A control method for a robot system

Abstract

The present invention relates to the field of manipulator intelligent control technology, specifically a control method for a robot system. In the present invention, by acquiring and analyzing the torque signals of the manipulator joints and connecting rods, combined with real-time operation data, the influence of each stress source under various operating conditions can be accurately evaluated, forming an in-depth understanding of the manipulator operating environment. Based on the above data decomposition method, the node energy release characteristics can be refined under different states, and the future energy response trend can be predicted, so as to facilitate pre-adjustment measures before stress changes. By weighted calculation of the driving force correction value of the joint, the driving force is adapted to the real-time changes in the stress distribution, thereby improving the precise control level of the driving force.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of manipulators, and in particular to a control method of a robot system. Background Art

[0002] The control technology of a robotic system usually involves precise control of a manipulator (usually a robotic arm or a robot's actuator). By combining sensor technology, artificial intelligence algorithms, and computer control systems to collect and process environmental information in real time, the manipulator is equipped with the ability to operate adaptively and autonomously, thereby making corresponding adjustments. It is widely used in industrial automation, intelligent manufacturing, and other fields.

[0003] Although existing robot control technology has achieved precise control of the manipulator, it is insufficient in the adaptability of stress and driving force, making it difficult for the manipulator to adapt to the rapidly changing stress in a complex environment, resulting in reduced operating accuracy. Existing systems usually set control parameters based on static or average values, which makes it difficult to effectively predict or respond to local stress fluctuations. At the same time, when path deviation occurs, it is difficult for the system to handle different types of deviations in a graded manner. It usually reacts slowly or over-adjusts to smaller deviations, and may not be able to effectively correct larger deviations, resulting in path deviation. This makes it impossible for the manipulator to achieve high-precision and stable execution effects in continuous path following. This limits the operational stability and reliability of the manipulator in actual operations. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a control method for a robot system.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: a control method of a robot system, comprising the following steps:

[0006] S1: Obtain the torque sensor signals of the manipulator joints and connecting rods and the manipulator operation data, evaluate the relative influence of the manipulator on each stress source under different operating conditions, obtain the stress source data, define the connection position of the manipulator joints and the end point position of the connecting rod as different nodes, analyze the energy release characteristics of the stress source data at each node, discretize the energy release characteristics of each node, obtain the energy response information of the node at different times or states, calculate the possible energy change trend of the node in the future based on the energy response information, and generate the local stress distribution prediction information of the node;

[0007] S2: collecting the torque, load and angle data of the manipulator joints to form a comprehensive stress state set of the joints, performing weighted calculation on the data in the comprehensive stress state set of the joints to determine the driving force correction value of the manipulator joints, adjusting the driving force correction value of the manipulator joints in real time according to the local stress distribution prediction information of the nodes, and generating a driving force adjustment result of the manipulator joints;

[0008] S3: During the movement of the manipulator, the driving force adjustment results of the manipulator joints are sampled multiple times and dynamically simulated under a conventional motion path, and the stability score of each test is calculated, and the configuration with the highest stability score of each test is used as the control parameter of the manipulator joint. The control parameters of the manipulator joint are used to adjust the driving force output of each node to generate an optimal configuration set of control parameters;

[0009] S4: Monitor in real time the motion path deviation of the manipulator affected by instantaneous changes in the complex path following of continuous operation. When the deviation data of the manipulator is monitored, refer to the optimal configuration set of the control parameters, calculate the current required deviation correction value according to the deviation amplitude, deviation direction persistence and response time of the deviation data, adjust the control parameters and correct the deviation, and generate the control optimization result of the robot system.

[0010] As a further solution of the present invention, the step of acquiring the energy response information is specifically as follows:

[0011] S111: collect torque sensor signals of the joints and connecting rods of the manipulator and remove noise, perform feature decomposition and spectrum analysis on the torque sensor signals, combine the data of the manipulator under different operating conditions acquired in real time, analyze its influence on stress sources including gravity, inertia, and friction, and generate stress source data;

[0012] S112: Based on the stress source data, the joint connection positions of the manipulator and the end points of the connecting rod are defined as nodes, and the energy release characteristics of the nodes are analyzed to generate energy release analysis results of each node;

[0013] S113: Based on the energy release analysis results of each node, discretize the data of the energy release characteristics of each node, divide the continuous data into fixed time intervals or specific intervals, and generate energy response information.

[0014] As a further solution of the present invention, the step of obtaining the local stress distribution prediction information of the node is specifically as follows:

[0015] S121: Based on the energy response information, use the formula:

[0016]

[0017] Calculate the node energy at the next moment in the energy response information , generate a chart of future energy change trends;

[0018] in, Indicates the current moment in the energy response information The node energy, represents the energy component of the gravity stress source at the node stationary state, represents the energy component of the inertial stress source, represents the energy component of the friction stress source, is the time increment based on the energy response information;

[0019] S122: Based on the data in the future energy change trend chart, analyze the energy increment and stress source contribution at each moment, combine the preset different stress source thresholds, judge the local stress distribution state of the node in the future period, and generate the local stress distribution prediction information of the node.

[0020] As a further solution of the present invention, the steps for obtaining the driving force adjustment result of the manipulator joint are specifically as follows:

[0021] S211: using a torque sensor, a load sensor and an angle encoder, respectively collecting torque, load and angle data of the manipulator joints, combining the local stress distribution prediction information of the nodes to perform synchronous processing item by item, and generating a comprehensive stress state set of the joints;

[0022] Based on the set of comprehensive stress states of the joint, the formula is used:

[0023]

[0024] Calculate the comprehensive stress state of the joint Correction value of item parameter , get the driving force correction value of the manipulator joint;

[0025] in, It is the comprehensive stress state of the joint. The weighting coefficient of the item parameter, It is the comprehensive stress state of the joint. The raw data of the item parameters, It is the comprehensive stress state of the joint. Dynamic adaptation factor of item parameters;

[0026] S213: Obtain the possible energy change trend of the node in the future from the local stress distribution prediction information of the node, adjust the driving force correction value of the manipulator joint according to the predicted energy change trend, and generate the driving force adjustment result of the manipulator joint.

[0027] As a further solution of the present invention, the step of obtaining the stability score of each test is specifically as follows:

[0028] S311: Based on the driving force adjustment result of the manipulator joint, a dynamic simulation test is performed, and the response of each manipulator joint in different paths is recorded one by one to generate a comprehensive path sampling data set;

[0029] S312: Based on the comprehensive path sampling data set, the formula is used:

[0030]

[0031] Calculation based on comprehensive path sampling data set Stability score of the test , get the stability score for each test;

[0032] in, Based on the comprehensive path sampling data set The response delay time of the test, It is a latency benchmark based on a comprehensive path sampling dataset, representing the maximum latency value among all tests. Based on the comprehensive path sampling data set The steady-state error of the test, is the steady-state error benchmark based on a comprehensive path sampling data set, representing the average steady-state error of multiple tests, Based on the comprehensive path sampling data set The oscillation frequency of the test, It is a system oscillation frequency benchmark based on a comprehensive path sampling data set.

[0033] As a further solution of the present invention, the step of obtaining the optimal configuration set of control parameters is specifically:

[0034] S321: Based on the stability score of each test, select the configuration with the highest score as the control parameter of the manipulator joint, compare the score results of different tests and determine the best configuration, and obtain the best configuration for subsequent adjustment;

[0035] S322: Based on the optimal configuration for subsequent adjustment, the driving force output of each node is adjusted using the configuration, the motion characteristics of each manipulator joint are limited by delay, steady-state error and oscillation frequency, and an optimal configuration set of control parameters is generated.

[0036] As a further solution of the present invention, the step of obtaining the deviation correction value currently required for calculation is specifically as follows:

[0037] S411: Real-time monitoring of the position data of the manipulator in complex path following based on continuous operation, comparing the actual position of each time node with the target path, identifying the deviation affected by instantaneous changes, and recording the deviation amplitude, offset direction and change trend to obtain deviation monitoring data;

[0038] S412: Based on the deviation monitoring data, refer to the optimal configuration set of the control parameters and use the formula:

[0039]

[0040] Calculate bias correction value , get the currently required deviation correction value;

[0041] in, is the deviation magnitude in the deviation monitoring data, It is the persistence coefficient set based on the deviation direction in the deviation monitoring data. is the response time in the deviation monitoring data, is the optimal response delay time in the optimal configuration set of control parameters.

[0042] As a further solution of the present invention, the steps for obtaining the control optimization result of the robot system are specifically as follows:

[0043] S421: Based on the currently required deviation correction value, the control parameter is adjusted, the deviation is divided into a plurality of different ranges, a corresponding magnitude correction is performed on each deviation situation, and adjustment results under different deviation scenarios are generated;

[0044] S422: According to the adjustment results under the different deviation scenarios, the correction process of the path deviation is dynamically optimized, and the control parameters are gradually adjusted in combination with the repetitive deviation analysis of the specific path segment to generate a control optimization result of the robot system.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are:

[0046] In the present invention, by acquiring and analyzing the torque signals of the manipulator joints and connecting rods, combined with real-time operation data, the influence of each stress source under various operating conditions can be accurately evaluated, and an in-depth understanding of the manipulator operating environment can be formed. Based on the above data decomposition method, the node energy release characteristics can be refined under different states, and the future energy response trend can be predicted, which is convenient for taking pre-adjustment measures before stress changes. By weighted calculation of the driving force correction value of the joint, the driving force is adapted to the real-time changes of the stress distribution, thereby improving the precise control level of the driving force. Based on the Monte Carlo method and stability scoring, the optimal driving force configuration is selected to optimize the node driving force output and reduce the fluctuation of the control parameters. By real-time monitoring of the path deviation, combined with the changes in the deviation amplitude and response time, the control parameters are adjusted in stages and the deviation correction is performed to achieve high accuracy and stability in continuous path following, thereby avoiding the errors caused by the accumulation of deviations and improving the stability and execution efficiency of the manipulator operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0048] Figure 2 A flow chart for obtaining energy response information of the present invention;

[0049] Figure 3 A flow chart of obtaining local stress distribution prediction information of a node according to the present invention;

[0050] Figure 4 A flow chart of obtaining the driving force adjustment result of the manipulator joint according to the present invention;

[0051] Figure 5 Flow chart for calculating the stability score for each test for the present invention;

[0052] Figure 6 A flow chart for obtaining an optimal configuration set of control parameters for the present invention;

[0053] Figure 7 A flow chart for calculating the currently required deviation correction value for the present invention;

[0054] Figure 8 The present invention is a flow chart for obtaining control optimization results of a robot system. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0056] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0057] See also Figure 1 The present invention provides a technical solution: a control method of a robot system, comprising the following steps:

[0058] S1: Obtain torque sensor signals of manipulator joints and connecting rods, perform feature decomposition and spectrum analysis on torque sensor signals, and combine with real-time manipulator operation data to evaluate the relative influence of manipulator on various stress sources, including gravity, inertia, and friction, under different operating conditions, including static, acceleration, deceleration, and constant speed, to obtain stress source data. Define the connection position of manipulator joints and the end position of connecting rods as different nodes, and analyze the energy release characteristics of stress source data at each node. Energy release characteristics include instantaneous energy release rate, accumulated energy release, energy release rate change, and vibration or fluctuation components of different frequencies. Discretize the energy release characteristics of each node, obtain energy response information of nodes at different times or states, and calculate the possible energy change trend of nodes in the future based on energy response information, which is used to judge the future local stress distribution state of each node and generate local stress distribution prediction information of nodes.

[0059] S2: collect the torque, load and angle data of the manipulator joint through the torque sensor, load sensor and angle encoder to form a comprehensive stress state set of the joint, perform weighted calculation on the torque, load and angle data in the comprehensive stress state set of the joint, determine the driving force correction value of the manipulator joint, and adjust the driving force correction value of the manipulator joint in real time according to the local stress distribution prediction information of the node, so that the driving force correction value of the manipulator joint can adapt to the change of the node stress in real time, and generate the driving force adjustment result of the manipulator joint;

[0060] S3: During the movement of the manipulator, the Monte Carlo method of Markov chain is used to perform multiple sampling and dynamic simulation tests on the driving force adjustment results of the manipulator joints under conventional motion paths. By comparing the response delay, steady-state error and oscillation frequency in different tests and sampling results, and calculating the stability score of each test, the configuration with the highest stability score in each test is used as the control parameter of the manipulator joint. The control parameters of the manipulator joint are used to adjust the driving force output of each node, optimize the overall distribution of the control parameters, and generate the optimal configuration set of the control parameters;

[0061] S4: Real-time monitoring of the motion path deviation of the manipulator affected by instantaneous changes in the complex path following of continuous operation. When the deviation data of the manipulator is monitored, the optimal configuration set of control parameters is referred to, and the current required deviation correction value is calculated according to the deviation amplitude, persistence of the deviation direction and response time of the deviation data. The control parameters are adjusted and the deviation is corrected by hierarchical compensation, including obvious deviation correction, moderate deviation correction and slight deviation correction. When the deviation amplitude exceeds the set range threshold and the deviation direction continuously deviates from the target path, it is judged as obvious deviation. When the deviation amplitude is within the middle range of the range threshold, it is judged as moderate deviation. When the deviation amplitude is close to zero and there is only a slight deviation, it is judged as slight deviation. The motion path is gradually optimized to generate the control optimization result of the robot system.

[0062] The local stress distribution prediction information of the nodes includes the stress distribution area, stress intensity level and stress change trend of each node under different operating conditions. The driving force adjustment results of the manipulator joints include the driving force distribution value, angle correction coefficient and load balancing parameters of each joint. The optimal configuration set of control parameters includes the optimal joint response time, stable angle range and error correction coefficient for different working conditions. The control optimization results of the robot system include the real-time path deviation adjustment value, cumulative deviation correction amount and final equilibrium state of each node.

[0063] See also Figure 2 , the specific steps for obtaining energy response information are:

[0064] S111: collect torque sensor signals of the joints and connecting rods of the manipulator and remove noise, perform feature decomposition and spectrum analysis on the torque sensor signals, combine the data of the manipulator under different operating conditions acquired in real time, analyze its influence on stress sources including gravity, inertia, and friction, and generate stress source data;

[0065] The torque sensor signals of the manipulator joints and connecting rods are obtained. The signal data is filtered and processed to remove noise to ensure data quality. Then, feature decomposition and spectrum analysis are performed. Combined with the real-time operation data of the manipulator, the relative influence of the stress sources of the manipulator under different operating conditions is specifically explained. First, in the static state, gravity is the main stress source, which is manifested as the steady-state low-frequency component of the torque sensor signal. The inertial force can be ignored when static, and the friction force only maintains a certain small value when the static state is maintained to maintain balance. Secondly, in the accelerated state, the influence of the inertial force is significantly enhanced, and the medium and high frequency components in the sensor signal increase to represent the inertial stress generated by the acceleration process of the connecting rod. The influence of gravity remains stable, while the friction force changes accordingly due to the start of joint movement, and the overall friction force increases slightly. Thirdly, in the deceleration state, the inertial force gradually decreases, but in terms of signal characteristics, the reverse stress of the inertial force will appear in the deceleration stage, and the friction force increases to cooperate with the deceleration action. The sensor signal shows the frequency characteristics unique to the deceleration stage, and the influence of gravity is stable and unchanged. Finally, under constant speed, the inertial force tends to be stable, and the friction force is proportional to the joint speed, reflecting the friction resistance characteristics under stable motion, and the influence of gravity remains stable. Under all different states, the various stress source signals based on spectrum analysis are decomposed to obtain the corresponding stress source data for subsequent stress source influence analysis in the subsequent operation condition changes.

[0066] S112: Based on the stress source data, the joint connection positions of the manipulator and the end positions of the connecting rod are defined as nodes, the energy release characteristics of the nodes are analyzed, and the energy release analysis results of each node are generated;

[0067] The joint connection positions and the end positions of the connecting rods of the manipulator are defined as key nodes. For example, the shoulder joint, elbow joint, wrist joint and the end positions of the gripper of the manipulator are respectively used as nodes. Each node position obtains specific stress source data through sensors, and different energy release characteristics are calculated based on the data. First, at the shoulder joint node, the instantaneous energy release rate is calculated using the MATLAB tool through the torque signal monitored by the sensor. Specifically, it is obtained by accumulating the square of the torque change value per unit time, which represents the instantaneous stress release intensity of the node. For example, if the instantaneous energy release rate continues to increase at the shoulder node, it may indicate that the joint is under higher inertial stress. After obtaining the instantaneous energy release rate, the cumulative energy release of the shoulder node is calculated based on the accumulated instantaneous energy release rate to judge the long-term energy release characteristics of the node. For example, if the cumulative energy release remains high for a long time and exceeds the set cumulative energy threshold, it may indicate that the node is at risk of excessive stress and further durability evaluation is required. Next, it is necessary to obtain the change in the instantaneous energy release rate at each moment. The MATLAB tool can be used to calculate the change in the energy release rate of the shoulder node to capture the fluctuation characteristics of the energy release rate of the shoulder joint under different states. For example, when the energy release rate changes violently and exceeds the preset fluctuation intensity threshold, it may reflect that the manipulator is in a frequent start-stop or acceleration and deceleration state, which may lead to increased wear or require higher control accuracy. In addition, through spectrum analysis, the signal at the shoulder node is decomposed into low-frequency and high-frequency components, and the MATLAB tool is used to represent the fluctuation components of different stress sources such as gravity (low frequency) and inertial force (medium and high frequency). For example, if the high-frequency component is significantly enhanced and exceeds the set high-frequency inertial threshold, it may indicate that the influence of inertial force is increased, especially in the acceleration stage; while the low-frequency component increases above the low-frequency gravity threshold, it usually indicates that the influence of gravity is more obvious, usually reflected in the static or constant speed state. Similarly, the instantaneous energy release rate, cumulative energy release, energy release rate change and vibration components of different frequencies are calculated at the elbow and wrist nodes respectively. For example, at the wrist node, if the instantaneous energy release rate of the high-frequency component is detected to increase and exceeds the wrist inertia threshold, it may indicate that the wrist joint is undergoing high-speed and small adjustments, and further optimization is needed to avoid excessive inertial impact; at the elbow node, if the cumulative energy release value exceeds the set elbow joint stress threshold, it may indicate that the load of the elbow joint frequently exceeds the standard range, and it is necessary to check the load of the joint. Through these judgments, the energy release characteristics of the manipulator at different nodes can be comprehensively evaluated, and a dynamic model of stress distribution can also be constructed for subsequent analysis and control.

[0068] S113: Based on the energy release analysis results of each node, discretize the data of the energy release characteristics of each node, divide the continuous data into fixed time intervals or specific intervals, and generate energy response information;

[0069] The energy release characteristics of each node are discretized to obtain the energy response information of the node at different times or states. For example, assuming that on the shoulder node, the energy release characteristics include multiple indicators such as instantaneous energy release rate, cumulative energy release, and energy release rate change. First, the continuously monitored instantaneous energy release rate is sampled at a fixed time interval (such as 0.1 seconds). The value corresponding to each sampling point represents the instantaneous energy release rate at that time point. This time interval is used as the sampling interval threshold, and the discretized data forms a time series. Next, the cumulative energy release is discretized, and the cumulative released energy is divided into different intervals. For example, the cumulative energy within every 10 seconds is called the cumulative energy interval, which is used to count the energy accumulation in different time periods. For the change of energy release rate, the change rate in each time interval is averaged, and the fluctuation characteristics of the energy release rate are represented by discrete change rate data. The defined discrete interval is called the rate change interval. Through the above discretization processing, the energy release data in each time interval is converted into easy-to-process discrete values, such as the instantaneous energy release rate of the shoulder node within 1 second, 2 seconds, ... 10 seconds, or the cumulative energy value within every 10 seconds, which is convenient for analysis and comparison in different time periods. For the manipulator operation data in different states (such as acceleration, constant speed, and deceleration), the discretized energy release characteristics can intuitively reflect the energy response characteristics in different states. For example, assuming that the instantaneous energy release rate of the shoulder node in the acceleration stage is higher than that in the static state every 1 second, it indicates that the inertial force has a more significant effect; while in the constant speed stage, if the value in the rate change interval is stable within a low fluctuation range, it shows that friction dominates in the constant speed state. This discretization processing method effectively converts continuous data into discrete energy response information that can be directly used for state assessment and trend prediction.

[0070] See also Figure 3 , the specific steps for obtaining the local stress distribution prediction information of the node are:

[0071] S121: Based on the energy response information, use the formula:

[0072]

[0073] Calculate the node energy at the next moment in the energy response information , generate a chart of future energy change trends;

[0074] in, Indicates the current moment in the energy response information The node energy is directly derived from the torque sensor signal on the node. Represents the energy component of the gravity stress source when the node is stationary. The gravity component is obtained through the low-frequency component of the spectrum when the node is stationary. For example, when the node is stationary, the torque sensor records the low-frequency signal, and the influence of the gravity stress source is determined by analyzing its amplitude. Represents the energy component of the inertial force stress source. The spectral components of the inertial force are mainly distributed in the high-frequency part. When the node is accelerated or decelerated, the inertial force influence is extracted by the amplitude of the high-frequency component. The energy component representing the friction stress source is extracted by the intermediate frequency component of the torque signal under the constant speed state of the node. It is the time increment based on the energy response information, that is, the time step of the system, which is usually set to a fixed sampling interval to convert the time step into discrete accumulated energy.

[0075] Assume that the torque sensor obtains the current moment in real time The energy reading is 20 joules.

[0076] Determine the stress source component values:

[0077] Gravity stress source :Get the amplitude of the low-frequency component when the node is stationary, and determine it as low frequency (0-10Hz) according to the frequency classification standard. The experiment records the low-frequency amplitude and obtains Joule / second.

[0078] Inertial stress source :The high-frequency component of the node under acceleration represents the inertial force. According to the classification standard, the high-frequency component with a frequency greater than 50Hz is defined as the inertial force. The experimental record high-frequency amplitude is Joule / second.

[0079] Friction stress source : Under constant speed, the friction force is expressed by the amplitude of the mid-frequency band, and 10-50Hz is selected as the mid-frequency band according to the division standard. The experimental results show that the friction force component Joule / second.

[0080] Setting the time step seconds, matching the sampling frequency of the system, and is used to gradually accumulate energy changes.

[0081] Calculate the energy state at the next moment:

[0082]

[0083] Continue to calculate the next time step and get:

[0084]

[0085] The results show that at time Based on the above, by gradually accumulating the components of each stress source, the energy gradually increases from 20 joules to 40 joules. The energy value of the node at the future moment is calculated. On this basis, the energy accumulation results at each moment can be organized into a time series data set. Then, by drawing a time series curve, the energy value at each moment is marked on the chart to form a future energy change trend chart. The specific operation is: set the horizontal axis to the time axis , the vertical axis is set as energy Value, mark the calculated energy value at each time point in turn And connected into a curve. This trend chart will intuitively display the changes in node energy in the future, clearly present the trend of energy accumulation or consumption, and facilitate the analysis of the energy demand and distribution characteristics of nodes in different states.

[0086] S122: Based on the data in the future energy change trend chart, analyze the energy increment and stress source contribution at each moment, combine the preset different stress source thresholds, determine the local stress distribution state of the node in the future period, and generate the local stress distribution prediction information of the node;

[0087] According to the calculation results, the local stress distribution state of each node at different times in the future can be determined. Joules and Taking Joule as an example, if the energy value shows a continuous upward trend, it indicates that the node may experience a higher stress load in future operations. By analyzing these energy increment data, the distribution of different stress sources can be further determined. For example, when the contribution rate of gravity, inertia and friction remains unchanged, as the node accelerates, changes in constant speed and other states, it is observed that the influence of the inertia stress source continues to increase, indicating that the local stress concentration of the node will increase under acceleration or frequent speed changes. In the future, the accumulated energy growth value at each moment can be used as a reference value for local stress. When it exceeds a specific threshold (such as the set safety stress value), it is predicted that the node is at risk of excessive stress.

[0088] See also Figure 4 , the specific steps for obtaining the driving force adjustment result of the manipulator joint are:

[0089] S211: The torque, load and angle data of the manipulator joints are collected through the torque sensor, load sensor and angle encoder respectively, and the local stress distribution prediction information of the node is combined to perform synchronous processing item by item to generate a comprehensive stress state set of the joints;

[0090] The actual torque, load and angle data of the manipulator joints are collected in real time through torque sensors, load sensors and angle encoders, and the real-time data are matched and combined with the local stress distribution prediction information of the nodes one by one. First, the torque sensor records the torque data of each joint, the load sensor collects the instantaneous load value borne by the joint, and the angle encoder obtains the actual angle and angular velocity of the joint. These real-time data are arranged in time series in the joint comprehensive stress state set and correspond to the previous stress distribution prediction value. The actual detected real-time data is compared and synchronized with the predicted stress information to form a comprehensive stress state set of the complete joint containing predicted and actual stress information, laying a data foundation for the subsequent driving force correction and joint control parameter optimization.

[0091] S212: A set of comprehensive stress states based on joints, using the formula:

[0092]

[0093] Calculate the comprehensive stress state of the joint Correction value of item parameter , get the driving force correction value of the manipulator joint;

[0094] in, It is the comprehensive stress state of the joint. The weighting coefficient of each parameter is used to reflect the degree of influence of each parameter on the correction value, indicating the importance or weight of torque, load and angle. For the weighting coefficient of torque, a series of different initial torque values ​​can be set and gradually increased, and the stability and response effect of the manipulator under different torque values ​​can be recorded. When the torque increases to a certain level, if the manipulator shakes or the stability decreases, the size of the torque weighting coefficient is appropriately adjusted until the manipulator can work smoothly under different torque levels. For the weighting coefficient of load, the stability of the manipulator under different load conditions can be tested by gradually increasing the joint load. For example, the load can be gradually increased, and the change of the driving force output with the load can be recorded. The weighting coefficient of the load is appropriately adjusted so that the manipulator can still maintain a stable response under a large load. For the weighting coefficient of angle, the angle and angular velocity of the joint can be gradually adjusted, and the posture and balance state of the manipulator can be recorded. When the angle changes greatly or the acceleration increases, the weighting coefficient of the angle is adjusted to ensure that the manipulator remains balanced in dynamic motion, so that the influence of the angle change on the driving force remains within a controllable range. It is the comprehensive stress state of the joint. The raw data of the item parameters, indicating the specific torque, load and angle values, It is the comprehensive stress state of the joint. The dynamic adaptation factor of the item parameter is used to adjust the influence of the stress source in different states, for example: , indicating that no adjustment is required; when the robot is in acceleration state, Indicates the increase of adaptability to inertia force; when the robot is in deceleration state, Indicates a reduced need for force.

[0095] Assume that the current torque of the manipulator is , the load is , angle data is , set the moment weight , load weight , Angle Weight , set in acceleration state To adapt to the inertia force, load adaptation factor at rest , Angle adaptation factor under deceleration state .

[0096] Calculate torque correction value :

[0097]

[0098] Calculate load correction value :

[0099]

[0100] Calculate angle correction :

[0101]

[0102] The results show that the calculated correction value is , , .

[0103] S213: acquiring the possible energy change trend of the node in the future from the local stress distribution prediction information of the node, adjusting the driving force correction value of the manipulator joint according to the predicted energy change trend, and generating the driving force adjustment result of the manipulator joint;

[0104] When the predicted energy level of the node at the next moment shows an increasing trend, it means that the node may face higher stress requirements in the future. To adapt to this change, the dynamic adaptation factor needs to be adjusted accordingly to optimize the driving force correction value of the joint. Specifically, if the predicted energy level increases, it may be due to the increase in inertia caused by acceleration or the increase in load. At this time, the value of the dynamic adaptation factor can be appropriately increased so that the torque correction value, load correction value and angle correction value can meet the new stress requirements. For example, during acceleration, by increasing the dynamic adaptation factor, the torque correction value can be increased, thereby providing a greater driving force to offset the inertia force and ensure the stable operation of the joint. Similarly, the load correction value will also be adjusted accordingly according to the predicted load increment to help the joint adapt to the possible increase in load, thereby avoiding the risk of overload. The angle correction value is adapted according to the angular velocity and angle change of the joint to ensure that the posture of the manipulator maintains stability and accuracy under high stress. On the contrary, when the predicted energy level shows a tendency to stabilize or decrease, the value of the dynamic adaptation factor can be appropriately reduced, so as to adjust the driving force correction value back to a stable state and avoid unnecessary energy consumption. For example, at constant speed or static state, the inertia force and friction force reach a balance. At this time, the dynamic adaptation factor is reduced so that the torque, load and angle correction values ​​return to the basic level, providing only the necessary driving force to maintain the stability and position accuracy of the joint.

[0105] See also Figure 5 , the steps to calculate the stability score for each test are as follows:

[0106] S311: Based on the driving force adjustment results of the manipulator joints, a dynamic simulation test is performed, and the response of each manipulator joint in different paths is recorded one by one to generate a comprehensive path sampling data set;

[0107] During the motion of the manipulator, the Monte Carlo method of Markov chain is used to perform multiple sampling and dynamic simulation tests of the driving force of the manipulator joints to optimize the driving force control configuration. First, a state transition model is constructed for the motion states of the manipulator joints (such as static, acceleration, deceleration and constant speed). By analyzing a large amount of historical data, the transition probability of each state is estimated, for example, the probability of remaining motionless in the static state is 70%, the probability of transferring to the acceleration state is 20%, and the probability of transferring to the deceleration and constant speed states is 5% respectively. These probabilities are defined in MATLAB through the state transition matrix, reflecting the different state transition tendencies of the manipulator under actual task conditions. Subsequently, the Monte Carlo method is used for sampling. First, multiple motion paths based on state transition probabilities are randomly generated. Each path represents the possible running trajectory of the manipulator under a specific driving force configuration. During each sampling process, the motion performance of the manipulator in the real control scenario is simulated according to the path through Simulink in MATLAB. In each path, the driving force change data, response delay time and steady-state error of the joints in different states are collected in real time. These data are stored in the test result set of multiple paths to form a simulation database. The key data for recording the oscillation frequency characteristics of the manipulator under different driving force configurations include: Average value of oscillation frequency: the main vibration frequency under each driving force configuration, expressed in Hertz (Hz). Oscillation amplitude: the vibration amplitude at the corresponding frequency, which is used to reflect the oscillation intensity. Steady-state reaching time: the time required to reach a stable oscillation frequency. Oscillation period: the repetition period of periodic vibration under the same driving force. Oscillation spectrum characteristic peak: the peak position and amplitude of the main frequency and sub-frequency in the spectrum analysis. Oscillation decay rate: the rate of change of oscillation amplitude with time or path distance. Instantaneous frequency in the start or stop phase: the change of vibration frequency at the beginning and end of the movement. The oscillation frequency characteristics of the manipulator under different driving force configurations are further recorded by dynamic simulation of each path, and spectrum analysis is performed on all states of the path to identify the main vibration components in each state. The simulation results of all sampled paths are analyzed comprehensively, and MATLAB calculates these results to evaluate the stability score of each test.

[0108] S312: Based on the comprehensive path sampling data set, the formula is used:

[0109]

[0110] Calculation based on comprehensive path sampling data set Stability score of the test , get the stability score for each test;

[0111] in, Based on the comprehensive path sampling data set The response delay time of each test is in seconds. The time difference from the driving force correction signal to the robot response is obtained through the delay sensor. For example, in a test, the response delay time is recorded as 1.3 seconds. Second, It is a system delay benchmark based on a comprehensive path sampling data set, indicating the maximum delay value in all tests. For example, assuming the maximum delay of the test is 2.0 seconds, Second, is the steady-state error of the jth test based on the comprehensive path sampling data set, and the unit is path deviation. The deviation between the actual path of the manipulator and the ideal path is measured by the sensor. For example, in one test, the recorded path deviation is 0.04, then , It is the steady-state error benchmark based on the comprehensive path sampling data set, which represents the average steady-state error of multiple tests. For example, assuming that the mean error of 10 tests is 0.046, then , Based on the comprehensive path sampling data set The oscillation frequency of the test, in Hertz, is obtained through a spectrum analyzer. For example, if the oscillation frequency recorded in a test is 1.6 Hz, then , is the system oscillation frequency benchmark based on the comprehensive path sampling data set, indicating the maximum oscillation frequency. For example, assuming that the maximum frequency in multiple tests is 2.0 Hz, then .

[0112] Assume that the following is recorded in three tests: Test 1: Second, , , test 2: Second, , , test 3: Second, , , the benchmark value is Second, , .

[0113] For Test 1: Latency Ratio: , steady-state error ratio: , oscillation frequency ratio: , stability score : .

[0114] For Test 2: Latency Ratio: , steady-state error ratio: , oscillation frequency ratio: , stability score : .

[0115] For Test 3: Latency Ratio: , steady-state error ratio: , oscillation frequency ratio: , stability score : .

[0116] See also Figure 6 , the steps for obtaining the optimal configuration set of control parameters are as follows:

[0117] S321: Based on the stability score of each test, select the configuration with the highest score as the control parameter of the manipulator joint, compare the score results of different tests and determine the best configuration, and obtain the best configuration for subsequent adjustment;

[0118] By analyzing the single stability scores of test 1, test 2, and test 3, the score results of each test configuration are obtained. Stability score of test 1 , stability score for test 2 , stability score for test 3 Among the three tests, the highest score was in Test 2, which was 2.89. Based on the score results, the configuration of Test 2 can be selected as the best driving force configuration under the current test conditions and used for setting the control parameters of the manipulator joints to ensure that the manipulator achieves optimal stability under a specific motion path.

[0119] S322: Based on the optimal configuration for subsequent adjustment, the driving force output of each node is adjusted using the configuration, the motion characteristics of each manipulator joint are limited by delay, steady-state error and oscillation frequency, and an optimal configuration set of control parameters is generated;

[0120] In each test, the configuration with the highest stability score is used as the control parameter of the manipulator joint. Assuming that the score of test 2 is the highest, its scoring configuration is used as the standard. Through the characteristic parameters it contains (such as response delay time, steady-state error and oscillation frequency), this optimal configuration is used to adjust the driving force output of each node of the manipulator system. In the specific adjustment process, the response delay time in the optimal configuration is allocated to the driving force adjustment of each joint node, so that each node can respond to the input signal within the configured response delay time and feedback correction in real time. The steady-state error is used to limit the motion path deviation of each node to ensure that the joint remains within the error allowable range in each action. In addition, by using the oscillation frequency characteristics of the optimal configuration, by analyzing the vibration characteristics of each node under driving force loading, the vibration frequency of the joint is adjusted one by one to ensure stable operation of the system. Finally, all parameters in the optimal configuration are synchronized to the driving force control of each joint node, and the optimal configuration set of control parameters is formed through the optimized delay, error and frequency adjustment combination.

[0121] See also Figure 7 , the specific steps for calculating the currently required deviation correction value are:

[0122] S411: Real-time monitoring of the position data of the manipulator in complex path following based on continuous operation, comparing the actual position of each time node with the target path, identifying the deviation affected by instantaneous changes, and recording the deviation amplitude, offset direction and change trend to obtain deviation monitoring data;

[0123] In the complex path following of continuous operation, the manipulator is monitored in real time to identify and respond to the deviation of the motion path affected by instantaneous changes. During the operation, the sensor continuously collects the position data of the manipulator and compares the actual position of each time node with the target path to determine whether there is a deviation. Once the current position of the manipulator is detected to deviate from the target path, the system will record the deviation amplitude, direction and the instantaneous change factors generated, such as mechanical shock, environmental vibration or other external interference to the manipulator. In the process of detecting the deviation, the system not only records the instantaneous position of the deviation, but also monitors the rate of change of the deviation, that is, whether the deviation has a tendency to expand or shrink in a short period of time. In addition, the system will analyze the directional stability of the deviation in real time, observe whether the deviation direction remains consistent or fluctuates frequently, so as to distinguish the deviation caused by continuous factors and instantaneous changes. Through these real-time monitoring and data analysis, the system can timely obtain the impact of instantaneous changes on the manipulator path, thereby providing accurate reference data for subsequent correction and control. Ultimately, these monitoring data will be used to calculate the necessary deviation correction values ​​to ensure that the manipulator can quickly return to the target path in complex path following.

[0124] S412: Based on the deviation monitoring data, refer to the optimal configuration set of control parameters and use the formula:

[0125]

[0126] Calculate bias correction value , get the currently required deviation correction value;

[0127] in, is the deviation amplitude in the deviation monitoring data, in millimeters (mm), indicating the distance difference between the current position of the robot and the target path. It is collected in real time by the position sensor and reflects the basic magnitude of the deviation. The acquisition method is: at each time node, the position sensor detects the actual position of the robot and compares it with the expected position of the target path to obtain the deviation amplitude. For example, if the sensor detects that the robot deviates from the target path by 8 mm, then mm, It is a persistence coefficient set based on the offset direction in the deviation monitoring data. It indicates the consistency of the offset direction. The value is between 0 and 1. This coefficient is used to reflect whether the offset direction remains stable. If the offset direction remains unchanged, Close to 1; if the direction changes frequently, Close to 0, for example, within a fixed time window, the proportion of statistical offset directions that remain consistent. For example, in 5 consecutive samples, if 4 offset directions are consistent, then , It is the response time in the deviation monitoring data, in seconds (s), which indicates the time interval from when the system detects the deviation to when it starts to correct it, reflecting the reaction speed of the system. The acquisition method is: the system records the start time of deviation detection and the start time of correction, and obtains the response time through the timestamp difference. For example, assuming that the deviation detection time is 2.5 seconds and the correction start time is 3.2 seconds, then Second, is the optimal response delay time in the optimal configuration set of control parameters, in seconds (s), It is an adjustment factor set based on the deviation monitoring data, which indicates the ratio difference between the current response time and the optimal response time. It is used to adjust the correction amount. Its purpose is to increase the correction strength when the response time is fast and to reduce the correction strength when the response time is slow to avoid over-adjustment. For example, if the response time is Seconds, optimal response time seconds, the adjustment factor is calculated as: .

[0128] Assume that the magnitude of the detected deviation is mm, persistence factor in the offset direction , response time s, optimal response time s.

[0129] Substituting into the formula:

[0130]

[0131] Calculate the adjustment factor :

[0132]

[0133] Calculate the correction value :

[0134]

[0135] The results show that the bias-corrected value mm: This value indicates that under the current instantaneous deviation, the system needs to adjust the robot position by 1.6 mm to compensate for the deviation and return it to the target path.

[0136] See also Figure 8 , the specific steps for obtaining the control optimization results of the robot system are:

[0137] S421: Based on the currently required deviation correction value, the control parameters are adjusted, the deviation is divided into a plurality of different ranges, a corresponding magnitude correction is performed on each deviation situation, and adjustment results under different deviation scenarios are generated;

[0138] According to the calculated deviation correction value, the control parameters are adjusted and the deviation is corrected by hierarchical compensation. Assuming that the deviation amplitude threshold set by the system is 1.0 mm, the correction strategy is based on three ranges: when the deviation amplitude exceeds 1.0 mm and the offset direction is continuously consistent, the system judges it as a significant deviation; when the deviation amplitude is between 0.5 and 1.0 mm, it is judged as a moderate deviation; when the deviation amplitude is less than 0.5 mm, it is judged as a slight deviation. For example, in a deviation detection, the system calculates that the deviation correction value is 1.6 mm, and the offset direction remains consistent in multiple sampling cycles, exceeding the threshold of 1.0 mm, so it is judged as a significant deviation. The system will perform a large correction of 1.6 mm, and the manipulator will quickly return to the target path by increasing the driving force. In another operation, if the correction value is 0.8 mm, and the offset direction occasionally changes but is generally stable, the system judges it as a moderate deviation, performs a medium amplitude correction of 0.8 mm, and moderately adjusts the driving force parameters to gradually reduce the impact of the deviation. In the case of a small deviation, for example, the correction value is 0.2 mm and the deviation direction is unstable, the system determines it as a small deviation and only makes a small adjustment of 0.2 mm to maintain the stability of the robot movement and the accuracy of path following. Through this hierarchical compensation method, the system can dynamically adjust the correction strategy according to the degree of deviation to achieve efficient path correction.

[0139] S422: dynamically optimizing the correction process of the path deviation according to the adjustment results under different deviation scenarios, and gradually adjusting the control parameters in combination with the repetitive deviation analysis of a specific path segment to generate a control optimization result of the robot system;

[0140] In the process of gradually optimizing the motion path, the system makes multiple fine-tuning adjustments to the deviation correction scheme based on continuous sampling and real-time feedback data. First, the system accumulates the data results of each correction, and compares the path deviation amplitude and direction before and after the correction to determine the effectiveness of the current correction strategy. Then, according to the deviation accumulation trend and the changes in real-time data, the control parameters are dynamically adjusted to adapt to the complexity and instantaneous changes of the path. In addition, the system will analyze the data of each correction to identify repetitive deviations on specific path segments so that early adjustments can be made under similar operating conditions. For example, if deviations occur frequently at the same complex curve, the system will preset fine-tuning values ​​at that location to further optimize the control strategy. Through this process of gradually optimizing the path, the system eventually forms a set of precise control parameters suitable for complex paths, realizes global optimization of the robot's motion path, and generates control optimization results for the robot system.

[0141] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A control method for a robot system, characterized in that: The following steps are involved: S1: Obtain the torque sensor signals of the manipulator joints and connecting rods and the manipulator operation data, evaluate the relative influence of the manipulator on each stress source under different operating conditions, obtain the stress source data, define the connection position of the manipulator joints and the end point position of the connecting rod as different nodes, analyze the energy release characteristics of the stress source data at each node, discretize the energy release characteristics of each node, obtain the energy response information of the node at different times or states, calculate the possible energy change trend of the node in the future based on the energy response information, and generate the local stress distribution prediction information of the node; The steps for obtaining the energy response information are specifically as follows: S111: collect torque sensor signals of the joints and connecting rods of the manipulator and remove noise, perform feature decomposition and spectrum analysis on the torque sensor signals, combine the data of the manipulator under different operating conditions acquired in real time, analyze its influence on stress sources including gravity, inertia, and friction, and generate stress source data; S112: Based on the stress source data, the joint connection positions of the manipulator and the end points of the connecting rod are defined as nodes, and the energy release characteristics of the nodes are analyzed to generate energy release analysis results of each node; S113: Based on the energy release analysis results of each node, discretize the data of energy release characteristics of each node, divide the continuous data into fixed time intervals or specific intervals, and generate energy response information; S2: collecting the torque, load and angle data of the manipulator joints to form a comprehensive stress state set of the joints, performing weighted calculation on the data in the comprehensive stress state set of the joints to determine the driving force correction value of the manipulator joints, adjusting the driving force correction value of the manipulator joints in real time according to the local stress distribution prediction information of the nodes, and generating a driving force adjustment result of the manipulator joints; The steps for obtaining the driving force adjustment result of the manipulator joint are specifically as follows: S211: using a torque sensor, a load sensor and an angle encoder, respectively collecting torque, load and angle data of the manipulator joints, combining the local stress distribution prediction information of the nodes to perform synchronous processing item by item, and generating a comprehensive stress state set of the joints; Based on the set of comprehensive stress states of the joint, the formula is used: Calculate the comprehensive stress state of the joint Correction value of item parameter , get the driving force correction value of the manipulator joint; in, It is the comprehensive stress state of the joint. The weighting coefficient of the item parameter, It is the comprehensive stress state of the joint. The raw data of the item parameters, It is the comprehensive stress state of the joint. Dynamic adaptation factor of item parameters; S213: acquiring the possible energy change trend of the node in the future from the local stress distribution prediction information of the node, adjusting the driving force correction value of the manipulator joint according to the predicted energy change trend, and generating a driving force adjustment result of the manipulator joint; S3: During the movement of the manipulator, the driving force adjustment results of the manipulator joints are sampled multiple times and dynamically simulated under a conventional motion path, and the stability score of each test is calculated, and the configuration with the highest stability score of each test is used as the control parameter of the manipulator joint. The control parameters of the manipulator joint are used to adjust the driving force output of each node to generate an optimal configuration set of control parameters; The steps for calculating the stability score of each test are specifically as follows: S311: Based on the driving force adjustment result of the manipulator joint, a dynamic simulation test is performed, and the response of each manipulator joint in different paths is recorded one by one to generate a comprehensive path sampling data set; S312: Based on the comprehensive path sampling data set, the formula is used: Calculation based on comprehensive path sampling data set Stability score of the test , get the stability score for each test; in, Based on the comprehensive path sampling data set The response delay time of the test, It is a latency benchmark based on a comprehensive path sampling dataset, representing the maximum latency value among all tests. Based on the comprehensive path sampling data set The steady-state error of the test, is the steady-state error benchmark based on a comprehensive path sampling data set, representing the average steady-state error of multiple tests, Based on the comprehensive path sampling data set The oscillation frequency of the test, is a system oscillation frequency benchmark based on a comprehensive path sampling data set; S4: Real-time monitoring of the motion path deviation of the manipulator affected by instantaneous changes in the complex path following of continuous operation. When the deviation data of the manipulator is monitored, the optimal configuration set of the control parameters is referred to, and the deviation correction value required at present is calculated according to the deviation amplitude, the persistence of the deviation direction and the response time of the deviation data, and the control parameters are adjusted and the deviation is corrected to generate the control optimization result of the robot system. The step of obtaining the deviation correction value currently required for calculation is specifically as follows: S411: Real-time monitoring of the position data of the manipulator in complex path following based on continuous operation, comparing the actual position of each time node with the target path, identifying the deviation affected by instantaneous changes, and recording the deviation amplitude, offset direction and change trend to obtain deviation monitoring data; S412: Based on the deviation monitoring data, refer to the optimal configuration set of the control parameters and use the formula: Calculate bias correction value , get the currently required deviation correction value; in, is the deviation magnitude in the deviation monitoring data, It is the persistence coefficient set based on the deviation direction in the deviation monitoring data. is the response time in the deviation monitoring data, is the optimal response delay time in the optimal configuration set of control parameters.

2. The control method of the robot system according to claim 1, characterized in that: The steps for obtaining the local stress distribution prediction information of the node are specifically as follows: S121: Based on the energy response information, use the formula: Calculate the node energy at the next moment in the energy response information , generate a chart of future energy change trends; in, Indicates the current moment in the energy response information The node energy, represents the energy component of the gravity stress source at the node stationary state, represents the energy component of the inertial stress source, represents the energy component of the friction stress source, is the time increment based on the energy response information; S122: Based on the data in the future energy change trend chart, analyze the energy increment and stress source contribution at each moment, combine the preset different stress source thresholds, judge the local stress distribution state of the node in the future period, and generate the local stress distribution prediction information of the node.

3. The control method of the robot system according to claim 1, characterized in that: The steps for obtaining the optimal configuration set of the control parameters are specifically as follows: S321: Based on the stability score of each test, select the configuration with the highest score as the control parameter of the manipulator joint, compare the score results of different tests and determine the best configuration, and obtain the best configuration for subsequent adjustment; S322: Based on the optimal configuration for subsequent adjustment, the driving force output of each node is adjusted using the configuration, the motion characteristics of each manipulator joint are limited by delay, steady-state error and oscillation frequency, and an optimal configuration set of control parameters is generated.

4. The control method of the robot system according to claim 1, characterized in that: The steps for obtaining the control optimization result of the robot system are specifically as follows: S421: Based on the currently required deviation correction value, the control parameter is adjusted, the deviation is divided into a plurality of different ranges, a corresponding magnitude correction is performed on each deviation situation, and adjustment results under different deviation scenarios are generated; S422: According to the adjustment results under the different deviation scenarios, the correction process of the path deviation is dynamically optimized, and the control parameters are gradually adjusted in combination with the repetitive deviation analysis of the specific path segment to generate a control optimization result of the robot system.

Images