Robot joint angle calibration method and system

Abstract

The invention discloses a robot joint angle calibration method and system, and relates to the technical field of robotics.The calibration method comprises the specific steps that S100, sensors are installed and data are collected, specifically, the sensors are installed at all joints of a robot, real-time compensation of thermomagnetic interference is achieved through intelligent self-adaptive model adjustment, and the calibration accuracy is improved; the method effectively reduces the influence of environmental factors on the joint angle of the robot, not only considers the influence of temperature on the joint angle, but also takes magnetic field interference into consideration, and can accurately describe the joint angle error of the robot under different working conditions by constructing a coupling model of thermomagnetic coupling and joint angle error, thereby improving the accuracy of the joint angle error. A theoretical basis is provided for subsequent thermomagnetic compensation, and meanwhile, an intelligent adaptive model adjustment mechanism enables the method to continuously learn and optimize model parameters, so that the method continuously adapts to a new environment and new conditions in the operation process of the robot, and the accuracy and reliability of calibration are further improved.

Description

Technical Field

[0001] The present invention relates to the field of robot technology, and in particular to a robot joint angle calibration method and system. Background Art

[0002] In the field of modern industrial automation, the rapid development and increasingly widespread application of robotics technology have put forward higher requirements on the robot's motion accuracy and stability, especially in the fields of high-precision manufacturing and precision assembly. The robot's joint angle calibration has become a key link to ensure its performance. However, since the robot is affected by a variety of environmental factors during actual operation, such as temperature changes and magnetic field interference, these factors often cause errors in the robot's joint angles, thereby affecting its motion accuracy and stability. Therefore, how to effectively calibrate the robot's joint angles to eliminate or reduce the impact of environmental factors on its motion accuracy has become an important issue that needs to be urgently solved in the current field of robotics technology.

[0003] Traditional robot joint angle calibration methods are mostly based on static or simple dynamic models, and often ignore the complex influence of environmental factors on the robot joint angle. For example, traditional calibration methods may only consider the effect of temperature on the thermal expansion of joint materials, but ignore the influence of magnetic field interference on the accuracy of joint sensors. In addition, traditional calibration methods often lack real-time and adaptability, and cannot be dynamically adjusted according to real-time environmental changes during the operation of the robot. Therefore, traditional methods often find it difficult to achieve ideal calibration effects under complex working conditions, which limits the application scope and motion performance of the robot.

[0004] Therefore, developing a robot joint angle calibration method and system will help promote the further development of robot technology and improve the level of industrial automation in my country. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the prior art and to provide a robot joint angle calibration method and system, which integrates multi-sensor data acquisition, intelligent adaptive model adjustment and predictive thermal magnetic compensation technology, and monitors and compensates for the impact of ambient temperature and magnetic field changes on joint angles in real time, thereby significantly improving the robot's motion accuracy and stability under complex working conditions. It not only optimizes the calibration process and improves the calibration efficiency, but also ensures the long-term reliability of the calibration system through continuous model updating and maintenance.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: On the one hand, a robot joint angle calibration method, the specific steps of the calibration method are:

[0007] S100, sensor installation and data acquisition: install sensors at each joint of the robot, select a data acquisition card with multiple analog input channels, set the sampling frequency, use the sampling program to collect sensor data in real time, and transmit the data to the robot control system through the USB3.0 interface;

[0008] S200, initial coupling model construction: build a simulation environment, simulate the working state of the robot joints under different working conditions, collect the joint angle error data under different thermal and magnetic conditions, build a coupling model of thermal and magnetic coupling and joint angle error based on the error data, and optimize the model parameters;

[0009] S300, intelligent adaptive model adjustment: continuously obtain real-time thermal and magnetic data and compensated joint angle feedback information, measure the difference between the model prediction and the actual measured joint angle through the loss function, update the model parameters through back propagation based on the difference, and regularly update and optimize the model according to changes in data characteristics;

[0010] S400, predictive thermal-magnetic compensation execution: collect historical thermal-magnetic data for preprocessing and feature extraction, build a prediction model and train it, receive real-time data and preprocess it, input it into the prediction model to predict short-term thermal-magnetic changes, calculate the pre-compensated joint angle error using the coupling model based on the predicted thermal-magnetic changes, and correct the error in advance;

[0011] S500, calibration process cycle and optimization: The robot starts and performs the above operations in sequence. During operation, the above process is continuously cycled to regularly inspect, maintain and optimize the entire calibration system.

[0012] Further, in the S100, sensors used in sensor installation and data collection and installation positions of the sensors;

[0013] Temperature sensor: Use a thermocouple temperature sensor and install it on the robot joint housing;

[0014] Magnetic field sensor: A Hall effect magnetic field sensor is used and installed inside the joint near the angle sensor;

[0015] Angle sensor: installed at the joints of the robot.

[0016] Furthermore, in the S200, the coupling model of thermal magnetic coupling and joint angle error is constructed in the initial coupling model construction, and the calculation formula is: Δθ=Δθ T +Δθ B +γ·Δθ T ·Δθ B , where Δθ is the total joint angle error, γ is the thermal-magnetic coupling interaction coefficient, and Δθ T is the joint angle error caused by thermal expansion, ΔθB Joint angle error Δθ caused by magnetic field interference B .

[0017] Furthermore, in the S200, the joint angle error Δθ caused by thermal expansion in the coupling model during the initial coupling model construction is T The calculation formula is: Δθ T =α T ·ΔT·(1+β T ·sin(ω T t+φ T ))·θ0, where α T is the coefficient of thermal expansion, ω T is the angular frequency, φ T is the initial phase for periodic changes, ΔT is the change in joint temperature, θ0 is the initial joint angle, β T is the modulation coefficient of the thermal expansion coefficient, t is time, ω T is the angular frequency of the change in thermal expansion coefficient; the joint angle error Δθ caused by magnetic field interference B The calculation formula is: Δθ B =α B ·B·(1+β B ·cos(ω B t+φ B ))·θ0, where α B is the magnetic field interference coefficient, B is the magnetic field strength, β B is the modulation coefficient of the magnetic field interference coefficient, ω B is the angular frequency of the change of the magnetic field interference coefficient, φ B is the initial phase of the change of the magnetic field interference coefficient, and θ0 is the initial joint angle.

[0018] Furthermore, in the intelligent adaptive model adjustment in S300, the difference between the model prediction and the actual measured joint angle is measured by the loss function, and the joint angle predicted by the model is θ model in is the model parameter vector, ΔT is the change in joint temperature, B is the magnetic field intensity, and the actual measured joint angle is θ measured , loss function The calculation formula is: Where N is the number of data samples, and are the actual measured joint angle and the model predicted joint angle of the ith sample, ω is the model parameter vector, ΔT is the change in joint temperature, and B is the magnetic field intensity.

[0019] Furthermore, in the S300, the specific steps of updating the model parameters by back-propagating the gradient according to the difference between the predicted and the actually measured joint angles in the intelligent adaptive model adjustment are: and the input thermomagnetic data (ΔT i ,B i ), calculate the joint angles predicted by the model Compare it with the actual measured joint angle Substitute into the loss function In the loss function The calculation formula is: in, is the ith measured joint angle, Is to use the current parameters and thermomagnetic data (ΔT i ,B i ) The model predicts the joint angle calculated by the coupling model, and gradually propagates back to the model from the loss function to calculate the loss function in the current model parameter vector ω k The gradient at Update the model parameter vector according to the gradient descent formula get The formula is: in is the model parameter vector, is the updated model parameter vector, η is the learning rate, is the loss function In the parameter vector The gradient at .

[0020] Furthermore, in the predictive thermal magnetic compensation in S400, the temperature change in the short term in the future is predicted by the thermal compensation model, and the historical temperature sequence is T1, T2, ..., T m , predict the temperature T at the next Δt time m+Δt The calculation formula is: where α j is the autoregressive term coefficient, β j is the sliding average coefficient, ∈ m-j is the prediction error at the mjth moment, that is, the difference between the actual temperature and the predicted temperature, δ is the cross-term coefficient, p, q, r are the model orders, ΔT m-j is the temperature change at the mjth moment, and the formula is: ΔT m-j =T m-j -T m-j-1 .

[0021] Furthermore, in the step S400, the predictive thermal magnetic compensation is performed by using a magnetic compensation model to predict the magnetic field change in the short term in the future, and the calculation formula is: where γ j is the magnetic field autoregressive term coefficient, δ j is the coefficient of the magnetic field sliding mean term, ∈′ m-j is the magnetic field prediction error at the mjth moment, and μ is the magnetic field cross-term coefficient.

[0022] On the other hand, a robot joint angle calibration system is applicable to a robot joint angle calibration method according to claims 1 to 8, the system comprising a sensor module, a data acquisition and transmission module, a model building and adjustment module, a predictive compensation module and a calibration process management module;

[0023] The sensor module: a sensor is installed at each joint of the robot, wherein the temperature sensor is a thermocouple temperature sensor, which is installed on the robot joint housing; the magnetic field sensor is a Hall effect magnetic field sensor, which is installed inside the joint near the angle sensor; the angle sensor is installed at the robot joint, which is responsible for collecting joint temperature, magnetic field strength and angle information respectively;

[0024] The data acquisition and transmission module: uses a data acquisition card with multiple analog input channels, sets the sampling frequency according to the robot joint movement and environmental changes, converts the sensor analog signal into a digital signal through the sampling program, and transmits it to the robot control system through the USB3.0 interface;

[0025] The model building and adjustment module: constructs a simulation environment, simulates the working state of the robot joint under different working conditions, collects the joint angle error data under different thermal and magnetic conditions, builds a coupling model of thermal and magnetic coupling and joint angle error based on the error data and optimizes the parameters; continuously obtains real-time thermal and magnetic and compensated joint angle feedback information, uses the loss function to measure the difference, calculates the gradient through back propagation, and updates the model parameters and optimizes them regularly according to the data characteristics;

[0026] The predictive compensation module collects historical thermal and magnetic data, predicts temperature and magnetic field changes according to the thermal compensation model and the magnetic compensation model, calculates the pre-compensated joint angle error according to the coupling model, and corrects the error in advance;

[0027] The calibration process management module: after the robot is started, each module is started in sequence and the process is circulated, each part of the system is regularly inspected, problems are maintained and repaired, and parameter algorithms are optimized.

[0028] Compared with the prior art, this robot joint angle calibration method and system has the following beneficial effects:

[0029] 1. The present invention realizes real-time compensation for thermal and magnetic interference through intelligent adaptive model adjustment, effectively reducing the influence of environmental factors on the joint angle of the robot. The present invention not only considers the influence of temperature on the joint angle, but also takes magnetic field interference into consideration. By constructing a coupling model of thermal and magnetic coupling and joint angle error, it can accurately describe the joint angle error of the robot under different working conditions, providing a theoretical basis for subsequent thermal and magnetic compensation. At the same time, the intelligent adaptive model adjustment mechanism enables the method to continuously learn and optimize model parameters, thereby continuously adapting to new environments and new conditions during the operation of the robot, further improving the accuracy and reliability of the calibration.

[0030] 2. The present invention realizes the advance adjustment of the robot's joint angle error by introducing a predictive thermal-magnetic compensation execution mechanism, thereby significantly improving the robot's operating efficiency and performance. This method uses historical thermal-magnetic data to build a prediction model, which can accurately predict the trend of thermal-magnetic interference in the short term in the future. According to the prediction results, the pre-compensated joint angle error can be calculated in advance, and the robot's joint angle can be automatically adjusted, thereby avoiding the motion error caused by thermal-magnetic interference. It not only reduces the number and time of adjustments during the operation of the robot, but also significantly improves the robot's motion accuracy and stability. In addition, the predictive thermal-magnetic compensation execution mechanism can effectively extend the service life of the robot, reduce joint wear and failure rate caused by frequent adjustments, and provide a strong guarantee for the long-term stable operation of the robot.

[0031] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 It is a framework diagram of a robot joint angle calibration method;

[0034] Figure 2 This is a flow chart of a robot joint angle calibration system. DETAILED DESCRIPTION

[0035] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0036] Embodiment 1:

[0037] Industrial robot arm joint angle calibration

[0038] In the welding workshop of an automobile manufacturing plant, there is an industrial robotic arm used to weld automobile parts.

[0039] Sensor installation and data acquisition (S100): Install thermocouple temperature sensors on the outer shells of each joint of the robot arm to measure the temperature changes at the joints; install Hall effect magnetic field sensors inside the joints near the angle sensors; ensure that the angle sensors are normally installed at the joints, select a data acquisition card with multiple analog input channels, set the sampling frequency to 100 Hz, collect sensor data in real time through the sampling program, and use the USB3.0 interface to stably transmit the data to the control system of the robot arm.

[0040] Initial coupling model construction (S200): Construct an environment that simulates the welding conditions of automotive parts, simulate the working conditions of the robot arm joints under different welding tasks and different working times, collect the joint angle error data under different thermal magnetic conditions generated during the welding process, such as data under the conditions of welding current changes, changes in the surrounding electromagnetic field environment, and long-term joint movement and heating. Combine the error data of mechanical engineering and electromagnetics to construct a coupling model of thermal magnetic coupling and joint angle error. The calculation formula is: Δθ=Δθ T +Δθ B +γ·Δθ T ·Δθ B , where Δθ is the total joint angle error, γ is the thermal-magnetic coupling interaction coefficient, and Δθ T is the joint angle error caused by thermal expansion, the formula is: Δθ T =α T ·ΔT·(1+β T ·sin(ω T t+φ T ))·θ0,α T is the coefficient of thermal expansion, ω T is the angular frequency, φ T is the initial phase for periodic changes, ΔT is the change in joint temperature, θ0 is the initial joint angle, β T is the modulation coefficient of the thermal expansion coefficient, t is time, ω T is the angular frequency of the change in thermal expansion coefficient, the joint angle error Δθ caused by magnetic field interference B , the calculation formula is: ΔθB =α B ·B·(1+β B ·cos(ω B t+φ B ))·θ0,α B is the magnetic field interference coefficient, B is the magnetic field strength, β B is the modulation coefficient of the magnetic field interference coefficient, ω B is the angular frequency of the change of the magnetic field interference coefficient, φ B It is the initial phase of the change of the magnetic field interference coefficient. By analyzing and fitting a large amount of collected data, optimizing the model parameters, and determining the specific values ​​of the thermal expansion coefficient, magnetic field interference coefficient, and thermal-magnetic coupling interaction coefficient in this scenario.

[0041] Intelligent adaptive model adjustment (S300): In the daily welding work of the robot arm, real-time thermal magnetic data and compensated joint angle feedback information are continuously obtained. Each time a welding task is completed (for example, welding a car door frame), the difference between the model prediction and the actual measured joint angle is measured through the loss function based on this batch of data. Let the joint angle predicted by the model be θ model in is the model parameter vector, and the actual measured joint angle is θ measured , the loss function is: Where N is the number of data samples, and are the actual measured joint angle of the i-th sample and the model predicted joint angle, ω is the model parameter vector, ΔT is the change in joint temperature, B is the magnetic field strength, and the model parameters are updated by back-propagating the gradient according to the difference. According to the rules of the back-propagation algorithm, the loss function is gradually back-propagated to the model to calculate the loss function in the current model parameter vector ω k The gradient at And update the model parameter vector according to the gradient descent formula to get The formula is: in is the model parameter vector, is the updated model parameter vector, η is the learning rate, is the loss function In the parameter vector The gradient at the location is calculated, and the model is updated and optimized according to the changes in data features.

[0042] Predictive thermal magnetic compensation execution (S400): collect historical thermal magnetic data of the robot arm during the welding process in the past week, preprocess and extract features, build a prediction model and use historical data for training, for example, establish a prediction model of temperature and magnetic field changes through time series analysis method, in subsequent welding tasks, receive real-time data processing and input the model to predict the trend of thermal magnetic interference in the short term, and predict the temperature change in the future in the short term through the thermal compensation model. Assume that the historical temperature sequence is T1, T2, …, T m , predict the temperature T at the next Δt time m+Δt The calculation formula is: where α j is the autoregressive term coefficient, β j is the sliding average coefficient, ∈ m-j is the prediction error at the mjth moment, that is, the difference between the actual temperature and the predicted temperature, δ is the cross-term coefficient, p, q, r are the model orders, ΔT m-j is the temperature change at the mjth moment, and the formula is: ΔT m-j =T m-j -T m-j-1 , the magnetic field changes in the short term in the future are predicted by the magnetic compensation model, and the calculation formula is: where γ j is the magnetic field autoregressive term coefficient, δ j is the coefficient of the magnetic field sliding mean term, ′ m-j is the magnetic field prediction error at the mjth moment, μ is the magnetic field cross-term coefficient, and the pre-compensated joint angle error is calculated using the coupling model according to the predicted thermal magnetic change, and the robot arm joint control parameters are adjusted in advance to improve the welding accuracy.

[0043] Calibration process loop and optimization (S500): After the robot arm is started every day, the above operations are performed in sequence. During the entire production shift, the above process is continuously cycled. After get off work every day, the entire calibration system is inspected and maintained to check whether the sensor is working properly and whether the data transmission is stable. At the same time, according to the production data of the day and the operation of the model, the system parameters and algorithms are optimized, such as adjusting the sampling frequency and optimizing the model structure, to adapt to the ever-changing welding conditions and improve the accuracy of joint angle calibration.

[0044] To sum up, in the application scenario of industrial robot arms in automobile manufacturing welding workshops, the present invention realizes effective data collection by reasonably installing thermocouple temperature sensors, Hall effect magnetic field sensors and angle sensors at the joints, and uses specific sampling frequencies and data transmission methods to provide a data basis for subsequent model construction. The constructed thermal-magnetic coupling and joint angle error coupling models can accurately reflect the working status of the robot arm after optimizing parameters through data collection under simulated working conditions. The intelligent adaptive model adjustment link updates parameters according to the loss function and back propagation algorithm to ensure model accuracy. Predictive thermal-magnetic compensation execution can respond to thermal-magnetic interference in advance. Combined with the calibration process cycle and optimization, the welding accuracy and stability of the robot arm can be continuously improved to meet the high-precision requirements of automobile manufacturing.

[0045] Embodiment 2:

[0046] Joint Angle Calibration of Logistics Handling Robot

[0047] In a large logistics warehouse, there is a group of robots used to move goods.

[0048] Sensor installation and data acquisition (S100): Install thermocouple temperature sensors on the outer shell of each joint of the logistics handling robot. Because the ambient temperature in the warehouse may change due to factors such as the cargo storage layout and the operation of the ventilation equipment, the joint temperature needs to be monitored; install Hall effect magnetic field sensors inside the joints near the angle sensors, taking into account the magnetic field interference generated by some electrical equipment in the warehouse; angle sensors are installed at the joints to obtain joint angle information, select a suitable data acquisition card, set the sampling frequency to 50Hz, collect sensor data through the sampling program, and transmit it to the robot control system via the USB3.0 interface to ensure the real-time and accuracy of the data.

[0049] Initial coupling model construction (S200): Simulate the working state of the logistics handling robot when carrying goods of different weights and shapes in different areas of the warehouse, collect the joint angle error data under the thermal magnetic environment generated by the operation of the warehouse's lighting system and electrical equipment, and build a coupling model of thermal magnetic coupling and joint angle error by combining robot kinematics and electromagnetic theory. The calculation formula is: Δθ=Δθ T +Δθ B +γ·Δθ T ·Δθ B For example, when transporting large metal goods, they may be affected by strong magnetic fields. Through a large number of experiments and data analysis, the various parameters in the model are determined and the model is optimized to adapt to the logistics and handling scenarios.

[0050] Intelligent adaptive model adjustment (S300): During the robot's daily cargo handling process, the robot continuously obtains thermal magnetic data and compensated joint angle feedback information. Every time the robot completes a handling task (such as carrying a box of standard-sized goods from the shelf area to the sorting area), the loss function is used to calculate the difference between the model prediction and the actual measured joint angle. The joint angle predicted by the model is θ model in is the model parameter vector, and the actual measured joint angle is θ measured , the loss function is: Update the model parameters by calculating the gradient through back propagation, according to the current model parameters and the input thermomagnetic data (ΔT i ,B i ), calculate the joint angles predicted by the model Compare it with the actual measured joint angle Substitute into the loss function In the loss function The calculation formula is: Starting from the loss function, we gradually propagate backwards to the model and calculate the loss function in the current model parameter vector ω k The gradient at Update the model parameter vector according to the gradient descent formula get The formula is: The model is optimized according to the changes in data features to ensure that the model can accurately reflect the changes in the joint angles of the robot in actual work.

[0051] Predictive thermal magnetic compensation execution (S400): collect historical thermal magnetic data of the robot working in the warehouse in the past month for preprocessing and feature extraction, build a prediction model and train it, and when the robot is running, receive real-time data for processing and input it into the model to predict the thermal magnetic interference trend in the short term (such as within the next 30 minutes), and predict the temperature change in the future in the short term through the thermal compensation model. Assume that the historical temperature sequence is T1, T2, …, T m , predict the temperature T at the next Δt time m+Δt The calculation formula is:

[0052] The magnetic field changes in the short term in the future are predicted by the magnetic compensation model. The calculation formula is: Based on the prediction results, the coupling model is used to calculate the pre-compensated joint angle error, and the robot joint control instructions are adjusted in advance to ensure the accuracy and stability of the handling process and avoid goods falling or colliding with shelves.

[0053] Calibration process cycle and optimization (S500): The logistics handling robot executes the calibration process after starting up every day, and it is cyclically carried out during the whole day's work. The entire calibration system is fully inspected and maintained every week, including checking the accuracy of the sensor and the performance of the data acquisition card. According to the operating data and model performance within a week, the system parameters and algorithms are optimized, such as adjusting the sampling frequency, improving the model training method, and improving the accuracy and reliability of the robot's joint angle calibration to adapt to the complex and changeable working environment of the logistics warehouse.

[0054] To summarize, for the handling robot scenario in a large logistics warehouse, sensors are installed at appropriate positions on the joint casing and inside, and the acquisition card collects and transmits data at a set frequency. The constructed model combines the actual logistics working conditions and error data to accurately reflect the impact of thermal magnetism on joint angles. During the execution of the handling task, the intelligent adaptive model is adjusted based on the loss function and gradient update mechanism to make the model fit the robot's operating rules. Predictive thermal magnetic compensation uses historical data to train the model, calculate the thermal magnetic changes and joint angle error pre-compensation values ​​in advance, and improve handling accuracy. Through the calibration process cycle and optimization, the system is inspected and maintained weekly, and the parameter algorithm is optimized to ensure that the robot can operate stably in a complex warehouse environment and improve logistics handling efficiency and quality.

[0055] Embodiment three:

[0056] Application of industrial robotic arms in production scenarios

[0057] In a modern automobile manufacturing plant, high-precision robotic arms are widely used in the welding and assembly of automobile parts. During the operation of one of the robotic arms, the accuracy of its joint angle was seriously affected due to the presence of a large number of heat-generating equipment and complex electromagnetic environment in the workshop.

[0058] To address this problem, the robot joint angle calibration system of the present invention is adopted. At each joint of the robotic arm, the thermocouple temperature sensor in the sensor module is tightly fitted on the joint housing to constantly monitor the temperature fluctuations of the joint caused by heat generated by equipment operation and changes in ambient temperature; the Hall effect magnetic field sensor is precisely installed inside the joint close to the angle sensor to effectively capture changes in the surrounding magnetic field strength; the angle sensor is precisely installed at the key position of the joint rotation to accurately obtain the initial angle data of the joint.

[0059] The multi-analog input channel data acquisition card of the data acquisition and transmission module sets the sampling frequency to 500Hz according to the rapid and frequent joint movements of the robot arm and the complex and changeable thermal and magnetic environment of the workshop. Through an efficient sampling program, it quickly converts the analog signals from each sensor into digital signals, and transmits them to the control system of the robot arm stably and at high speed through the USB3.0 interface, ensuring that the control system can obtain real-time status information of the joints in a timely manner.

[0060] The model building and adjustment module first constructs a simulation environment that highly simulates the actual working scene of the robotic arm. In this environment, the temperature range in the workshop from 20℃ to 80℃, the fluctuation of magnetic field strength between 0.1T and 1T, and the working conditions of the robotic arm under different loads and movement speeds are accurately simulated. A large amount of robotic arm joint angle error data under these different thermal and magnetic conditions is collected to build a coupling model of thermal and magnetic coupling and joint angle error, and the model parameters are finely optimized through complex data analysis and optimization algorithms. In the subsequent operation of the robotic arm, real-time thermal and magnetic data and compensated joint angle feedback information are continuously obtained, and a specially designed loss function is used to strictly measure the difference between the model prediction and the actual measured joint angle. According to the precise calculation process of the back propagation algorithm, the gradient is accurately calculated and the model parameters are updated. At regular intervals, the model is comprehensively updated and optimized according to the changes in data features caused by the adjustment of the robotic arm's work tasks or significant changes in the workshop environment, so as to always maintain the model's high-precision prediction capability.

[0061] The predictive compensation module carefully collects historical thermal and magnetic data during the operation of the robot arm and stores it in a special database. Through the thermal compensation model and the magnetic compensation model, it accurately predicts the future temperature and magnetic field changes at the robot arm joints. For example, in a long-term continuous welding task, based on the changing trend of the historical temperature data and the current working status, it is predicted in advance that the joint temperature will rise by 20°C in the next 10 minutes. At the same time, the magnetic field strength will be increased by 0.3T due to the newly started large-scale electromagnetic equipment nearby. Then, the pre-compensated joint angle error is quickly calculated based on the coupling model, and the compensation command is sent to the robot arm joint controller in time, so that the joint can make precise adjustments in advance, effectively avoiding the adverse effects of thermal and magnetic interference on joint angle measurement and control.

[0062] The calibration process management module starts the above modules in strict order every time the robot arm is started, and continuously cycles the calibration process throughout the production process. Every 24 hours, the entire calibration system is comprehensively and carefully tested, and the accuracy of the sensors is calibrated and verified using high-precision standard measuring instruments to ensure that the measurement errors of temperature sensors, magnetic field sensors, and angle sensors are always controlled within a very small range; the data transmission stability of the data acquisition and transmission module is checked using professional data monitoring software to ensure that there is no packet loss or delay in data transmission; the model prediction accuracy of the model construction and adjustment module is strictly evaluated through a series of test data sets, and the distribution of prediction errors is analyzed. The system continuously optimizes and improves the system parameters and algorithms based on the data and experience accumulated from the long-term operation of the robot arm, ensuring that the entire calibration system can continuously provide the robot arm with high-precision joint angle calibration results on the basis of long-term stable operation, thereby ensuring that the robot arm can accurately and efficiently complete the welding and assembly tasks in the automobile manufacturing process, greatly improving the quality and efficiency of automobile production.

[0063] To sum up, in the application scenario of the robotic arm in the automobile manufacturing plant, the robot joint angle calibration system of the present invention plays a key role. The sensor module accurately collects the joint temperature, magnetic field strength and angle information. The data acquisition and transmission module ensures the timely and accurate transmission of data. The model construction and adjustment module continuously optimizes to adapt to environmental changes. The predictive compensation module responds to thermal and magnetic interference in advance. The calibration process management module ensures the stable operation of the system. The system effectively overcomes the influence of the complex thermal and magnetic environment of the workshop on the joint angle of the robotic arm, significantly improves the working accuracy and stability of the robotic arm, and effectively promotes the high-quality development of automobile production. It has important application value and broad promotion prospects in the field of industrial robots.

[0064] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A robot joint angle calibration method, characterized in that: The specific steps of this calibration method are: S100, sensor installation and data acquisition: install sensors at each joint of the robot, select a data acquisition card with multiple analog input channels, set the sampling frequency, use the sampling program to collect sensor data in real time, and transmit the data to the robot control system through the USB3.0 interface; S200, initial coupling model construction: build a simulation environment, simulate the working state of the robot joints under different working conditions, collect the joint angle error data under different thermal and magnetic conditions, build a coupling model of thermal and magnetic coupling and joint angle error based on the error data, and optimize the model parameters; S300, intelligent adaptive model adjustment: continuously obtain real-time thermal and magnetic data and compensated joint angle feedback information, measure the difference between the model prediction and the actual measured joint angle through the loss function, update the model parameters through back propagation based on the difference, and regularly update and optimize the model according to changes in data characteristics; S400, predictive thermal-magnetic compensation execution: collect historical thermal-magnetic data for preprocessing and feature extraction, build a prediction model and train it, receive real-time data and preprocess it, input it into the prediction model to predict short-term thermal-magnetic changes, calculate the pre-compensated joint angle error using the coupling model based on the predicted thermal-magnetic changes, and correct the error in advance; S500, calibration process cycle and optimization: The robot starts and performs the above operations in sequence. During operation, the above process is continuously cycled to regularly inspect, maintain and optimize the entire calibration system.

2. A robot joint angle calibration method according to claim 1, characterized in that: S100, sensors used in sensor installation and data collection and installation positions of the sensors; Temperature sensor: Use a thermocouple temperature sensor and install it on the robot joint housing; Magnetic field sensor: A Hall effect magnetic field sensor is used and installed inside the joint near the angle sensor; Angle sensor: installed at the joint of the robot.

3. A robot joint angle calibration method according to claim 1, characterized in that: In the S200, the coupling model of thermal magnetic coupling and joint angle error is constructed in the initial coupling model construction, and the calculation formula is: Δθ=Δθ T +Δθ B +γ·Δθ T ·Δθ B , where Δθ is the total joint angle error, γ is the thermal-magnetic coupling interaction coefficient, and Δθ T is the joint angle error caused by thermal expansion, Δθ B Joint angle error Δθ caused by magnetic field interference B .

4. A robot joint angle calibration method according to claim 3, characterized in that: S200, joint angle error Δθ caused by thermal expansion in the coupling model during initial coupling model construction T The calculation formula is: Δθ T =α T ·ΔT·(1+β T ·sin(ω T t+φ T ))·θ0, where α T is the coefficient of thermal expansion, ω T is the angular frequency, φ T is the initial phase for periodic changes, ΔT is the change in joint temperature, θ0 is the initial joint angle, β T is the modulation coefficient of the thermal expansion coefficient, t is time, ω T is the angular frequency of the change in thermal expansion coefficient; the joint angle error Δθ caused by magnetic field interference B The calculation formula is: Δθ B =α B ·B·(1+β B ·cos(ω B t+φ B ))·θ0, where α B is the magnetic field interference coefficient, B is the magnetic field strength, β B is the modulation coefficient of the magnetic field interference coefficient, ω B is the angular frequency of the change of the magnetic field interference coefficient, φ B is the initial phase of the change of the magnetic field interference coefficient, and θ0 is the initial joint angle.

5. A robot joint angle calibration method according to claim 1, characterized in that: In the S300, the difference between the model prediction and the actual measured joint angle is measured by the loss function in the intelligent adaptive model adjustment. The joint angle predicted by the model is assumed to be in is the model parameter vector, ΔT is the change in joint temperature, B is the magnetic field intensity, and the actual measured joint angle is θ measured , loss function The calculation formula is: Where N is the number of data samples, and are the actual measured joint angle and the model predicted joint angle of the ith sample, ω is the model parameter vector, ΔT is the change in joint temperature, and B is the magnetic field intensity.

6. A robot joint angle calibration method according to claim 1, characterized in that: In the S300, the specific steps of updating the model parameters by back-propagating the gradient according to the difference between the predicted and the actually measured joint angles in the intelligent adaptive model adjustment are: and the input thermomagnetic data (ΔT i ,B i ), calculate the joint angles predicted by the model Compare it with the actual measured joint angle Substitute into the loss function In the loss function The calculation formula is: in, is the ith measured joint angle, Is to use the current parameters and thermomagnetic data (ΔT i ,B i ) The model predicts the joint angle calculated by the coupling model, and gradually propagates back to the model from the loss function to calculate the loss function in the current model parameter vector ω k The gradient at Update the model parameter vector according to the gradient descent formula get The formula is: in is the model parameter vector, is the updated model parameter vector, η is the learning rate, is the loss function In the parameter vector The gradient at .

7. A robot joint angle calibration method according to claim 1, characterized in that: In the above S400, the predictive thermal magnetic compensation is performed by using a thermal compensation model to predict the temperature change in the short term in the future. Assume that the historical temperature sequence is T1, T2, ..., T m , predict the temperature T at the next Δt time m+Δt The calculation formula is: where α j is the autoregressive term coefficient, β j is the sliding average term coefficient, ∈ m-j is the prediction error at the mjth moment, that is, the difference between the actual temperature and the predicted temperature, δ is the cross-term coefficient, p, q, r are the model orders, ΔT m-j is the temperature change at the mjth moment, and the formula is: ΔT m-j =T m-j -T m-j-1 .

8. A robot joint angle calibration method according to claim 1, characterized in that: In the step S400, the predictive thermal magnetic compensation is performed by using a magnetic compensation model to predict the magnetic field change in the short term in the future. The calculation formula is: where γ j is the magnetic field autoregressive term coefficient, δ j is the coefficient of the magnetic field sliding mean term, ∈′ m-j is the magnetic field prediction error at the mjth moment, and μ is the magnetic field cross-term coefficient.

9. A robot joint angle calibration system, the system being applicable to a robot joint angle calibration method as claimed in claims 1 to 8, characterized in that: The system includes a sensor module, a data acquisition and transmission module, a model building and adjustment module, a predictive compensation module and a calibration process management module; The sensor module: a sensor is installed at each joint of the robot, wherein the temperature sensor is a thermocouple temperature sensor, which is installed on the robot joint housing; the magnetic field sensor is a Hall effect magnetic field sensor, which is installed inside the joint near the angle sensor; the angle sensor is installed at the robot joint, which is responsible for collecting joint temperature, magnetic field strength and angle information respectively; The data acquisition and transmission module: uses a data acquisition card with multiple analog input channels, sets the sampling frequency according to the robot joint movement and environmental changes, converts the sensor analog signal into a digital signal through the sampling program, and transmits it to the robot control system through the USB3.0 interface; The model building and adjustment module: constructs a simulation environment, simulates the working state of the robot joint under different working conditions, collects the joint angle error data under different thermal and magnetic conditions, builds a coupling model of thermal and magnetic coupling and joint angle error based on the error data, and optimizes the parameters; Continuously obtain real-time thermal and magnetic information and compensated joint angle feedback information, use loss function to measure the difference, calculate gradients through back propagation to update model parameters and regularly optimize according to data features; The predictive compensation module collects historical thermal and magnetic data, predicts temperature and magnetic field changes according to the thermal compensation model and the magnetic compensation model, calculates the pre-compensated joint angle error according to the coupling model, and corrects the error in advance; The calibration process management module: after the robot is started, each module is started in sequence and the process is circulated, each part of the system is regularly inspected, problems are maintained and repaired, and parameter algorithms are optimized.

Images