velocity equation;

[0163] The calculation formula of the joint torque is:

[0164]

[0165] In the formula, F r is the joint torque; are the weights of different segments of the human-special clothing upper limb respectively; are the three components of the center point velocity of the hand in the generalized coordinate system; are the external forces decomposed by the hand operation respectively in Three components in a direction; They are respectively three components of the angular velocity of the center point of the hand; They are respectively the constant angular velocity vector coefficients;

[0166] The calculation formula of the angular velocity equation is:

[0167]

[0168] In the formula, They are respectively three components of the velocity of the center point of the hand in the generalized coordinate system; J1, J2, and J3 are respectively the inertias of the upper arm, forearm, and hand; They are respectively two components of the angular velocity of the center point of the hand.

[0169] In this embodiment, in the multi-source heterogeneous data acquisition module 001, the miniaturized multi-modal sensor network includes: a temperature sensor, a pressure sensor, a vibration sensor, a laser displacement sensor, and a distributed fiber Bragg grating sensor.

[0170] In this embodiment, in the multi-source heterogeneous data processing module 002, in the process of processing the multi-source heterogeneous data through the dimensionless normalization strategy to eliminate the dimension difference, the expression of the dimensionless normalization strategy is:

[0171]

[0172] In the formula, x is the original data; x norm is the data after dimensionless normalization; min(x) and max(x) are respectively the minimum and maximum values of the data.

[0173] In this embodiment, in the multi-source heterogeneous data processing module 002, in the process of compressing the consistency data through the PCA dimensionality reduction technology, the mathematical model of PCA dimensionality reduction is:

[0174] X = UΣV T

[0175] In the formula, X is the original data matrix; U and V are respectively the left singular matrix and the right singular matrix; Σ is the singular value matrix; T is the transpose matrix.

[0176] It should be noted that the information interaction, execution process, etc. between the above system modules, because they are based on the same concept as the method embodiment in Embodiment 1 of the present application, the technical effects brought by them are the same as those of the method embodiment of the present application. For the specific content, reference can be made to the description in the method embodiment shown above in the present application, and details are not described here again.

[0177] Embodiment 3

[0178] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium, in which program code of a method for astronaut digital twin training based on multi-dimensional data fusion is stored, and the program code includes instructions for executing a method for astronaut digital twin training based on multi-dimensional data fusion according to Embodiment 1 or any possible implementation thereof.

[0179] The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center integrating one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state disk (SSD)).

[0180] Embodiment 4

[0181] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor;

[0182] The processor and the memory communicate with each other through a bus; the memory stores program instructions executable by the processor, and the processor can execute a method for astronaut digital twin training based on multi-dimensional data fusion according to Embodiment 1 or any possible implementation thereof by calling the program instructions.

[0183] Specifically, the processor can be implemented by hardware or by software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor, which is implemented by reading software code stored in the memory. The memory can be integrated in the processor or can exist independently outside the processor.

[0184] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from a website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.).

[0185] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing system. They can be concentrated on a single computing system or distributed on a network composed of multiple computing systems. Optionally, they can be implemented by program code executable by the computing system. Thus, they can be stored in a storage system and executed by the computing system. And in some cases, the steps shown or described can be executed in a different order than here, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module to be implemented. In this way, the present invention is not limited to any specific combination of hardware and software.

[0186] Although the present invention has been described in detail above with general descriptions and specific embodiments, based on the present invention, some modifications or improvements can be made, which are obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection required by the present invention.

Claims

1. An astronaut digital twin training method based on multi-dimensional data fusion, characterized in that, Including: Dividing space mission types and astronaut characteristic groups through a stratified sampling strategy to obtain space mission types at set levels and astronaut characteristic groups at set levels; calculating the sample size through a statistical strategy; randomly screening from the astronaut characteristic groups at the set levels according to the sample size to obtain an analysis sample; collecting data from the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data; Processing the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate dimension differences and obtain consistent data; compressing the consistent data through PCA dimensionality reduction technology to obtain compressed data; constructing a Mandel-Cochran-Grubbs three-level outlier detection unit; processing outliers and missing values in the compressed data through the three-level outlier detection unit to obtain a standardized input data set; Based on the standardized input data set, constructing a space ergonomics standardization model based on a hierarchical structure and a metadata model; Based on the space ergonomics standardization model, disassembling the astronaut operation trajectory into atomic actions according to the human terminal; constructing an atomic action library by integrating the atomic actions, attributes, interaction relationships, and logical hierarchical relationships; Based on the atomic action library, evaluating and calculating the task load through the VACP model to obtain a task load index; generating an ergonomics evaluation result by analyzing the differences between the sky and the ground of the atomic actions; According to the ergonomics evaluation result and the time difference analysis of the atomic action library, dynamically adjusting the task sequence and resource allocation to optimize the astronaut training process.

2. The digital twin training method for astronauts based on multi-dimensional data fusion according to claim 1, wherein In the process of constructing a space ergonomics standardization model based on a hierarchical structure and a metadata model, it includes: constructing a virtual human upper limb bone and muscle model on the OpenSim platform, and setting the segment mass and centroid parameters of the upper limb joints in the virtual human upper limb bone and muscle model; representing the upper limb force-generating muscle groups through Hill-type muscle-tendon descriptions; defining the muscle attribute characteristics of the muscle-tendon through the Schutte dynamic muscle model; constructing an upper limb joint fatigue model based on the segment mass, centroid parameters, and muscle attribute characteristics of the upper limb joints; simulating the muscle fatigue accumulation effect through the upper limb joint fatigue model; The expression of the upper limb joint fatigue model is: where, Γ cem (t) is the maximum joint force application ability of the joint at time t; C is an integral constant; e is the natural exponential function; k is a coefficient; n is the co - contraction factor; t0 is the start time of the movement; Γ jolin is the joint torque of the joint under external force load being examined; θ(t) is the function of the joint angle changing with time; M load is the external force load; du is the differential symbol; Γ MVC is the maximum joint torque of the joint.

3. The astronaut digital twin training method based on multi-dimensional data fusion according to claim 2, wherein, It also includes: Constructing a space suit upper limb mechanics model according to a rigid body geometry model, a kinematics model, and a joint damping torque hysteresis model; in the space suit upper limb mechanics model, expressing the rotation of the joint through Euler angles; Calculating the joint torque according to the weights of the human and space suit upper limb limbs, the damping torque of the space suit, and the external force; determining the numerical solution of the joint torque according to the angular velocity equation through a joint torque model based on Kane's equation; The calculation formula of the joint torque is: Where, F r is the joint torque; are the weights of different segments of the upper limb of the human-special clothing respectively; are the three components of the central point velocity of the hand in the generalized coordinate system; are the external forces of the hand operation decomposed into the three components in the direction respectively; are the three components of the angular velocity of the central point of the hand respectively; are the constant angular velocity vector coefficients respectively; The calculation formula of the angular velocity equation is: wherein, are respectively the three components of the velocity of the center point of the hand in the generalized coordinate system; J1, J2, and J3 are respectively the inertias of the upper arm, forearm, and hand; are respectively the two components of the angular velocity of the center point of the hand.

4. The digital twin training method for astronauts based on multi-dimensional data fusion according to claim 3, wherein, The miniaturized multi-modal sensor network includes: a temperature sensor, a pressure sensor, a vibration sensor, a laser displacement sensor, and a distributed fiber Bragg grating sensor.

5. A method for training astronauts' digital twins based on multi-dimensional data fusion according to claim 4, characterized in that, In the process of processing the multi-source heterogeneous data through the dimensionless normalization strategy to eliminate the dimensional difference, the expression of the dimensionless normalization strategy is: where x is the original data; x norm is the dimensionless data; min(x) and max(x) are the minimum and maximum values of the data, respectively.

6. The digital twin training method for astronauts based on multi-dimensional data fusion according to claim 5, characterized in that In the process of compressing the consistency data through the PCA dimensionality reduction technology, the mathematical model of PCA dimensionality reduction is: X = UΣV T In the formula, X is the original data matrix; U and V are the left singular matrix and the right singular matrix respectively; Σ is the singular value matrix; T is the transpose matrix.

7. An astronaut digital twin training system based on multi-dimensional data fusion, which adopts a method for training an astronaut digital twin based on multi-dimensional data fusion according to any one of claims 1-6, and is characterized in that It includes: A multi-source heterogeneous data acquisition module, which is used to divide the space mission types and astronaut characteristic groups through a stratified sampling strategy to obtain space mission types at a set level and astronaut characteristic groups at a set level; calculate the sample size through a statistical strategy; randomly screen according to the sample size from the astronaut characteristic groups at the set level to obtain an analysis sample; collect data on the analysis sample through a miniaturized multi-modal sensor network to obtain multi-source heterogeneous data; A multi-source heterogeneous data processing module, which is used to process the multi-source heterogeneous data through a dimensionless normalization strategy to eliminate the dimensional difference and obtain consistency data; compress the consistency data through the PCA dimensionality reduction technology to obtain compressed data; construct a Mandel-Cochran-Grubbs three-level outlier detection unit; process the compressed data for outliers and missing values through the three-level outlier detection unit to obtain a standardized input data set; An aerospace ergonomics standardization model construction module, which is used to construct an aerospace ergonomics standardization model based on the hierarchical structure and metadata model according to the standardized input data set; An atomic action library construction module, which is used to decompose the astronaut operation trajectory into atomic actions according to the human body terminal based on the aerospace ergonomics standardization model; integrate the atomic actions, attributes, interaction relationships, and logical hierarchical relationships to construct an atomic action library; A man-machine ergonomics evaluation module, which is used to evaluate and calculate the task load through the VACP model based on the atomic action library to obtain a task load index; generate an ergonomics evaluation result through the analysis of the differences between space and ground for the atomic actions; A task process optimization module, which is used to dynamically adjust the task sequence and resource allocation according to the ergonomics evaluation result and the time difference analysis of the atomic action library to optimize the astronaut training process.

8. A digital twin training system for astronauts based on multi-dimensional data fusion according to claim 7, characterized in that, It also includes: A virtual human upper limb bone and muscle model construction and application module, which is used to construct a virtual human upper limb bone and muscle model on the OpenSim platform and set the segment mass and centroid parameters of the upper limb joints in the virtual human upper limb bone and muscle model; represent the upper limb force-generating muscle groups through a Hill-type muscle-tendon description; define the muscle property characteristics of the muscle-tendon through the Schutte dynamic muscle model; construct an upper limb joint fatigue model based on the segment mass, centroid parameters, and muscle property characteristics of the upper limb joints; simulate the muscle fatigue accumulation effect through the upper limb joint fatigue model; The expression of the upper limb joint fatigue model is: The expression of the upper limb joint fatigue model is: where Γ cem (t) is the maximum joint force application ability of the joint in the state at time t; C is an integral constant; e is the natural exponential function; k is a coefficient; n is a co-contraction factor; t0 is the starting time of the movement; Γ jolin is the joint torque of the joint under external load being investigated; θ(t) is the function of the joint angle changing with time; M load is the external load; du is the differential symbol; Γ MVC is the maximum joint torque of the joint.

9. The digital twin training system for astronauts based on multi-dimensional data fusion according to claim 8, characterized in that, It also includes: The upper limb mechanical model construction and processing module of the spacesuit is used to construct the upper limb mechanical model of the spacesuit according to the rigid body geometric model, kinematic model and joint damping torque hysteresis model; in the upper limb mechanical model of the spacesuit, the joints are rotationally represented by Euler angles; Calculate the joint torque according to the weights of the human and the upper limb of the spacesuit, the damping torque of the spacesuit and the external force; determine the numerical solution of the joint torque according to the angular velocity equation through the joint torque model based on Kane's equation; The calculation formula of the joint torque is: Where, F r is the joint torque; are the weights of different segments of the upper limb of the human-special clothing respectively; are the three components of the central point velocity of the hand in the generalized coordinate system; are the three components of the external force of the hand operation decomposed in the direction respectively; are the three components of the angular velocity of the central point of the hand respectively; are the constant angular velocity vector coefficients respectively; The calculation formula of the angular velocity equation is: In the formula, are respectively the three components of the velocity of the center point of the hand in the generalized coordinate system; J1, J2, and J3 are respectively the inertias of the upper arm, forearm, and hand; are respectively the two components of the angular velocity of the center point of the hand.

10. The astronaut digital twin training system based on multi-dimensional data fusion according to claim 9, wherein, In the multi-source heterogeneous data acquisition module, the miniaturized multi-modal sensor network includes: temperature sensors, pressure sensors, vibration sensors, laser displacement sensors and distributed fiber Bragg grating sensors.

Images