End effector force control method and device based on improved PSO algorithm

Abstract

The invention relates to the technical field of surface finishing machining, in particular to an end effector force control method and device based on an improved PSO algorithm, and the method comprises the steps: S1, setting the expected contact force of an end effector; s2, collecting actual contact force data, and calculating a contact force deviation and a contact force deviation change rate; s3, obtaining a fuzzy control quantity according to the contact force deviation and the contact force deviation change rate modulus, and calculating three parameter change values of PID control; s4, iteratively acquiring three parameter initial values of PID control based on an improved PSO algorithm; and S5, calculating parameters of the optimized fuzzy adaptive PID according to the initial value and the change value of the PID control, and performing contact force control on the end effector. According to the method, the control precision is improved, meanwhile, the response speed is high, better robustness is achieved, and a better control effect can be obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of surface finishing, and in particular to a method and device for controlling the force of an end effector based on an improved PSO algorithm. Background Art

[0002] At present, intelligent industrial robots have become the main solution for surface finishing operations due to their low cost, high flexibility, and easy integration with various end effectors and sensors. When using industrial robots to perform surface finishing operations, the key to determining the finishing effect is the control of the grinding and polishing contact force and motion position. The end effector needs to decouple the contact force and motion position, and accurately control the robot's position and the contact force between the end effector and the workpiece.

[0003] PID controller is a feedback loop control system widely used in industrial control. It controls through three key parameters (adjustment coefficient P, integral time constant I, differential time constant D) in a linear combination manner. It has the characteristics of simple structure, good stability, and no need for precise system model.

[0004] However, surface finishing is a dynamic environment where multiple controlled objects change together. The contact force between the end effector and the workpiece is affected by many factors such as the robot's position, posture, and acceleration. When the characteristics of the controlled object change, the traditional PID controller cannot be optimized in real time, and its parameters need to be re-tuned to achieve the best. The main method for optimizing the parameters of the PID controller in the prior art is the manual tuning method, which requires the user to have skilled techniques, is inefficient, and has low parameter accuracy.

[0005] The PID control method can be superimposed with the fuzzy reasoning step to optimize the control parameters of the PID. Compared with the traditional PID method, it can achieve more accurate control effects, but there are still shortcomings. The design of the fuzzy controller depends on fuzzy rules, which are usually manually formulated by experienced experts. For complex systems, it may be very difficult to design a suitable set of fuzzy rules. At the same time, if the initial PID parameters are not selected properly, it may cause problems such as slow system response, excessive overshoot or oscillation, affecting the control accuracy and stability. In some cases, the fuzzy controller may fall into a local optimal solution and fail to find the global optimal control strategy, resulting in control performance that is not as expected. At the same time, the debugging process of the fuzzy PID controller is relatively complicated, which increases the development cycle and cost.

[0006] Therefore, it is necessary to develop a more efficient end-effector contact force control method. Summary of the invention

[0007] The present invention provides an end effector force control method and device based on an improved PSO algorithm, which can solve the current problems of difficulty in constant force control of contact force, unstable surface quality, poor product consistency, etc.

[0008] The present invention provides the following technical solutions:

[0009] A method for controlling the force of an end effector based on an improved PSO algorithm comprises the following steps:

[0010] S1, set the desired contact force f of the end effector;

[0011] S2, collecting the actual normal contact force data f0, calculating the contact force deviation e by comparing f with f0, taking the derivative of the contact force deviation e with time as the variable, and obtaining the contact force deviation change rate ec;

[0012] S3, according to e and ec, look up the table in the fuzzy rule base to obtain the fuzzy control quantity, and calculate the three parameter change values ​​of PID control according to the fuzzy control quantity

[0013] S4. Iteratively obtain the initial values ​​of the three parameters of PID control based on the improved PSO algorithm

[0014] S5. Calculate and optimize the parameter K of the fuzzy adaptive PID based on the initial value and change value of the PID control P , K I , K D , and the contact force of the end effector is controlled.

[0015] Furthermore, in S5, the parameter K of the fuzzy adaptive PID P , K I , K D Calculate according to the following formula:

[0016]

[0017] Further, the S4 includes:

[0018] S41, introducing optimization functions Φ1 and Φ2, where optimization function Φ1 is used to enhance the search capability of particles; and optimization function Φ2 is used to increase the convergence speed of the algorithm;

[0019] S42. Establish velocity evolution model and position evolution model of particles;

[0020] The particle velocity evolution model is as follows:

[0021] v ij (t+1)=Φ1[v ij (t)+c1r1(p ij (t)-xij (t))]+Φ2r2c2(p gj (t)-x ij (t))

[0022] The particle velocity evolution model is as follows:

[0023] x ij (t+1)=x ij (t)+v ij (t+1)

[0024] Among them, i represents the i-th particle, j represents the j-th dimension, and x ij (t) is the position of the particle at time t, v ij (t) is the velocity of the particle at time t, c1 and c2 are independent random functions between (0, 1); r1 is the cognitive coefficient, r2 is the social learning coefficient, r1 and r2 take values ​​between (0, 2), p ij (t) is the best position experienced by particle i before time t, p gj (t) is the global optimal value of the jth dimension before the tth iteration;

[0025] S43. Initialize the parameter values ​​of the improved PSO algorithm:

[0026] For any i and j, if in [-X max ,X max ], the particles are generated in a uniform distribution, and the initial position of the i-th particle in the j-dimension is x ij (0), the initial velocity is v ij (0), the fitness value of the i-th particle is J i For the first PSO calculation, the parameters have not been iterated, and the fitness value of the i-th particle is J i (0), the optimal fitness value of the i-th particle is J ibest , and J i (0) equals J ibest ; J ibest is the optimal fitness value of the i-th particle, J gbest is the optimal fitness value of the global particle, p ij (t) is the best position experienced by particle i before time t, p gj (t) is the global optimal value of the jth dimension before the tth iteration;

[0027] Compare the fitness values ​​of all particles in the population and obtain the best fitness value J of the global particle gbest , at this time the best fitness value J of the global particle gbest Equal to P gj (0), P in the initialization velocity model formula gj (t);

[0028] S44. Calculate the fitness value J of each particle for each evolution i :

[0029]

[0030] Where, e(f) = f - f0(t);

[0031] If J i <J ibest , then J i =J ibest , p ij (t) = x ij (t);

[0032] If J i <J gbest , then J i =J gbest , p gj (t) = x ij (t);

[0033] S45. Update the speed and position of each particle according to the speed evolution model and the position evolution model, where x i1 (t+1), x i2 (t+2), x i3 (t+3) are the 3 initial values ​​of the PID controller

[0034] S46, when t <t max Repeat S44 and S45 when t=t max The improved PSO algorithm ends when the last evolution is completed. As the initial value of the fuzzy adaptive PID, t max is the cutoff evolutionary generation.

[0035] Furthermore, the optimization functions Φ1 and Φ2 are specifically as follows:

[0036]

[0037] Among them, η is a coefficient between (0, 1), t is the evolutionary generation, and t max is the cutoff evolutionary generation.

[0038] The principles and advantages of the present invention are:

[0039] The traditional PID controller is improved. The changes of the three parameters of the PID controller are calculated based on fuzzy control. The initial quantities of the three parameters of the PID controller are solved based on the improved PSO algorithm. Finally, the initial quantities and the changes are superimposed to obtain the resetting parameters of the PID controller.

[0040] The present invention introduces a fuzzy adaptive control strategy on the basis of the basic PID controller, which can realize the precise control of the contact force of the end effector. At the same time, the initial parameters of the PID controller are solved based on the improved PSO algorithm, and the optimization function is introduced on the basis of the traditional PSO algorithm, which can effectively avoid the problems of premature convergence and falling into local optimality of the standard PSO algorithm, and at the same time improve the convergence speed of the algorithm, and can quickly and accurately find the optimal PID initial parameter value, which provides control accuracy while also having a faster response speed, better robustness, and better control effect.

[0041] On the other hand, the present invention provides an end effector force control device based on an improved PSO algorithm, comprising a data acquisition module, a control module and an execution module, wherein:

[0042] The data acquisition module includes a pressure sensor disposed on the end effector for collecting the actual contact force; and also includes a human-machine interaction interface disposed on the control end of the end effector for obtaining the expected contact force;

[0043] The control module includes an AD conversion module, a programmable logic controller and a communication module. The programmable logic controller includes a fuzzy adaptive PID controller based on an improved PSO algorithm. The AD conversion module receives the collected data of the actual contact force and completes the analog-to-digital conversion and sends it to the programmable logic controller. The fuzzy adaptive PID controller based on the improved PSO algorithm generates a control signal of the contact force according to the data of the expected contact force and the actual contact force, and sends it to the execution module for execution through the communication module.

[0044] The execution module includes an active force control device arranged on the end effector, and adjusts the contact force with the processed workpiece according to the contact force control signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of a force control method of the end effector based on the improved PSO algorithm;

[0046] Figure 2 This is a structural schematic diagram of an end-effector force control device based on the improved PSO algorithm. DETAILED DESCRIPTION

[0047] The following is further described in detail through specific implementation methods:

[0048] Embodiment 1

[0049] A force control method for the end effector based on the improved PSO algorithm, such as Figure 1 As shown, the following steps are included:

[0050] S1, set the desired contact force f of the end effector;

[0051] The operator sets various working parameters on the interactive interface of the end effector, including the expected contact force f, tool weight, safe stroke, etc.

[0052] S2, collecting actual contact force data f0 in the normal direction through a force sensor set at the end effector, comparing f with f0, calculating the contact force deviation e, taking the derivative of the contact force deviation e with time as a variable, and obtaining the contact force deviation change rate ec;

[0053] S3, according to e and ec, look up the table in the fuzzy rule base to obtain the fuzzy control quantity, and calculate according to the fuzzy control quantity to obtain the change values ​​of the three parameters of PID control

[0054] The contact force deviation e is divided into 5 fuzzy sets, namely: negative large NB, negative small NS, zero O, positive small PS, positive large PB, and the change levels are specified as -3, -2, -1, 0, 1, 2, 3; the control quantity u is divided into 5 fuzzy sets, namely: negative large NB, negative small NS, zero O, positive small PS, positive large PB, and the change levels are specified as: -4, -3, -2, -1, 0, 1, 2, 3, 4, then the fuzzy table of the control quantity can be obtained as follows:

[0055] Table 1 Fuzzy table of control quantity

[0056] -4 -3 -2 -1 0 1 2 3 4 PB 0 0 0 0 0 0 0 0.5 1 PS 0 0 0 0 0 0.5 1 0.5 0 ZO 0 0 0 0.5 1 0.5 0 0 0 NS 0 0.5 1 0.5 0 0 0 0 0 NB 1 0.5 0 0 0 0 0 0 0

[0057] The fuzzy rules are specified as follows: (1) if the deviation e is negative and large, then the control quantity u is negative and large;

[0058] (2) If the deviation e is small, then the control quantity u is small;

[0059] (3) If the deviation e is 0, the control amount u is 0;

[0060] (4) If the deviation e is small, then the control quantity u is small;

[0061] (5) If the deviation e is positive and large, then the control quantity u is positive and large.

[0062] S4. Iteratively obtain the initial values ​​of the three parameters of PID control based on the improved PSO algorithm

[0063] The specific steps include:

[0064] S41, introducing optimization functions Φ1 and Φ2;

[0065] The PSO optimization algorithm is a swarm intelligence algorithm designed by simulating the predation behavior of bird flocks, which is used to solve optimization problems. The particle in the PSO algorithm represents a solution, and the global optimal solution is found by moving the position and speed of the particle in the search space. The position of the particle represents a solution to the problem being solved, while the speed of the particle represents the direction and distance the particle moves in the next iteration. The advantage of the PSO algorithm is that it is easy to implement and understand, and is applicable to various types of problems. However, the conventional PSO algorithm has the following defects: it cannot balance the global and local search capabilities of the particles well; it is easy to fall into iteration stagnation too early and fail to obtain the optimal solution.

[0066] This application improves the basic PSO algorithm by adding optimization functions Φ1 and Φ2. The optimization function Φ1 allows the particle swarm to consider the example's own factors in the early stage of evolution and enhance the search ability of the particles to select suitable seeds. The specific formula is as follows:

[0067]

[0068] Among them, η is a coefficient between (0, 1), t is the evolutionary generation, and t max is the cutoff value of the evolutionary generation, Φ1 decreases with the increase of the evolutionary generation;

[0069] The role of the optimization function Φ2 is to allow the particle swarm to consider sharing information in the later stage of evolution and increase the convergence speed of the algorithm. The specific formula is as follows:

[0070]

[0071] Among them, η is a coefficient between (0, 1), t is the evolutionary generation, and t max is the cutoff evolutionary generation, Φ1 decreases with the increase of evolutionary generation, and Φ2 increases with the increase of evolutionary generation;

[0072] S42. Establish velocity evolution model and position evolution model of particles;

[0073] The particle velocity evolution model is as follows:

[0074] v ij (t+1)=Φ1[v ij (t)+c1r1(p ij (t)-x ij (t))]+Φ2r2c2(p gj (t)-x ij (t)) (3)

[0075] The particle velocity evolution model is as follows:

[0076] x ij (t+1)=x ij (t)+vij (t+1) (4)

[0077] Among them, i represents the i-th particle, j represents the j-th dimension, and x ij (t) is the position of the particle at time t, v ij (t) is the velocity of the particle at time t, c1 and c2 are independent random functions between (0, 1); r1 is the cognitive coefficient, r2 is the social learning coefficient, r1 and r2 take values ​​between (0, 2), p ij (t) is the best position experienced by particle i before time t, p gj (t) is the global optimal value of the jth dimension before the tth iteration, j = 1, 2, 3;

[0078] S43, initializing the improved PSO algorithm parameter values;

[0079] Assume that in n-dimensional space, the number of particles in the group is m, and the end of the evolutionary generation is t. max , set η to be between (0, 1);

[0080] For any i and j, if in [-X max ,X max ], the particles are generated in a uniform distribution, and the initial position of the i-th particle in the j-dimension is x ij (0), the initial velocity is v ij (0), the fitness value of the i-th particle is J i For the first PSO calculation, the parameters have not been iterated, and the fitness value of the i-th particle is J i (0), the optimal fitness value of the i-th particle is J ibest , and J i (0) equals J ibest .J ibest is the optimal fitness value of the i-th particle, J gbest is the optimal fitness value of the global particle, p ij (t) is the best position experienced by particle i before time t, p gj (t) is the global optimal value of the j-th dimension before the t-th iteration.

[0081] Compare the fitness values ​​of all particles in the population and obtain the best fitness value J of the global particle gbest , at this time the best fitness value J of the global particle gbest Equal to P gj (0), P in the initialization velocity model formula gj (t).

[0082] S44. Calculate the fitness value J of each particle for each evolution i .

[0083] Ji Calculated based on the unit step response according to the squared deviation integral function:

[0084]

[0085] Among them, e(f)=f-f0(t).

[0086] If J i <J ibest , then J i =J ibest , p ij (t) = x ij (t);

[0087] If J i <J gbest , then J i =J gbest , p gj (t) = x ij (t).

[0088] S45. Update the speed and position of each particle according to the speed evolution model and the position evolution model.

[0089] In this application, the PID controller has three parameter values, v i1 (t+1), v i2 (t+2), v i3 (t+3) represents the speed change of the three parameters of the PID controller, x i1 (t+1), x i2 (t+2), x i3 (t+3) represents the position of the three parameters of the PID controller, x i1 (t+1), x i2 (t+2), x i3 (t+3) are the 3 initial values ​​of the PID controller

[0090] S46, when t <t max Repeat S44 and S45 when t=t max The improved PSO algorithm ends when the last evolution is completed. As the initial value of fuzzy adaptive PID.

[0091] S5. Calculate and optimize the parameter K of the fuzzy adaptive PID based on the initial value and change value of the PID control P , K I , K D , and the contact force of the end effector is controlled.

[0092] Parameter K of fuzzy adaptive PID P, K I , K D The calculation formula is:

[0093]

[0094] Embodiment 2

[0095] This embodiment is a force control device for an end effector based on an improved PSO algorithm, including a data acquisition module, a control module and an execution module. Figure 2 As shown, where:

[0096] The data acquisition module is set at the end effector, including a pressure sensor integrated in the end effector and a human-machine interface at the control end. Among them, the human-machine interface is used by the operator to set various working parameters, including the expected contact force, tool weight, safe stroke, etc. The end effector is equipped with an active force control device and a pneumatic grinding head. The active force control device drives the pneumatic grinding head to contact the surface of the workpiece. A pressure sensor is also installed between the active force control device and the grinding head tool to collect the normal pressure of the pneumatic grinding head, that is, the actual contact force data.

[0097] The control module includes an AD conversion module, a programmable logic controller and a communication module. The AD conversion module uses a data acquisition card to receive the collected data of the actual contact force, and sends it to the programmable logic controller after completing the analog-to-digital conversion. The programmable logic controller includes a fuzzy adaptive PID controller based on an improved PSO algorithm. The fuzzy adaptive PID controller generates a control signal of the contact force according to the data of the expected contact force and the actual contact force, and sends it to the execution module for execution through the communication module.

[0098] The execution module is an active force control device installed on the end effector. It drives the electromagnetic proportional valve to control the cylinder according to the contact force control signal and adjusts the contact force with the workpiece in real time.

[0099] The detailed working process of each module is as described in Example 1.

[0100] The above are only embodiments of the present invention. The invention is not limited to the field involved in this implementation case. The common knowledge such as the known specific structure and characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A method for end effector force control based on an improved PSO algorithm, characterized in that: The steps include: S1, set the desired contact force f of the end effector; S2, collecting the actual normal contact force data f0, calculating the contact force deviation e by comparing f with f0, taking the derivative of the contact force deviation e with time as the variable, and obtaining the contact force deviation change rate ec; S3, according to e and ec, look up the table in the fuzzy rule base to obtain the fuzzy control quantity, and calculate the three parameter change values ​​of PID control according to the fuzzy control quantity S4. Iteratively obtain the initial values ​​of the three parameters of PID control based on the improved PSO algorithm S5. Calculate and optimize the parameter K of the fuzzy adaptive PID based on the initial value and change value of the PID control P , K I , K D , contact force control of the end effector.

2. The end effector force control method based on the improved PSO algorithm according to claim 1 is characterized in that: In S5, the parameter K of the fuzzy adaptive PID P , K I , K D Calculate according to the following formula:

3. The end effector force control method based on the improved PSO algorithm according to claim 2 is characterized in that: The S4 includes: S41, introducing optimization functions Φ1 and Φ2, where optimization function Φ1 is used to enhance the search capability of particles; and optimization function Φ2 is used to increase the convergence speed of the algorithm; S42. Establish velocity evolution model and position evolution model of particles; The particle velocity evolution model is as follows: v ij (t+1)=Φ1[v ij (t)+c1r1(p ij (t)-x ij (t))]+Φ2r2c2(p gj (t)-x ij (t)) The particle velocity evolution model is as follows: x ij (t+1)=x ij (t)+v ij (t+1) Among them, i represents the i-th particle, j represents the j-th dimension, and x ij (t) is the position of the particle at time t, v ij (t) is the velocity of the particle at time t, c1 and c2 are independent random functions between (0, 1); r1 is the cognitive coefficient, r2 is the social learning coefficient, r1 and r2 take values ​​between (0, 2), p ij (t) is the best position experienced by particle i before time t, p gj (t) is the global optimal value of the jth dimension before the tth iteration; S43. Initialize the parameter values ​​of the improved PSO algorithm: For any i and j, if in [-X max ,X max ], the particles are generated in a uniform distribution, and the initial position of the i-th particle in the j-dimension is x ij (0), the initial velocity is v ij (0), the fitness value of the i-th particle is J i ; For the first PSO calculation, the parameters have not been iterated, and the fitness value of the i-th particle is J i (0), the optimal fitness value of the i-th particle is J ibest , and J i (0) equals J ibest ; J ibest is the optimal fitness value of the i-th particle, J gbest is the optimal fitness value of the global particle, p ij (t) is the best position experienced by particle i before time t, p gj (t) is the global optimal value of the jth dimension before the tth iteration; Compare the fitness values ​​of all particles in the population and obtain the best fitness value J of the global particle gbest , at this time the best fitness value J of the global particle gbest Equal to P gj (0), P in the initialization velocity model formula gj (t); S44. Calculate the fitness value J of each particle for each evolution i : Where, e(f) = f - f0(t); If J i <J ibest , then J i =J ibest , p ij (t) = x ij (t); If J i <J gbest , then J i =J gbest , p gj (t) = x ij (t); S45. Update the speed and position of each particle according to the speed evolution model and the position evolution model, where x i1 (t+1), x o2 (t+2), x i3 (t+3) are the 3 initial values ​​of the PID controller S46, when t <t max Repeat S44 and S45 when t=t max The improved PSO algorithm ends when the last evolution is completed. As the initial value of the fuzzy adaptive PID, t max is the cutoff evolutionary generation.

4. The end effector force control method based on the improved PSO algorithm according to claim 3 is characterized by: The optimization functions Φ1 and Φ2 are specifically as follows: Among them, η is a coefficient between (0, 1), t is the evolutionary generation, and t max is the cutoff evolutionary generation.

5. An end effector force control device based on an improved PSO algorithm, characterized in that: The end effector force control device based on the improved PSO algorithm implements the end effector force control method based on the improved PSO algorithm as claimed in any one of claims 1 to 4, including a data acquisition module, a control module and an execution module, wherein: The data acquisition module includes a pressure sensor disposed on the end effector for collecting the actual contact force; and also includes a human-machine interaction interface disposed on the control end of the end effector for obtaining the expected contact force; The control module includes an AD conversion module, a programmable logic controller and a communication module. The programmable logic controller includes a fuzzy adaptive PID controller based on an improved PSO algorithm. The AD conversion module receives the collected data of the actual contact force and completes the analog-to-digital conversion and sends it to the programmable logic controller; the fuzzy adaptive PID controller based on the improved PSO algorithm generates a contact force control signal according to the data of the expected contact force and the actual contact force, and sends it to the execution module for execution through the communication module; the execution module includes an active force control device arranged on the end effector, which adjusts the contact force with the processed workpiece according to the contact force control signal.

Images