Digital twinning-oriented man-machine interaction adaptive method

Abstract

The invention discloses a digital twinning-oriented man-machine interaction adaptive method. The method comprises the following steps of 1, data collection and preprocessing: collecting sensor data; step 2, constructing a data twinborn model: modeling; 3, developing an intelligent decision support system: designing a decision algorithm; 4, learning optimization based on time sequence data: time sequence data analysis: performing deep analysis on historical operation data of the mechanical arm by using a time sequence analysis technology; 5, interactive interface design: a dynamic interactive interface: developing a user-friendly dynamic interactive interface; and 6, real-time monitoring and adjustment are conducted, wherein a real-time monitoring system is used for achieving real-time monitoring of the mechanical arm and the operation environment of the mechanical arm. By analyzing the digital twin model and the sensor data in real time, the difference between the current state and the target can be quickly recognized, so that optimization suggestions are generated, the intelligent decision support can help the mechanical arm to efficiently execute tasks in different working modes, and the production efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of robot arm interaction, and in particular to a human-machine interaction adaptive method for digital twins. Background Art

[0002] With the rapid development of industrial automation and intelligent manufacturing, robotic arms have been widely used in various production environments. In order to improve the operational efficiency and safety of robotic arms, researchers are increasingly focusing on how to effectively combine the physical world with digital models, which is where digital twin technology comes in. Digital twins are virtual models based on real-time data that can reflect the state of a physical system in real time and predict its future performance through simulation.

[0003] Current research focuses on the construction of digital twin models and the integration of real-time data. However, how to effectively use these models and data for intelligent decision support remains an urgent problem to be solved. Existing technologies often lack in-depth analysis of operational data and dynamic decision-making capabilities based on data feedback, which results in the inability of the robot arm to flexibly respond to changing working environments when performing tasks, reducing overall operational efficiency and safety. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above-mentioned problems existing in the existing human-computer interaction adaptive method for digital twins, the present invention is proposed.

[0006] Therefore, the purpose of the present invention is to provide a human-computer interaction adaptive method for digital twins, which can quickly identify the gap between the current state and the target by real-time analysis of the digital twin model and sensor data, thereby generating optimization suggestions. This intelligent decision support can help the robotic arm efficiently perform tasks in different working modes and improve production efficiency.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a human-computer interaction adaptive method for digital twins, comprising the following steps:

[0008] Step 1: Data collection and preprocessing:

[0009] Sensor data collection: collect real-time data of the robot arm through machine vision, force sensors, and acceleration sensors, including position, speed, acceleration, and force feedback;

[0010] Data preprocessing: denoising, standardizing and feature extraction of collected data to improve data quality and effectiveness of subsequent analysis;

[0011] Step 2: Build a data twin model:

[0012] Modeling: Building a digital twin model of the robotic arm based on the collected sensor data to simulate its performance under different working conditions;

[0013] Step 3: Intelligent decision support system development:

[0014] Decision algorithm design: Develop a decision algorithm based on machine learning that can analyze digital twin models and sensor data in real time to identify the gap between the current state and the target;

[0015] Suggestion generation: The algorithm generates optimization suggestions based on the analysis results, such as adjusting the working mode of the robot arm and modifying operating parameters to improve efficiency and safety;

[0016] Step 4: Learning optimization based on time series data:

[0017] Time series data analysis: Use time series analysis technology to conduct in-depth analysis of the robot's historical operation data to identify successful operation patterns and failure cases;

[0018] Step 5: Interactive interface design:

[0019] Dynamic interactive interface: Develop a user-friendly dynamic interactive interface to display the digital twin status, recommended actions and potential problems in real time;

[0020] Step 6: Real-time monitoring and adjustment:

[0021] Real-time monitoring system: Realize real-time monitoring of the robot arm and its operating environment, obtain key data in a timely manner and feed it back to the intelligent decision support system;

[0022] Dynamically adjust strategies: Based on real-time monitoring, the system can automatically adjust decision-making strategies to respond to emergencies or environmental changes.

[0023] As a preferred solution of the human-computer interaction adaptive method for digital twins of the present invention, the step 2 of constructing a digital twin model of the robotic arm includes:

[0024] 1) Geometric model:

[0025] Description: Use CAD software or modeling tools to build a 3D geometric model of the robotic arm, defining the size, shape, and connection relationship of each component;

[0026] 2) Kinematic model:

[0027] Forward Kinematics:

[0028] T=T 0 ·T 1 (θ 1 )·T 2 (θ 2 )…T n (θ n )

[0029] Where T is the transformation matrix of the end effector, T n (θ n ) is the transformation matrix of the nth joint, θ n is the joint angle;

[0030] Inverse Kinematics:

[0031] Solve T = T 0 …T n (θ 1 ,θ 2 ,…,θ n )

[0032] 3) Kinetic model:

[0033] Newton–Euler method:

[0034] Calculate the forces and torques at each joint:

[0035] F net =m·a+F gravity +F friction

[0036] Among them, F net is the net force, m is the mass, a is the acceleration, F gravity is the influence of gravity, F friction is the friction force;

[0037] 4) Control model:

[0038] PID Controller:

[0039]

[0040] Where u(t) is the control input, e(t) is the error, and K p , K i , K d are the proportional, integral and derivative gains respectively, τ is the joint torque;

[0041] Model Predictive Control:

[0042] The model-based optimization control problem is formulated as:

[0043]

[0044] Among them, y k is the system output, y ref is the reference output, N is the number of prediction steps, and λ is the regularization parameter;

[0045] 5) Environmental interaction model:

[0046] Interaction force model:

[0047] Based on Hooke's law and friction model, the relationship between force and displacement is described as:

[0048]

[0049] Among them, F is the interaction force, k is the elastic coefficient, x is the displacement, and b is the damping coefficient;

[0050] 6) Data-driven modeling:

[0051] Machine Learning Models:

[0052] Use regression algorithms to construct a mapping relationship between input (sensor data) and output (robotic arm behavior):

[0053]

[0054] in, is the predicted value, X is the input feature, and θ is the model parameter.

[0055] As a preferred solution of the human-computer interaction adaptive method for digital twins described in the present invention, the decision algorithm in step three specifically includes:

[0056] S1. Input feature selection:

[0057] Extract features from sensor data and digital twin models as input to the algorithm. These features include:

[0058] The current position information of the robot arm (joint angle, end effector position); current speed and acceleration; force feedback (such as grasping force, collision detection); environmental information (such as obstacle position, target object status)

[0059] S2. Target state definition:

[0060] Define the desired target state, including target position, target velocity, and desired interaction force;

[0061] S3. Gap calculation:

[0062] Calculate the error between the current state and the target state;

[0063] S4. Machine learning model training:

[0064] Use deep learning methods to train the decision model. Based on the gap between input features and target states, the model learns how to adjust the behavior of the robotic arm.

[0065] S5. Decision-making process:

[0066] Make real-time decisions based on trained models;

[0067] S6. Feedback mechanism:

[0068] Update and optimize the decision model in real time based on the actual performance of the robot arm and newly collected data;

[0069] S7. Risk Assessment and Management:

[0070] During the decision-making process, a risk assessment mechanism is added to make decision adjustments based on the uncertainty and potential risks of the current state to ensure the safety of the robot arm operation.

[0071] As a preferred solution of the human-computer interaction adaptive method for digital twins described in the present invention, the method for calculating the gap in S3 includes:

[0072] e(t)=y target -y current

[0073] Where e(t) is the error, y target is the target state, y current is the current status.

[0074] As a preferred solution of the human-computer interaction adaptive method for digital twins described in the present invention, the deep learning algorithm in S4 uses a deep Q network:

[0075]

[0076] Among them, Q(s t ,a t ) is in state s t Next, perform action a t the current estimated value of

[0077] α is the learning rate, which determines the degree to which new information affects existing knowledge;

[0078] γ is the discount factor, which represents the present value of future rewards;

[0079] r t is the current reward, r t =R(s t ,a t ).

[0080] As a preferred solution of the human-computer interaction adaptive method for digital twins of the present invention, the real-time decision in S5 includes:

[0081] Determine action: Choose action: Choose the best action a t To minimize the error:

[0082]

[0083] Execute action: Apply the selected action to the robot and update the current state.

[0084] As a preferred solution of the human-computer interaction adaptive method for digital twins of the present invention, the model update in S6 is:

[0085]

[0086] Among them, θ is the model parameter, η is the learning rate, and L(θ) is the loss function.

[0087] As a preferred solution of the human-computer interaction adaptive method for digital twins described in the present invention, the timing analysis technology in step 4 specifically includes:

[0088] A1. Data collection and preprocessing:

[0089] Data collation: The historical operation data of the robot arm is collated into a time series data format, including timestamp, joint angle, end effector position, speed, and force feedback information;

[0090] Missing value processing: interpolation or deletion of missing data to ensure data integrity;

[0091] A2. Feature extraction:

[0092] Statistical features: Extract statistical features from time series data, such as mean, variance, peak, maximum, and minimum values;

[0093] Frequency domain features: Use fast Fourier transform to convert time series data into frequency domain and extract frequency features;

[0094] A3. Pattern recognition:

[0095] Cluster analysis: Use clustering algorithms to cluster historical data and identify successful and failed operation patterns;

[0096] Template matching: Create templates of successful and failed operations and use distance metrics to identify the similarity between new operations and the templates.

[0097] A4. Anomaly Detection:

[0098] Statistical-based methods: Use control charts to monitor changes in time series data and identify abnormal operations;

[0099] Based on machine learning methods: using isolation forest or one-class support vector machine for anomaly detection;

[0100] A5. Trend analysis:

[0101] Moving average method:

[0102]

[0103] Among them, MA t is the moving average at time point t, y is the time series data value, and n is the window size;

[0104] Exponential smoothing:

[0105] S t =αy t +(1-α)S t-1

[0106] Among them, S t is the smoothing value, α is the smoothing constant (0<α<1)

[0107] A6. Timing Modeling:

[0108] Autoregressive integrated moving average model:

[0109] Set the model parameters and perform simulation fitting:

[0110] y t =c+φ 1 y t-1 +…+θ p y t-p +θ 1 ε t-1 +…+θ q ε t-q +ε t

[0111] Among them, c is a constant term, φ and θ are model parameters, and ε is white noise;

[0112] Establish LSTM model for time series prediction:

[0113] h t =f(W h h t-1 +W x x t +b)

[0114] Among them, h t is the hidden state at the current moment, x t is the current input, W hand W x is the weight matrix and b is the bias.

[0115] As a preferred solution of the human-computer interaction adaptive method for digital twins described in the present invention, the timing analysis result evaluation and optimization includes:

[0116] Performance metrics: Use mean squared error to evaluate model performance:

[0117]

[0118] Among them, y i is the actual value, is the predicted value;

[0119] Feedback mechanism: Based on the analysis results, feedback is given to the external control system to optimize subsequent operation strategies.

[0120] Beneficial effects of the present invention:

[0121] 1. Improve operational efficiency: By analyzing the digital twin model and sensor data in real time, the gap between the current state and the target can be quickly identified, thereby generating optimization suggestions. This intelligent decision support can help the robotic arm perform tasks efficiently in different working modes and improve production efficiency.

[0122] 2. Enhanced safety: The real-time monitoring system can obtain key data in a timely manner and feed it back to the intelligent decision support system. In emergency situations, the system can automatically adjust the decision-making strategy to effectively reduce the risk of accidents and ensure operational safety.

[0123] 3. Optimize human-machine interaction: The design of the dynamic interactive interface enables users to obtain real-time information on the robot's status, recommended operations, and potential problems. This transparent information flow can enhance users' sense of trust and improve the efficiency of human-machine collaboration.

[0124] 4. Achieve adaptive control: Learning optimization based on time series data can help the system identify successful operation modes and failed cases, thereby continuously improving the decision-making model. Through this adaptive capability, the robotic arm can continuously optimize its performance in different working environments and improve its overall flexibility.

[0125] 5. Reduce maintenance costs: Through effective data analysis and prediction of maintenance needs, the failure rate and downtime of the robotic arm can be reduced, maintenance costs can be reduced, and the economic benefits of production can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0126] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0127] Figure 1 This is a schematic diagram of the method structure of the human-computer interaction adaptive method for digital twins of the present invention. DETAILED DESCRIPTION

[0128] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0129] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0130] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0131] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0132] Reference Figure 1 , provides a human-computer interaction adaptive method for digital twins, including the following steps:

[0133] Step 1: Data collection and preprocessing:

[0134] Sensor data collection: collect real-time data of the robot arm through machine vision, force sensors, and acceleration sensors, including position, speed, acceleration, and force feedback;

[0135] Data preprocessing: denoising, standardizing and feature extraction of collected data to improve data quality and effectiveness of subsequent analysis;

[0136] Step 2: Build a data twin model:

[0137] Modeling: Building a digital twin model of the robotic arm based on the collected sensor data to simulate its performance under different working conditions;

[0138] Step 3: Intelligent decision support system development:

[0139] Decision algorithm design: Develop a decision algorithm based on machine learning that can analyze digital twin models and sensor data in real time to identify the gap between the current state and the target;

[0140] Suggestion generation: The algorithm generates optimization suggestions based on the analysis results, such as adjusting the working mode of the robot arm and modifying operating parameters to improve efficiency and safety;

[0141] Step 4: Learning optimization based on time series data:

[0142] Time series data analysis: Use time series analysis technology to conduct in-depth analysis of the robot's historical operation data to identify successful operation patterns and failure cases;

[0143] Step 5: Interactive interface design:

[0144] Dynamic interactive interface: Develop a user-friendly dynamic interactive interface to display the digital twin status, recommended actions and potential problems in real time;

[0145] Step 6: Real-time monitoring and adjustment:

[0146] Real-time monitoring system: Realize real-time monitoring of the robot arm and its operating environment, obtain key data in a timely manner and feed it back to the intelligent decision support system;

[0147] Dynamically adjust strategies: Based on real-time monitoring, the system can automatically adjust decision-making strategies to respond to emergencies or environmental changes.

[0148] Among them, building a digital twin model of the robotic arm in step 2 includes:

[0149] 1) Geometric model:

[0150] Description: Use CAD software or modeling tools to build a 3D geometric model of the robotic arm, defining the size, shape, and connection relationship of each component;

[0151] 2) Kinematic model:

[0152] Forward Kinematics:

[0153] T=T 0 ·T 1 (θ 1 )·T 2 (θ 2 )…T n (θn )

[0154] Where T is the transformation matrix of the end effector, T n (θ n ) is the transformation matrix of the nth joint, θ n is the joint angle;

[0155] Inverse Kinematics:

[0156] Solve T = T 0 …T n (θ 1 ,θ 2 ,…,θ n )

[0157] 3) Kinetic model:

[0158] Newton–Euler method:

[0159] Calculate the forces and torques at each joint:

[0160] F net =m·a+F gravity +F friction

[0161] Among them, F net is the net force, m is the mass, a is the acceleration, F gravity is the influence of gravity, F friction is the friction force;

[0162] 4) Control model:

[0163] PID Controller:

[0164]

[0165] Where u(t) is the control input, e(t) is the error, and K p , K i , K d are the proportional, integral and derivative gains respectively, τ is the joint torque;

[0166] Model Predictive Control:

[0167] The model-based optimization control problem is formulated as:

[0168]

[0169] Among them, y k is the system output, y ref is the reference output, N is the number of prediction steps, and λ is the regularization parameter;

[0170] 5) Environmental interaction model:

[0171] Interaction force model:

[0172] Based on Hooke's law and friction model, the relationship between force and displacement is described as:

[0173]

[0174] Among them, F is the interaction force, k is the elastic coefficient, x is the displacement, and b is the damping coefficient;

[0175] 6) Data-driven modeling:

[0176] Machine Learning Models:

[0177] Use regression algorithms to construct a mapping relationship between input (sensor data) and output (robotic arm behavior):

[0178]

[0179] in, is the predicted value, X is the input feature, and θ is the model parameter.

[0180] Furthermore, the decision algorithm in step 3 specifically includes:

[0181] S1. Input feature selection:

[0182] Extract features from sensor data and digital twin models as input to the algorithm. These features include:

[0183] The current position information of the robot arm (joint angle, end effector position); current speed and acceleration; force feedback (such as grasping force, collision detection); environmental information (such as obstacle position, target object status)

[0184] S2. Target state definition:

[0185] Define the desired target state, including target position, target velocity, and desired interaction force;

[0186] S3. Gap calculation:

[0187] Calculate the error between the current state and the target state;

[0188] S4. Machine learning model training:

[0189] Use deep learning methods to train the decision model. Based on the gap between input features and target states, the model learns how to adjust the behavior of the robotic arm.

[0190] S5. Decision-making process:

[0191] Make real-time decisions based on trained models;

[0192] S6. Feedback mechanism:

[0193] Update and optimize the decision model in real time based on the actual performance of the robot arm and newly collected data;

[0194] S7. Risk Assessment and Management:

[0195] During the decision-making process, a risk assessment mechanism is added to make decision adjustments based on the uncertainty and potential risks of the current state to ensure the safety of the robot arm operation.

[0196] Among them, the gap calculation method in S3 includes:

[0197] e(t)=y target -y current

[0198] Where e(t) is the error, y target is the target state, y current is the current status.

[0199] Furthermore, the deep learning algorithm in S4 uses a deep Q network:

[0200]

[0201] Among them, Q(s t ,a t ) is in state s t Next, perform action a t the current estimated value of

[0202] α is the learning rate, which determines the degree to which new information affects existing knowledge;

[0203] γ is the discount factor, which represents the present value of future rewards;

[0204] r t is the current reward, r t =R(s t ,a t ).

[0205] Furthermore, real-time decision making in S5 includes:

[0206] Determine action: Choose action: Choose the best action a t To minimize the error:

[0207]

[0208] Execute action: Apply the selected action to the robot and update the current state.

[0209] Among them, the model update in S6:

[0210]

[0211] Among them, θ is the model parameter, η is the learning rate, and L(θ) is the loss function.

[0212] The timing analysis technology in step 4 specifically includes:

[0213] A1. Data collection and preprocessing:

[0214] Data collation: The historical operation data of the robot arm is collated into a time series data format, including timestamp, joint angle, end effector position, speed, and force feedback information;

[0215] Missing value processing: interpolation or deletion of missing data to ensure data integrity;

[0216] A2. Feature extraction:

[0217] Statistical features: Extract statistical features from time series data, such as mean, variance, peak, maximum, and minimum values;

[0218] Frequency domain features: Use fast Fourier transform to convert time series data into frequency domain and extract frequency features;

[0219] A3. Pattern recognition:

[0220] Cluster analysis: Use clustering algorithms to cluster historical data and identify successful and failed operation patterns;

[0221] Template matching: Create templates of successful and failed operations and use distance metrics to identify the similarity between new operations and the templates.

[0222] A4. Anomaly Detection:

[0223] Statistical-based methods: Use control charts to monitor changes in time series data and identify abnormal operations;

[0224] Based on machine learning methods: using isolation forest or one-class support vector machine for anomaly detection;

[0225] A5. Trend analysis:

[0226] Moving average method:

[0227]

[0228] Among them, MA t is the moving average at time point t, y is the time series data value, and n is the window size;

[0229] Exponential smoothing:

[0230] S t=αy t +(1-α)S t-1

[0231] Among them, S t is the smoothing value, α is the smoothing constant (0<α<1)

[0232] A6. Timing Modeling:

[0233] Autoregressive integrated moving average model:

[0234] Set the model parameters and perform simulation fitting:

[0235] y t =x+φ 1 y t-1 +…+φ p y t-p +θ 1 ε t-1 +…+θ q ε t-q +ε t

[0236] Among them, c is a constant term, φ and θ are model parameters, and ε is white noise;

[0237] Establish LSTM model for time series prediction:

[0238] h t =f(W h h t-1 +W x x t +b)

[0239] Among them, h t is the hidden state at the current moment, x t is the current input, W h and W x is the weight matrix and b is the bias.

[0240] Furthermore, timing analysis result evaluation and optimization include:

[0241] Performance metrics: Use mean squared error to evaluate model performance:

[0242]

[0243] Among them, y i is the actual value, is the predicted value;

[0244] Feedback mechanism: Based on the analysis results, feedback is given to the external control system to optimize subsequent operation strategies.

[0245] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A human-computer interaction adaptive method for digital twins, characterized in that: The following steps are involved: Step 1: Data collection and preprocessing: Sensor data collection: collect real-time data of the robot arm through machine vision, force sensors, and acceleration sensors, including position, speed, acceleration, and force feedback; Data preprocessing: denoising, standardizing and feature extraction of collected data to improve data quality and effectiveness of subsequent analysis; Step 2: Build a data twin model: Modeling: Building a digital twin model of the robotic arm based on the collected sensor data to simulate its performance under different working conditions; Step 3: Intelligent decision support system development: Decision algorithm design: Develop a decision algorithm based on machine learning that can analyze digital twin models and sensor data in real time to identify the gap between the current state and the target; Suggestion generation: The algorithm generates optimization suggestions based on the analysis results, such as adjusting the working mode of the robot arm and modifying operating parameters to improve efficiency and safety; Step 4: Learning optimization based on time series data: Time series data analysis: Use time series analysis technology to conduct in-depth analysis of the robot's historical operation data to identify successful operation patterns and failure cases; Step 5: Interactive interface design: Dynamic interactive interface: Develop a user-friendly dynamic interactive interface to display the digital twin status, recommended actions and potential problems in real time; Step 6: Real-time monitoring and adjustment: Real-time monitoring system: Realize real-time monitoring of the robot arm and its operating environment, obtain key data in a timely manner and feed it back to the intelligent decision support system; Dynamically adjust strategies: Based on real-time monitoring, the system can automatically adjust decision-making strategies to respond to emergencies or environmental changes.

2. The human-computer interaction adaptive method for digital twins according to claim 1 is characterized in that: The digital twin model of the robotic arm constructed in step 2 includes: 1) Geometric model: Description: Use CAD software or modeling tools to build a 3D geometric model of the robotic arm, defining the size, shape, and connection relationship of each component; 2) Kinematic model: Forward Kinematics: T=T0·T1(θ1)·T2(θ2)…T n (i n ) Where T is the transformation matrix of the end effector, T n (θ n ) is the transformation matrix of the nth joint, θ n is the joint angle; Inverse Kinematics: Solve for T = T0…T n (θ1, θ2, …, θ n ) 3) Kinetic model: Newton-Euler method: Calculate the forces and torques at each joint: F net =m·a+F grauity +F friction Among them, F net is the net force, m is the mass, a is the acceleration, F gravity is the influence of gravity, F friction is the friction force; 4) Control model: PID Controller: Where u(t) is the control input, e(t) is the error, and K p , K i , K d are the proportional, integral and derivative gains respectively, τ is the joint torque; Model Predictive Control: The model-based optimization control problem is formulated as: Among them, y k is the system output, y ref is the reference output, N is the number of prediction steps, and λ is the regularization parameter; 5) Environmental interaction model: Interaction force model: Based on Hooke's law and friction model, the relationship between force and displacement is described as: Among them, F is the interaction force, k is the elastic coefficient, x is the displacement, and b is the damping coefficient; 6) Data-driven modeling: Machine Learning Models: Use regression algorithms to construct a mapping relationship between input (sensor data) and output (robotic arm behavior): in, is the predicted value, X is the input feature, and θ is the model parameter.

3. The human-computer interaction adaptive method for digital twins according to claim 2 is characterized in that: The decision algorithm in step 3 specifically includes: S1. Input feature selection: Extract features from sensor data and digital twin models as input to the algorithm. These features include: The current position information of the robot arm (joint angle, end effector position); current speed and acceleration; force feedback (such as grasping force, collision detection); environmental information (such as obstacle position, target object status) S2. Target state definition: Define the desired target state, including target position, target velocity, and desired interaction force; S3. Gap calculation: Calculate the error between the current state and the target state; S4. Machine learning model training: Use deep learning methods to train the decision model. Based on the gap between input features and target states, the model learns how to adjust the behavior of the robotic arm. S5. Decision-making process: Make real-time decisions based on trained models; S6. Feedback mechanism: Update and optimize the decision model in real time based on the actual performance of the robot arm and newly collected data; S7. Risk Assessment and Management: During the decision-making process, a risk assessment mechanism is added to make decision adjustments based on the uncertainty and potential risks of the current state to ensure the safety of the robot arm operation.

4. The human-computer interaction adaptive method for digital twins according to claim 3 is characterized in that: The method for calculating the gap in S3 includes: e(t)=y target -and current Where e(t) is the error, y target is the target state, y current is the current status.

5. The human-computer interaction adaptive method for digital twins according to claim 4 is characterized in that: The deep learning algorithm in S4 uses a deep Q network: Among them, Q(s t , a t ) is in state s t Next, perform action a t the current estimated value of α is the learning rate, which determines the degree to which new information affects existing knowledge; γ is the discount factor, which represents the present value of future rewards; r t is the current reward, r t =R(s t , a t ).

6. The human-computer interaction adaptive method for digital twins according to claim 5 is characterized in that: The real-time decision making in S5 includes: Determine action: Choose action: Choose the best action a t To minimize the error: Execute action: Apply the selected action to the robot and update the current state.

7. The human-computer interaction adaptive method for digital twins according to claim 3 is characterized in that: The S6 model updates: Among them, θ is the model parameter, η is the learning rate, and L(θ) is the loss function.

8. The human-computer interaction adaptive method for digital twins according to claim 1, characterized in that: The timing analysis technology in step 4 specifically includes: A1. Data collection and preprocessing: Data collation: The historical operation data of the robot arm is collated into a time series data format, including timestamp, joint angle, end effector position, speed, and force feedback information; Missing value processing: interpolation or deletion of missing data to ensure data integrity; A2. Feature extraction: Statistical features: Extract statistical features from time series data, such as mean, variance, peak, maximum, and minimum values; Frequency domain features: Use fast Fourier transform to convert time series data into frequency domain and extract frequency features; A3. Pattern recognition: Cluster analysis: Use clustering algorithms to cluster historical data and identify successful and failed operation patterns; Template matching: Create templates of successful and failed operations and use distance metrics to identify the similarity between new operations and the templates. A4. Anomaly Detection: Statistical-based methods: Use control charts to monitor changes in time series data and identify abnormal operations; Based on machine learning methods: using isolation forest or one-class support vector machine for anomaly detection; A5. Trend analysis: Moving average method: Among them, MA t is the moving average at time point t, y is the time series data value, and n is the window size; Exponential smoothing: S t =αy t +(1-α)S t-1 Among them, S t is the smoothing value, α is the smoothing constant (0<α<1) A6. Timing Modeling: Autoregressive integrated moving average model: Set the model parameters and perform simulation fitting: y t =c+φ1y t-1 +…+φ p y t-p +θ1ε t-1 +…+θ q e t-q +e t Among them, c is a constant term, φ and θ are model parameters, and ε is white noise; Establish LSTM model for time series prediction: h t =f(W h h t-1 +W x x t +b) Among them, h t is the hidden state at the current moment, x t is the current input, W h and W x is the weight matrix and b is the bias.

9. The human-computer interaction adaptive method for digital twins according to claim 8 is characterized in that: The timing analysis result evaluation and optimization includes: Performance metrics: Use mean squared error to evaluate model performance: Among them, yi is the actual value, is the predicted value; Feedback mechanism: Based on the analysis results, feedback is given to the external control system to optimize subsequent operation strategies.

Images