A force control robot control method and system based on fuzzy control

Abstract

The present invention relates to a force control robot control method and system based on fuzzy control, which collects the shape data of the controlled object in real time, creates a data processing queue, performs fuzzy processing on the shape data of the controlled object, and outputs a control signal through a fuzzy processing module; a signal prediction module predicts the control signal within the expected time period and creates a control signal queue; the predicted control signal is verified to determine whether the predicted control signal is abnormal. The control signal is sequentially transmitted to the signal conversion module for conversion from digital signal to analog signal, and then transmitted to the control module, and the actuator performs corresponding control; the real-time force data is fed back to the control module through the feedback module, and if the real-time force data is not within the error range, compensation adjustment is performed through the control signal based on the angle and the difference. The execution delay of the force control robot is reduced, and the execution efficiency of the force control robot is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of robot control, and in particular relates to a force control robot control method and system based on fuzzy control. Background Art

[0002] A force-controlled robot is a robot that obtains various data through various sensors and transmits them to the control system. The control system analyzes and processes the data obtained by the sensors, generates control signals and transmits them to the actuator to perform the corresponding actions. The relevant data at a certain execution time is obtained from each sensor, and after data transmission, data analysis and processing are performed, and finally a control signal at a certain time is generated to control the actuator. For the next execution time, various data are obtained through various sensors and transmitted to the control system. After the control system analyzes and processes the obtained data, it generates a control signal at that time to control the actuator. In this way, the force-controlled robot achieves continuous action. Therefore, on the one hand, it will take a certain amount of time to complete the execution of the relevant actions from data acquisition to the generation of control signals. That is to say, after a certain period of time, the actuator will perform the corresponding action for the signal collected by the sensor in real time, which will cause a delay in the execution time. On the other hand, the corresponding data analysis and processing must be performed at each moment to perform the corresponding action, which makes the execution efficiency of the force-controlled robot low.

[0003] Therefore, how to improve the existing control method of the force-controlled robot, reduce the execution delay of the force-controlled robot, and improve the execution efficiency of the force-controlled robot is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The purpose of the present invention is to provide a force-controlled robot control method and system based on fuzzy control, so as to improve the existing force-controlled robot control method, reduce the execution delay of the force-controlled robot, and improve the execution efficiency of the force-controlled robot.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0006] In a first aspect, a force control robot control method based on fuzzy control is provided, comprising the following steps:

[0007] S1: The shape data of the controlled object is collected in real time through the data collection module, and the shape data includes external shape, size and material, and a data processing queue is created based on the collected shape data of the controlled object, and the data is sequentially transmitted to the fuzzy processing module through the data processing queue;

[0008] S2: the fuzzy processing module performs fuzzy processing on the shape data of the controlled object, and outputs a control signal through the fuzzy processing module; the signal prediction module predicts the control signal within the expected time period based on the control signal within the specified time period, and creates a control signal queue;

[0009] S3: verifying the predicted control signal through the shape data of the controlled object collected in real time by the data acquisition module, and judging whether the predicted control signal is abnormal. If so, extracting the corresponding shape data of the controlled object and transmitting it to the fuzzy processing to output the control signal after fuzzy processing, and replacing the corresponding control signal in the control signal queue based on the output control signal after fuzzy processing. If not, executing step S4;

[0010] S4: the control signal queue sequentially transmits the control signal to the signal conversion module for converting the digital signal into an analog signal, and transmits the converted analog signal to the control module, and the control module controls the actuator of the force control robot accordingly based on the analog signal;

[0011] S5: collecting real-time force data of the execution action of the actuator, and feeding back the real-time force data to the control module through the feedback module, the control module compares the acquired real-time force data with the standard force data corresponding to the control signal, and determines whether the real-time force data is within the error range. If so, the control module continues to control the actuator of the force-controlled robot through subsequent control signals. If not, executing step S6;

[0012] S6: Compare the real-time force data with the standard force data corresponding to the control signal, including the direction and magnitude of the force, calculate the angle between the direction of the force of the real-time force data and the direction of the force of the standard force data and the difference between the magnitude of the force of the real-time force data and the magnitude of the force of the standard force data, and perform compensation adjustment through the control signal based on the angle and the difference, so that the angle and magnitude of the real-time force data and the standard force data corresponding to the control signal are within the error range.

[0013] Preferably, there are several controlled objects in step S1, and the data processing queue is created sequentially based on the several controlled objects. The order of data processing in the data processing queue is the processing order of the several controlled objects.

[0014] Preferably, the fuzzy processing module in step S2 performs fuzzy processing on the shape data of the controlled object, and the specific process of outputting the control signal through the fuzzy processing module is as follows:

[0015] S21: creating a fuzzy processing database and a fuzzy processing rule base for the shape data of the controlled object, wherein the fuzzy processing database stores the membership vector values ​​of all fuzzy subsets of the input shape data of the controlled object and the output control variables;

[0016] S22: converting the shape data of the controlled object in the data processing queue into a corresponding fuzzy vector through a fuzzification interface, and solving a fuzzy relation equation for the fuzzy vector based on a fuzzy processing database and a fuzzy processing rule base;

[0017] S23: Obtaining a fuzzy control signal based on the solution result of the fuzzy relationship equation.

[0018] Preferably, the fuzzy processing module in step S2 performs fuzzy processing on the shape data of the controlled object using a two-dimensional fuzzy controller, and two input variables of the two-dimensional fuzzy controller are the deviation and the deviation change of the size of the controlled object and the input standard value.

[0019] Preferably, the specific process of the signal prediction module in step S2 predicting the control signal within the expected time period based on the control signal within the specified time period is as follows:

[0020] The control signal within a specified time period is obtained, and part of the control signal within the specified time period is input into the signal prediction module to train the model, and the parameters of the signal prediction module are iteratively updated through a gradient descent optimization algorithm, and another part of the control signal within the specified time period is input into the signal prediction module, and the control signal is output through the signal prediction module to obtain a prediction result, and it is determined whether the prediction result is within the expected range, if so, the control signal within the expected time period is output, if not, the parameters of the signal prediction module are re-optimized and updated through a particle swarm algorithm.

[0021] Preferably, the specific process of iteratively updating the parameters of the signal prediction module by the gradient descent optimization algorithm is as follows:

[0022] Initialize the weights and biases of the signal prediction module, that is, set a set of weight values ​​and bias values ​​of the signal prediction module, input part of the control signal within the specified time period into the signal prediction module for forward propagation to obtain a predicted value of the control signal, and calculate the error between the predicted value and the standard value through a loss function, calculate the gradient based on the calculated error through back propagation, and the gradient is the influence of the change in weight and bias on the loss function value, and use the gradient to update the weights and biases of the signal prediction module, and the updated weight values ​​and bias values ​​are obtained by subtracting the product of the learning rate and the gradient from the current weight and current bias of the signal prediction module, respectively, and the learning rate is a set parameter, and finally iterate the above process until a preset number of iterations is met or the loss function value is less than a preset threshold.

[0023] Preferably, in step S3, the specific process of verifying the predicted control signal by using the controlled object shape data collected in real time by the data acquisition module to determine whether the predicted control signal has an abnormality is as follows:

[0024] S31: establishing a mapping relationship table between shape data and control signals based on historical controlled object shape data and historical control signals;

[0025] S32: Retrieve the mapping relationship table based on the current shape data, extract the corresponding control signal in the mapping relationship table, compare the corresponding control signal with the predicted control signal, and determine whether the extracted corresponding control signal is within a preset confidence interval. If so, it indicates that the predicted control signal is normal. If not, it indicates that the predicted control signal is abnormal.

[0026] Preferably, the specific process of performing compensation adjustment based on the angle and the difference through the control signal in step S6 is as follows:

[0027] S61: based on the angle between the force direction of the real-time force data and the force direction of the standard force data, calculating a first ratio of the angle to the angle between the force direction of the standard force data and a certain coordinate direction;

[0028] S62: performing reverse compensation on the force direction signal in the control signal based on the first ratio, wherein the reverse compensation means that the force direction of the real-time force data is compensated in the opposite direction to the force direction of the standard force data;

[0029] S63: based on the difference between the magnitude of the force of the real-time force data and the magnitude of the force of the standard force data, calculating a second ratio of the difference to the magnitude of the force of the standard force data;

[0030] S64: Perform reverse compensation on the force signal in the control signal based on the second ratio, wherein the reverse compensation means that when the force of the real-time force data is greater than the force of the standard force data, the force signal in the control signal is adjusted down by a corresponding ratio; and when the force of the real-time force data is less than the force of the standard force data, the force signal in the control signal is adjusted up by a corresponding ratio.

[0031] In a second aspect, a force-controlled robot control system based on fuzzy control is provided, which is used to implement any one of the force-controlled robot control methods based on fuzzy control, and is characterized in that it includes a data acquisition module, a data processing queue, a fuzzy processing module, a control signal queue, a signal prediction module, a signal verification module, a signal conversion module, a control module, an actuator and a feedback module, wherein the data acquisition module is connected to the data processing queue, the data processing queue is connected to the fuzzy processing module, the fuzzy processing module is respectively connected to the control signal queue and the signal prediction module, the signal prediction module is connected to the signal verification module, the control signal queue is connected to the signal conversion module, the signal conversion module is connected to the control module, the control module is connected to the actuator, and the actuator is connected to the feedback module;

[0032] The data acquisition module is used to collect the shape data of the controlled object in real time, and the shape data includes external shape, size and material; the fuzzy processing module is used to perform fuzzy processing on the shape data of the controlled object, and output a control signal through the fuzzy processing module; the control module is used to control the action of the actuator based on the control signal, and perform compensation adjustment through the control signal based on the angle and difference, so that the angle and size of the real-time force data and the standard force data corresponding to the control signal are within the error range; the actuator is used to perform corresponding actions based on the control signal of the control module; the feedback module is used to feed back the real-time action performed by the actuator to the control module.

[0033] The beneficial effects of the present invention include:

[0034] The force control robot control method and system based on fuzzy control provided by the present invention collects the shape data of the controlled object in real time, creates a data processing queue, performs fuzzy processing on the shape data of the controlled object, and outputs a control signal through a fuzzy processing module; the signal prediction module predicts the control signal within the expected time period and creates a control signal queue; the predicted control signal is verified to determine whether the predicted control signal is abnormal. The control signal is sequentially transmitted to the signal conversion module for conversion from digital signal to analog signal, and then transmitted to the control module, and the actuator performs corresponding control; the real-time force data is fed back to the control module through the feedback module, and the acquired real-time force data is compared with the standard force data corresponding to the control signal. If the real-time force data is not within the error range, the real-time force data is compared with the corresponding standard force data, and compensation adjustment is performed through the control signal based on the angle and the difference.

[0035] First, the shape data of the controlled object is fuzzy processed by the fuzzy processing module, and the control signal is output by the fuzzy processing module. There is no need to establish an accurate mathematical model between the shape data of the controlled object and the control signal in the force control process of the force control robot. On the one hand, the control logic of the force control robot is simplified, and the control system is simple in structure, easy to implement and maintain, and the cost is lower. On the other hand, even if the shape data of the controlled object changes or the external environment interferes, the robot's control system has high stability.

[0036] Secondly, the signal prediction module predicts the control signal within the expected time period based on the control signal within the specified time period. The fuzzy processing module does not need to perform fuzzy processing on the shape data of the controlled object at each moment. After the force control robot has worked for a specified time, it can be predicted based on the control signal within the specified time. The fuzzy processing process is a time-consuming stage in the process of the force control robot. Therefore, by predicting the control signal, the control system time of the force control robot will be greatly reduced, the execution delay of the force control robot will be reduced, and the execution efficiency of the force control robot will be improved.

[0037] Thirdly, the predicted control signal is verified by the real-time collected shape data of the controlled object to determine whether there is any abnormality in the predicted control signal, thereby further ensuring the accuracy of the predicted control signal, and processing the inaccurate, i.e., abnormal, control signals to ensure the accuracy of the final control signal received by the control module.

[0038] Finally, the real-time force data is fed back to the control module through the feedback module. The control module compares the acquired real-time force data with the standard force data corresponding to the control signal. When the real-time force data is not within the error range, the control module performs signal compensation processing, thereby further improving the action accuracy of the actuator. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of the force control robot control method based on fuzzy control of the present invention.

[0040] Figure 2 It is a schematic diagram of the architecture of the force control robot control system based on fuzzy control of the present invention.

[0041] Figure 3 It is a schematic diagram of the flow of the present invention for performing compensation adjustment through control signals based on the angle and the difference. DETAILED DESCRIPTION

[0042] The following is combined with Figure 1~Figure 3 The present invention is further described in detail:

[0043] Example 1

[0044] See attached Figure 1 As shown, a force control robot control method based on fuzzy control includes the following steps:

[0045] S1: The shape data of the controlled object is collected in real time through the data acquisition module. The shape data includes external shape, size and material. A data processing queue is created based on the collected shape data of the controlled object, and the data is transmitted to the fuzzy processing module in sequence through the data processing queue. For a force-controlled robot, the execution of related actions can be quantified by the magnitude and direction of the force applied to the controlled object, and the magnitude and direction of the applied force are generally determined based on the external shape, size and material of the controlled object. Therefore, the acquisition module first collects the shape data of the controlled object in real time, and transmits it to the fuzzy processing module for subsequent fuzzy processing to generate the corresponding control signal.

[0046] S2: The fuzzy processing module performs fuzzy processing on the shape data of the controlled object, and outputs the control signal through the fuzzy processing module; the signal prediction module predicts the control signal within the expected time period based on the control signal within the specified time period, and creates a control signal queue. The shape data of the controlled object is fuzzy processed by the fuzzy processing module, and the control signal is output through the fuzzy processing module. Since there is no need to establish an accurate mathematical model between the shape data of the controlled object and the control signal during the force control process of the force control robot, the control logic of the force control robot is simplified, the control system is simpler, easier to implement and maintain, and the cost is lower. In the case of changes in the shape data of the controlled object or interference from the external environment, the robot's control system can still have high stability. The signal prediction module predicts the control signal within the expected time period based on the control signal within the specified time period. The fuzzy processing module does not need to perform fuzzy processing on the shape data of the controlled object at each moment. After the force control robot has worked for a specified time, it can be predicted based on the control signal within the specified time. The fuzzy processing process is a time-consuming stage in the process of the force control robot. Therefore, by predicting the control signal, the control system time of the force control robot is effectively reduced, the execution delay of the force control robot is reduced, and the execution efficiency of the force control robot is improved.

[0047] S3: Verify the predicted control signal through the controlled object shape data collected in real time by the data acquisition module, and determine whether the predicted control signal is abnormal. If so, extract the corresponding controlled object shape data and transmit it to the fuzzy processing to output the control signal after fuzzy processing, and replace the corresponding control signal in the control signal queue based on the output control signal after fuzzy processing. If not, execute step S4. Verify the predicted control signal through the controlled object shape data collected in real time, and determine whether the predicted control signal is abnormal, so as to further ensure the accuracy of the predicted control signal, and process the inaccurate control signal, that is, the abnormal control signal, to ensure the accuracy of the final control signal received by the control module.

[0048] S4: The control signal queue sequentially transmits the control signal to the signal conversion module for converting the digital signal into an analog signal, and transmits the converted analog signal to the control module, and the control module controls the actuator of the force control robot accordingly based on the analog signal. By creating a control signal queue, the control module has better fluency and further improves the control efficiency of the system.

[0049] S5: Collect real-time force data of the execution action of the actuator, and feed back the real-time force data to the control module through the feedback module. The control module compares the acquired real-time force data with the standard force data corresponding to the control signal to determine whether the real-time force data is within the error range. If so, the control module continues to control the actuator of the force-controlled robot through subsequent control signals. If not, execute step S6.

[0050] S6: Compare the real-time force data with the standard force data corresponding to the control signal, including the direction and magnitude of the force, calculate the angle between the direction of the force of the real-time force data and the direction of the force of the standard force data, and the difference between the magnitude of the force of the real-time force data and the magnitude of the force of the standard force data, and make compensation adjustments through the control signal based on the angle and the difference, so that the angle and magnitude of the real-time force data and the standard force data corresponding to the control signal are within the error range. By feeding back the real-time force data to the control module through the feedback module, the control module compares the acquired real-time force data with the standard force data corresponding to the control signal, and when the real-time force data is not within the error range, the control module performs signal compensation processing, thereby further improving the action accuracy of the actuator.

[0051] In this embodiment, there are several controlled objects in step S1, and the data processing queue is created sequentially based on the several controlled objects. The order of data processing in the data processing queue is the processing order of the several controlled objects.

[0052] Example 2

[0053] On the basis of Example 1, the fuzzy processing module in step S2 performs fuzzy processing on the shape data of the controlled object, and the specific process of outputting the control signal through the fuzzy processing module is as follows:

[0054] S21: Create a fuzzy processing database and a fuzzy processing rule base for the shape data of the controlled object, wherein the fuzzy processing database stores the membership vector values ​​of all fuzzy subsets of the input shape data of the controlled object and the output control variables. Since the input shape data must be fuzzified before it can be used to solve the control output during fuzzy processing, it is actually the input interface of the fuzzy controller to realize the conversion of the real deterministic input into a fuzzy vector.

[0055] S22: Based on converting the shape data of the controlled object in the data processing queue into a corresponding fuzzy vector through the fuzzification interface, and solving the fuzzy relation equation for the fuzzy vector based on the fuzzy processing database and the fuzzy processing rule base. The fuzzy processing database stores the membership vector values ​​of all fuzzy subsets of all input and output variables, that is, the set of corresponding values ​​after the domain level is discretized, and if the domain is a continuous domain, it is a membership function. In the process of solving the fuzzy relation equation of rule reasoning, the corresponding data is provided to the inference engine.

[0056] S23: Obtaining a fuzzy control signal based on the solution result of the fuzzy relationship equation.

[0057] In this embodiment, the fuzzy processing module in step S2 performs fuzzy processing on the shape data of the controlled object using a two-dimensional fuzzy controller, and the two input variables of the two-dimensional fuzzy controller are the deviation and deviation change of the size of the controlled object from the input standard value. Since the deviation and deviation change of the size of the controlled object from the input standard value can be combined to better reflect the dynamic characteristics, the control signal generated by the fuzzy processing module after fuzzy processing has a higher accuracy.

[0058] Example 3

[0059] On the basis of Embodiment 1 or Embodiment 2, the specific process of the signal prediction module in step S2 predicting the control signal within the expected time period based on the control signal within the specified time period is as follows:

[0060] The control signal within the specified time period is obtained, and part of the control signal within the specified time period is input into the signal prediction module to train the model, and the parameters of the signal prediction module are iteratively updated through the gradient descent optimization algorithm, and another part of the control signal within the specified time period is input into the signal prediction module, and the control signal is output through the signal prediction module to obtain the prediction result, and it is determined whether the prediction result is within the expected range. If so, the control signal within the expected time period is output, and if not, the parameters of the signal prediction module are re-optimized and updated through the particle swarm algorithm. By training the signal prediction module until it has the prediction accuracy that meets the requirements, the predicted control signal is accurate, and the accuracy of the subsequent execution action is guaranteed.

[0061] In this embodiment, the specific process of iteratively updating the parameters of the signal prediction module by the gradient descent optimization algorithm is as follows:

[0062] Initialize the weights and biases of the signal prediction module, that is, set a set of weight values ​​and bias values ​​of the signal prediction module, input part of the control signal within the specified time period into the signal prediction module for forward propagation to obtain the predicted value of the control signal, and calculate the error between the predicted value and the standard value through the loss function, and calculate the gradient based on the calculated error through back propagation. The gradient is the influence of the change of weights and biases on the loss function value, and use the gradient to update the weights and biases of the signal prediction module. The updated weight value and bias value are obtained by subtracting the product of the learning rate and the gradient from the current weight and current bias of the signal prediction module, respectively. The learning rate is a set parameter, and finally iterate the above process until the preset number of iterations is met or the loss function value is less than the preset threshold. By continuously adjusting the parameters of the signal prediction module to minimize the loss function, the prediction of the model is infinitely close to the true value, thereby improving the accuracy of the prediction.

[0063] In step S3, the specific process of verifying the predicted control signal by using the controlled object shape data collected in real time by the data acquisition module to determine whether the predicted control signal has an abnormality is as follows:

[0064] S31: establishing a mapping relationship table between shape data and control signals based on historical controlled object shape data and historical control signals;

[0065] S32: Based on the current shape data, the mapping relationship table is retrieved, and the corresponding control signal in the mapping relationship table is extracted. The corresponding control signal is compared with the predicted control signal to determine whether the extracted corresponding control signal is within the preset confidence interval. If so, it indicates that the predicted control signal is normal. If not, it indicates that the predicted control signal is abnormal. The predicted control signal is verified by the real-time collected shape data of the controlled object to determine whether the predicted control signal is abnormal, further ensuring the accuracy of the predicted control signal, and processing the inaccurate, i.e., abnormal, control signal to further ensure the accuracy of the final control signal received by the control module.

[0066] See also Figure 3 The specific process of performing compensation adjustment based on the angle and the difference through the control signal in step S6 is as follows:

[0067] S61: based on the angle between the force direction of the real-time force data and the force direction of the standard force data, calculating a first ratio of the angle to the angle between the force direction of the standard force data and a certain coordinate direction;

[0068] S62: performing reverse compensation on the force direction signal in the control signal based on the first ratio, wherein the reverse compensation means that the force direction of the real-time force data is compensated in the opposite direction to the force direction of the standard force data;

[0069] S63: based on the difference between the magnitude of the force of the real-time force data and the magnitude of the force of the standard force data, calculating a second ratio of the difference to the magnitude of the force of the standard force data;

[0070] S64: Perform reverse compensation on the force signal in the control signal based on the second ratio, wherein reverse compensation means that when the force of the real-time force data is greater than the force of the standard force data, the force signal in the control signal is adjusted down by a corresponding ratio, and when the force of the real-time force data is less than the force of the standard force data, the force signal in the control signal is adjusted up by a corresponding ratio. By feeding back the real-time force data to the control module through the feedback module, the control module compares the acquired real-time force data with the standard force data corresponding to the control signal, and when the real-time force data is not within the error range, the control module performs signal compensation processing, thereby further improving the action accuracy of the actuator.

[0071] See also Figure 2 A force-controlled robot control system based on fuzzy control is used to implement any one of the force-controlled robot control methods based on fuzzy control, characterized in that it includes a data acquisition module, a data processing queue, a fuzzy processing module, a control signal queue, a signal prediction module, a signal verification module, a signal conversion module, a control module, an actuator and a feedback module. The data acquisition module is connected to the data processing queue, the data processing queue is connected to the fuzzy processing module, the fuzzy processing module is respectively connected to the control signal queue and the signal prediction module, the signal prediction module is connected to the signal verification module, the control signal queue is connected to the signal conversion module, the signal conversion module is connected to the control module, the control module is connected to the actuator, and the actuator is connected to the feedback module.

[0072] A data acquisition module is used to collect the shape data of the controlled object in real time, and the shape data includes external shape, size and material; a fuzzy processing module is used to perform fuzzy processing on the shape data of the controlled object and output a control signal through the fuzzy processing module; a control module is used to control the action of the actuator based on the control signal, and to perform compensation adjustment through the control signal based on the angle and difference, so that the angle and size of the real-time force data and the standard force data corresponding to the control signal are within the error range; an actuator is used to perform corresponding actions based on the control signal of the control module; a feedback module is used to feed back the real-time action performed by the actuator to the control module.

[0073] In summary, the force control robot control method and system based on fuzzy control provided by the present invention performs fuzzy processing on the shape data of the controlled object through the fuzzy processing module, and outputs the control signal through the fuzzy processing module. It is not necessary to establish an accurate mathematical model between the shape data of the controlled object and the control signal in the force control process of the force control robot, which simplifies the control logic of the force control robot, and the control system is simple in structure, easy to implement and maintain, and has lower cost. Even in the case of changes in the shape data of the controlled object or interference from the external environment, the control system of the robot has high stability. The control signal within the expected time period is predicted based on the control signal within the specified time period by the signal prediction module. The fuzzy processing module does not need to perform fuzzy processing on the shape data of the controlled object at each moment. After the force control robot works for a specified time, it can be predicted through the control signal within the specified time. The fuzzy processing process is a stage that takes a long time in the process of the force control robot. Therefore, the prediction of the control signal will greatly reduce the control system time of the force control robot, reduce the execution delay of the force control robot, and improve the execution efficiency of the force control robot. By verifying the predicted control signal with the real-time collected shape data of the controlled object, it is determined whether the predicted control signal has any abnormality, further ensuring the accuracy of the predicted control signal, and processing the inaccurate, i.e., abnormal, control signal to ensure the accuracy of the final control signal received by the control module. By feeding back the real-time force data to the control module through the feedback module, the control module compares the acquired real-time force data with the standard force data corresponding to the control signal. When the real-time force data is not within the error range, the control module performs signal compensation processing, further improving the action accuracy of the actuator.

Claims

1. A force control robot control method based on fuzzy control, characterized in that: The following steps are involved: S1: The shape data of the controlled object is collected in real time through the data collection module, and the shape data includes external shape, size and material, and a data processing queue is created based on the collected shape data of the controlled object, and the data is sequentially transmitted to the fuzzy processing module through the data processing queue; S2: the fuzzy processing module performs fuzzy processing on the shape data of the controlled object, and outputs a control signal through the fuzzy processing module; the signal prediction module predicts the control signal within the expected time period based on the control signal within the specified time period, and creates a control signal queue; S3: verifying the predicted control signal through the shape data of the controlled object collected in real time by the data acquisition module, and judging whether the predicted control signal is abnormal. If so, extracting the corresponding shape data of the controlled object and transmitting it to the fuzzy processing to output the control signal after fuzzy processing, and replacing the corresponding control signal in the control signal queue based on the output control signal after fuzzy processing. If not, executing step S3; S4: the control signal queue sequentially transmits the control signal to the signal conversion module for converting the digital signal into an analog signal, and transmits the converted analog signal to the control module, and the control module controls the actuator of the force control robot accordingly based on the analog signal; S5: collecting real-time force data of the execution action of the actuator, and feeding back the real-time force data to the control module through the feedback module, the control module compares the acquired real-time force data with the standard force data corresponding to the control signal, and determines whether the real-time force data is within the error range. If so, the control module continues to control the actuator of the force-controlled robot through subsequent control signals. If not, executing step S6; S6: comparing the real-time force data with the standard force data corresponding to the control signal, including the direction and magnitude of the force, calculating the angle between the direction of the force of the real-time force data and the direction of the force of the standard force data and the difference between the magnitude of the force of the real-time force data and the magnitude of the force of the standard force data, and performing compensation adjustment through the control signal based on the angle and the difference, so that the angle and magnitude of the real-time force data and the standard force data corresponding to the control signal are within the error range; The specific process of the fuzzy processing module in step S2 performing fuzzy processing on the shape data of the controlled object and outputting the control signal through the fuzzy processing module is as follows: S21: creating a fuzzy processing database and a fuzzy processing rule base for the shape data of the controlled object, wherein the fuzzy processing database stores the membership vector values ​​of all fuzzy subsets of the input shape data of the controlled object and the output control variables; S22: converting the shape data of the controlled object in the data processing queue into a corresponding fuzzy vector through a fuzzification interface, and solving a fuzzy relation equation for the fuzzy vector based on a fuzzy processing database and a fuzzy processing rule base; S23: Obtaining a fuzzy control signal based on the solution result of the fuzzy relationship equation; The specific process of the signal prediction module in step S2 predicting the control signal within the expected time period based on the control signal within the specified time period is as follows: Obtaining a control signal within a specified time period, and inputting part of the control signal within the specified time period into the signal prediction module to train the model, and iteratively updating the parameters of the signal prediction module through a gradient descent optimization algorithm, and inputting another part of the control signal within the specified time period into the signal prediction module, and outputting the control signal through the signal prediction module to obtain a prediction result, and judging whether the prediction result is within an expected range, if so, outputting the control signal within the expected time period, if not, re-optimizing and updating the parameters of the signal prediction module through a particle swarm algorithm; In step S3, the specific process of verifying the predicted control signal by using the controlled object shape data collected in real time by the data acquisition module to determine whether the predicted control signal has an abnormality is as follows: S31: establishing a mapping relationship table between shape data and control signals based on historical controlled object shape data and historical control signals; S32: Retrieve the mapping relationship table based on the current appearance data, extract the corresponding control signal in the mapping relationship table, compare the corresponding control signal with the predicted control signal, and determine whether the extracted corresponding control signal is within a preset confidence interval. If so, it indicates that the predicted control signal is normal, if not, it indicates that the measured control signal is abnormal.

2. A force control robot control method based on fuzzy control according to claim 1, characterized in that: There are several controlled objects in step S1, and the data processing queue is created in sequence based on the several controlled objects. The order of data processing in the data processing queue is the processing order of the several controlled objects.

3. The method for controlling a force-controlled robot based on fuzzy control according to claim 1, characterized in that: In step S2, the fuzzy processing module performs fuzzy processing on the shape data of the controlled object using a two-dimensional fuzzy controller, and two input variables of the two-dimensional fuzzy controller are the deviation amount and the deviation change amount of the size of the controlled object and the input standard value.

4. The method for controlling a force-controlled robot based on fuzzy control according to claim 1, characterized in that: The specific process of iteratively updating the parameters of the signal prediction module by the gradient descent optimization algorithm is as follows: Initialize the weights and biases of the signal prediction module, that is, set a set of weight values ​​and bias values ​​of the signal prediction module, input part of the control signal within the specified time period into the signal prediction module for forward propagation to obtain a predicted value of the control signal, and calculate the error between the predicted value and the standard value through a loss function, calculate the gradient based on the calculated error through back propagation, and the gradient is the influence of the change in weight and bias on the loss function value, and use the gradient to update the weights and biases of the signal prediction module, and the updated weight values ​​and bias values ​​are obtained by subtracting the product of the learning rate and the gradient from the current weight and current bias of the signal prediction module, respectively, and the learning rate is a set parameter, and finally iterate the above process until a preset number of iterations is met or the loss function value is less than a preset threshold.

5. The method for controlling a force-controlled robot based on fuzzy control according to claim 1, characterized in that: The specific process of performing compensation adjustment based on the angle and the difference through the control signal in step S6 is as follows: S61: based on the angle between the force direction of the real-time force data and the force direction of the standard force data, calculating a first ratio of the angle to the angle between the force direction of the standard force data and a certain coordinate direction; S62: performing reverse compensation on the force direction signal in the control signal based on the first ratio, wherein the reverse compensation means that the force direction of the real-time force data is compensated in the opposite direction to the force direction of the standard force data; S63: based on the difference between the magnitude of the force of the real-time force data and the magnitude of the force of the standard force data, calculating a second ratio of the difference to the magnitude of the force of the standard force data; S64: Perform reverse compensation on the force signal in the control signal based on the second ratio, wherein the reverse compensation means that when the force of the real-time force data is greater than the force of the standard force data, the force signal in the control signal is adjusted down by a corresponding ratio; and when the force of the real-time force data is less than the force of the standard force data, the force signal in the control signal is adjusted up by a corresponding ratio.

6. A force-controlled robot control system based on fuzzy control, used to implement a force-controlled robot control method based on fuzzy control as described in any one of claims 1 to 5, characterized in that: It includes a data acquisition module, a data processing queue, a fuzzy processing module, a control signal queue, a signal prediction module, a signal inspection module, a signal conversion module, a control module, an actuator and a feedback module, wherein the data acquisition module is connected to the data processing queue, the data processing queue is connected to the fuzzy processing module, the fuzzy processing module is respectively connected to the control signal queue and the signal prediction module, the signal prediction module is connected to the signal inspection module, the control signal queue is connected to the signal conversion module, the signal conversion module is connected to the control module, the control module is connected to the actuator, and the actuator is connected to the feedback module; The data acquisition module is used to collect the shape data of the controlled object in real time, and the shape data includes external shape, size and material; the fuzzy processing module is used to perform fuzzy processing on the shape data of the controlled object and output a control signal through the fuzzy processing module; The control module is used to control the action of the actuator based on the control signal, and to perform compensation adjustment through the control signal based on the angle and the difference, so that the angle and size of the real-time force data and the standard force data corresponding to the control signal are within the error range; the actuator is used to perform corresponding actions based on the control signal of the control module; the feedback module is used to feed back the real-time action performed by the actuator to the control module.

Images