Bionic mechanical arm elbow joint adjusting angular velocity control method

Abstract

The invention belongs to the technical field of mechanical arm control, relates to a bionic mechanical arm elbow joint adjusting angular velocity control method, and solves the problem that an existing method is difficult to meet the collaborative requirements of smooth starting, smooth transition and quick response in bionic actions. Acquiring starting and target positions to generate a motion trail; deducing an expected angle sequence based on forward and reverse kinematics, calculating a target angular velocity, and collecting current angle and angular velocity information in real time; an angle error signal is constructed, an angular velocity control instruction is generated by using a position type PID controller, and PID parameters are dynamically set through an improved algorithm based on state optimization; and a control instruction is input into the elbow joint driving unit, and dynamic adjustment and closed-loop control of the angular speed are achieved. The effect of the method is verified through experiments, and the experiments prove that the precision, robustness and dynamic response capability of angular velocity control of the bionic mechanical arm can be remarkably improved through the method.

Description

Technical Field

[0001] This invention relates to the field of robotic arm control technology, and more particularly to a method for controlling the angular velocity adjustment of the elbow joint of a bionic robotic arm. Background Technology

[0002] With the development of biomimetic robotics technology, robotic arms are increasingly being used in medical rehabilitation, industrial collaboration, and human-computer interaction. To achieve precise trajectory tracking of the robotic arm's end effector, dynamic adjustment of the joint angles and angular velocities is typically required. Among these, angular velocity control, as a crucial link between trajectory planning and drive execution, directly affects the robotic arm's response speed and motion stability.

[0003] Traditional angular velocity control methods are mostly based on closed-loop regulation of joint position errors, often using proportional-integral-derivative (PID) controllers. However, in simulating natural human movements, due to the nonlinear regulatory characteristics and phased response features of the human muscle and nervous system, traditional PID control suffers from problems such as high regulation rigidity, poor adaptability, and uncoordinated response phases, making it difficult to meet the coordinated requirements of smooth start, stable transition, and rapid response in biomimetic movements. In addition, some existing control methods lack a dynamic update mechanism for the feedback model during the joint angular velocity control process, failing to adjust control commands in real time according to different movement stages, leading to accumulated trajectory deviations and decreased system robustness. Therefore, there is an urgent need for an angular velocity control method that integrates human biomimetic characteristics and has strong adaptive capabilities to improve the dynamic performance and control accuracy of biomimetic robotic arms.

[0004] To improve control performance, intelligent optimization algorithms have been gradually introduced into the angular velocity control of robotic arms. State-Based Optimization (SBO) is an emerging intelligent optimization method that achieves rapid search for the global optimum by simulating the state interaction behavior between individuals in a group. The SBO algorithm introduces an elite interaction stage (global search stage) and a resource acquisition stage (local development stage) to jointly model the historical state of individuals, the global state of the group, and environmental adaptability, thereby achieving efficient optimization in high-dimensional complex spaces. Compared with traditional particle swarm optimization or genetic algorithms, SBO has significant advantages in convergence speed, global search capability, and avoidance of local extrema. Summary of the Invention

[0005] This invention aims to solve the problems of unstable trajectory tracking, insensitive control response, and poor adaptability during motion phase in existing bionic robotic arm joint angular velocity control methods. It proposes a bionic robotic arm joint adjustable angular velocity control method, which combines the characteristics of human upper limb movements and achieves efficient, smooth, and high-precision control of joint movements by constructing a two-dimensional position model of the end effector and a dynamic angular velocity control strategy.

[0006] To achieve the above-mentioned objectives, the present invention employs the following technical solution: This application relates to a method for controlling the angular velocity adjustment of the elbow joint of a bionic robotic arm, the specific method being as follows: Using the elbow position of the bionic robotic arm as the origin, a two-dimensional position model of the end of the elbow joint of the bionic robotic arm is constructed to obtain the starting position and the ending position, and to generate the motion trajectory data of the bionic robotic arm; the motion trajectory is composed of N1 points (x,y) in the two-dimensional position model; Based on the principle of inverse kinematics, the expected angle sequence of the robotic arm's elbow joint in the trajectory is calculated, the corresponding angular velocity data is derived from the expected angle sequence, and the current angular velocity information of the elbow joint is collected in real time. An angular velocity error signal is constructed based on the current angular velocity and the target angular velocity. The error signal is then calculated by a positional PID controller to generate an angular velocity control command. The positional PID controller dynamically adjusts the control parameters using an improved state optimization algorithm. The angular velocity control command is input to the elbow joint drive unit of the bionic robotic arm to achieve dynamic adjustment and closed-loop control of the joint angular velocity.

[0007] The bionic robotic arm establishes a Cartesian two-dimensional position model with the elbow position as the origin; it obtains the starting and ending positions of the target action, and generates continuous motion trajectory data by simulating the trajectory change trend of the lower arm controlled by the human elbow joint during natural movement. The trajectory data consists of several discrete points (x, y) in a two-dimensional plane, and these points are connected to form a curved motion trajectory; wherein the X-axis extends along the horizontal direction and the Y-axis extends along the vertical direction.

[0008] The process involves: knowing the coordinates of the starting and target points, calculating the initial angle based on the starting and origin positions, and then calculating the final elbow joint angle based on the target position and the elbow joint origin position; selecting a point as the target position based on N points (x, y) in the motion trajectory, and calculating the angle change sequence between the target position and the starting position according to the inverse kinematics principle; performing the rate of change of the angle difference with respect to time on adjacent points in the angle change sequence to obtain angle change rate data, and generating angular velocity data at the target position, which is then the target angular velocity data; This invention studies the elbow joint angular velocity control problem of a bionic robotic arm. Therefore, it only includes the expected angle sequence of the elbow joint under the trajectory and further calculates the target angular velocity data. At the same time, the real-time angular position and angular velocity of the elbow joint are collected as feedback information to provide dynamic state input for subsequent closed-loop control.

[0009] Preferably, the elbow joint of the bionic robotic arm is driven by a servo motor. The motion angle of the elbow joint is calculated in the two-dimensional position model at the end of the elbow joint of the bionic robotic arm based on the position of the end of the elbow joint and the origin position, and is recorded as the motion angle data of the elbow joint at the current moment. The current angular velocity information is calculated based on the difference between the current angle data and the angle data of the previous sampling. The current angular velocity value is obtained by comparing the angle data before and after the sampling interval with the sampling interval and using the difference method.

[0010] Preferably, the positional PID controller calculates the error signal to generate an angular velocity control command. The positional PID controller calculates the angular error signal using proportional control parameters, integral control parameters, and derivative control parameters, and outputs the angular velocity control command at the current moment. The control parameters of the PID controller are dynamically adjusted by an improved state optimization algorithm. The improved state optimization algorithm constructs an evaluation function based on feedback state variables, and achieves online optimization of the proportional control parameters, integral control parameters, and derivative control parameters through the adaptive adjustment strategy designed in this invention. This ensures that the control parameters for calculating the angular velocity control command achieve optimal accuracy, thereby enhancing the robustness and adjustment response capability of the bionic robotic arm elbow joint control system to dynamic disturbances.

[0011] Preferably, the adaptive adjustment strategy includes an elite emergence jump mechanism based on status fluctuation entropy and elite degeneration index and a dimension-aware perturbation update method.

[0012] Preferably, the control parameters of the PID controller are dynamically adjusted by an improved state-based optimization algorithm, specifically as follows: A mapping is established between the individual's position vector and the control parameter vector of the PID controller. The PID controller's control parameters include proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd. Each set of control parameters constructs a three-dimensional vector. After mapping, the position vector of the three-dimensional vector is... The dimension value is the first of the individual position vectors. Dimension The fitness value of each agent's position vector is calculated using an evaluation function. The fitness values ​​of each agent are then sorted sequentially. The agent's position vector with the lowest fitness value is taken as the global optimum. The optimal individuals are selected as the elite set; In the process of optimization using a randomly initialized, improved state-based optimization algorithm, the individual position is determined by the first position vector of the individual. Dimension Randomly assign a value, setting the value range to be between the maximum value ub and the minimum value lb; We design an elite emergence jump mechanism based on status fluctuation entropy and elite degeneration index to enable individuals to update their positions by jumping to new elite directions; and we construct a position update strategy for the resource acquisition stage through a dimension-aware perturbation update method to enable individuals to update their positions during the resource acquisition stage. The response of the updated proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd to the dynamic adjustment of joint angular velocity is calculated by evaluating the function value. This result is recorded as the updated fitness value. Then, the individual position is updated again through the elite emergence jump mechanism and the dimension-aware perturbation update method. The iteration count increases by one each time the iteration count is returned until the current iteration count reaches the maximum iteration count. At this point, the dynamic optimization ends, and the values ​​of proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd corresponding to the minimum fitness value are output. This value is recorded as the optimal combination of control parameters for the PID controller.

[0013] Preferably, the evaluation function adopts an overall integral absolute error design. Based on the deviation between the actual elbow joint adjustment angular velocity of the controlled object and the target elbow joint adjustment angular velocity during the control cycle, the integral absolute error (IAE) in the target response process is calculated, and the IAE is used as the individual fitness value.

[0014] Preferably, the positional PID controller is reconstructed by combining the optimal control parameters of the PID controller to calculate the error signal, generate an angular velocity control command, and input it into the elbow joint drive unit of the bionic robotic arm. The elbow joint drive unit adjusts the servo actuator to achieve dynamic adjustment and closed-loop control of the joint angular velocity. Specifically, the angular velocity control process of the bionic robotic arm elbow joint is based on the dynamic response model of the elbow joint drive unit of the bionic robotic arm. The dynamic response model uses the angular velocity control command generated by the positional PID controller as the input drive voltage signal, which acts on the servo actuator to dynamically control the actuator current. The actuator current is converted into motor output torque through the motor torque constant. The output torque drives the robotic arm elbow joint to generate angular velocity changes around the rotation axis. Finally, a response mathematical model between the control signal and the angular velocity is constructed based on the second-order dynamic response model to output the adjustable angular velocity of the bionic robotic arm elbow joint.

[0015] Preferably, the specific method for the elite emergence jump mechanism based on status fluctuation entropy and elite degeneration index is as follows: Extract individuals In the near Local state change data sequence within the next iteration cycle The distance between adjacent states is calculated to construct the state fluctuation entropy index of the individual, and a weighting coefficient is constructed based on the relative fluctuation intensity of each iteration. Using the distance and weighting coefficients as input, an individual... State fluctuation entropy index ; Randomly select an elite individual from the elite set, and then... The state fluctuation entropy index is used to calculate the elite individuals associated with the individual. State fluctuation entropy And calculate the state difference factor based on the state fluctuation entropy index of the individual and the associated elite individuals. ; Get the current individual Construct trajectory vectors from historical positions in the current and two previous iteration cycles. The current evolution curvature factor of the individual is calculated based on the directional angle of the trajectory vector and the path length. Based on the individual's current evolutionary curvature factor and combined with the curvature response adjustment factor, an attraction weighting factor for the individual's attraction to elite individuals is established using a nonlinear function. The attraction weighting factor is used to control the degree of guidance an individual receives at their current location; For each elite individual Perform state monitoring and count the number of consecutive iterations that fail to achieve positive evolutionary improvement. The number of iterations is compared with the preset maximum tolerance threshold. Perform ratio calculations to obtain trend indicators used to determine whether degradation has occurred. If the trend index of any elite individual meets the degradation condition, based on the state difference factor of its associated individuals, and excluding itself, the elite individual whose state is most different from the current elite individual is selected as the jump target, and a new leap elite index is constructed accordingly. Based on the elite index and attraction weight factor, combined with the standard perturbation structure, the current individual... The position vector is updated to enable individuals to make updates and jumps under the new elite direction.

[0016] Preferably, the original resource acquisition stage location update strategy constructs a mask variable vector based on a preset number of dimensions and performs location updates based on the mask values. The generation method of the mask variables relies entirely on random functions, without considering the influence of each variable dimension, namely the proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd, on the optimization results of the evaluation function, resulting in a lack of specificity in the resource fusion update. Based on this, a dimension-aware perturbation update method is proposed, in which each dimension value is subjected to target-oriented perturbation to strengthen the adjustment of the differences in each dimension value.

[0017] Preferably, the dimension-aware perturbation update method specifically comprises: Based on the current individual The sensitivity of the evaluation function value to changes during each optimization process is assessed, and each variable dimension is constructed. Dimensional contribution metrics The dimensional contribution index is the contribution of each variable dimension value to the dynamic adjustment effect value of the joint angular velocity, where the effect value is evaluated by an evaluation function. Calculation feedback; Based on the aforementioned dimensional contribution index, a modulation factor is constructed. And for the current individual in each dimension The value is updated as follows to achieve dimension-aware perturbation update, wherein the modulation factor is proportional to the dimension contribution index and is used to adjust the direction of dimension value adjustment; wherein the mathematical model of dimension-aware perturbation update is: ; in, This represents the updated value of the j-th dimension at the position of the i-th individual. This represents the value of the j-th dimension at the current i-th individual position. This is the globally optimal solution for the j-th dimension in the current population.

[0018] Preferably, in the process of controlling the angular velocity of the robotic arm elbow joint, the sensitivity of each dimension of the control parameter to the response in different control processes is significantly different. This is manifested in that some dimension disturbances have a significant promoting effect on error convergence, while other dimensions exhibit unstable behaviors such as disturbance reversal and disturbance amplification. Therefore, a dimension contribution index is introduced to quantify the dynamic contribution of each dimension in the historical adjustment process. Further analysis shows that by calculating the partial derivatives of different dimensions in the historical function changes, the dimension contribution index can identify the main control parameters that are sensitive to error in the current stage, thereby giving them a greater adjustment weight in the disturbance update.

[0019] Preferably, the modulation factor is calculated for each dimension of each individual, based on the change in position of that dimension over M consecutive iterations. The unit direction of the dimension is defined based on the magnitude of the change over M consecutive steps. By evaluating the unit directional changes in this dimension during historical iterations, a directional consistency index is constructed based on the perturbation directional consistency integral. The directional consistency index for each individual takes a value between -1 and 1. Based on the dimensional contribution index of the target individual in this dimension, and combined with the calculated directional consistency index, a moderating factor is generated. ;in, Let be the modulation factor for the i-th individual in the t-th iteration. The contribution index for the dimension of the i-th individual in the t-th iteration.

[0020] Preferably, the evaluation function adopts an overall integral absolute error design. Based on the deviation between the actual elbow joint adjustment angular velocity of the controlled object and the target elbow joint adjustment angular velocity during the control cycle, the integral absolute error (IAE) in the target response process is calculated, and the IAE is used as the individual fitness value.

[0021] Based on the above-described invention, the beneficial effects of the present invention can be summarized as follows: First, by constructing a two-dimensional position model of the elbow joint end, and combining forward and inverse kinematics to derive the desired angle and angular velocity sequence, the natural movement characteristics of the human elbow are simulated, generating a continuous and smooth control target trajectory, which effectively improves the compliance and realism of bionic motion, and is suitable for rehabilitation training, human-computer collaboration and other scenarios. Secondly, an elite emergence jump mechanism based on status fluctuation entropy and elite degeneration index is proposed, and combined with the dimension-aware perturbation update method, the adaptive update of PID controller parameters is realized. By adaptively adjusting the controller parameters, the problem of insensitivity of fixed parameters to angular velocity response during the movement of the elbow joint of the bionic robotic arm is effectively avoided, so that the angular velocity control process has strong dynamic response capability and anti-interference performance. Finally, the angular velocity control command is used to drive the change of current in the servo actuator, and the torque is converted into torque through the motor torque constant, which ultimately drives the change of elbow joint angular velocity. A second-order dynamic response mathematical model from PID control signal to angular velocity output is constructed to ensure that the control command can be accurately and in real time mapped to the angular velocity control of the bionic robotic arm elbow joint. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the implementation of a bionic robotic arm elbow joint angular velocity control method according to the present invention.

[0023] Figure 2 This is a structural block diagram of a bionic robotic arm elbow joint angular velocity control method according to the present invention.

[0024] Figure 3 This is a simulation diagram of the elbow joint angular velocity control of a bionic robotic arm according to the present invention.

[0025] Figure 4 This is a fitness diagram of the elbow joint adjustment angular velocity control of a bionic robotic arm according to the present invention.

[0026] Figure 5 The diagram shows the results of optimizing the control parameters of the elbow joint position PID control algorithm to improve the previous control algorithm.

[0027] Figure 6 The diagram shows the results of optimizing the control parameters of the elbow joint position PID control algorithm after the improvement of the control algorithm.

[0028] Figure 7This diagram illustrates the control effect of a bionic robotic arm elbow joint angular velocity adjustment control method according to the present invention. Detailed Implementation

[0029] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0030] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention. In the description of this invention, it should be understood that the terms "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0031] To achieve precise control of the angular velocity of the elbow joint of a bionic robotic arm, this application relates to a method for adjusting the angular velocity of the elbow joint of a bionic robotic arm, see [link to relevant documentation]. Figure 1 The implementation of the joint angular velocity control method is described in detail.

[0032] S1. Using the elbow position of the bionic robotic arm as the origin, construct a two-dimensional position model of the end of the elbow joint of the bionic robotic arm, obtain the starting position and the ending position, and generate the motion trajectory data of the bionic robotic arm; the motion trajectory is composed of N points (x,y) in the two-dimensional position model; S2. Based on the principle of inverse kinematics, calculate the expected angle sequence of the robotic arm elbow joint in the trajectory, derive the corresponding angular velocity data based on the expected angle sequence, and collect the current angular velocity of the elbow joint in real time. S3. Construct an angular velocity error signal based on the collected current angular velocity and target angular velocity, and calculate the error signal using a position-based PID controller to generate an angular velocity control command; The position-based PID controller dynamically adjusts the control parameters using an improved state-based optimization algorithm; S4. Input the angular velocity control command to the elbow joint drive unit of the bionic robotic arm to realize dynamic adjustment and closed-loop control of the joint angular velocity.

[0033] In some embodiments of this application, the bionic robotic arm elbow joint angular velocity control system includes a trajectory planning module, a joint angle solving module, a sensing and acquisition module, an angular velocity controller module, and an elbow joint drive unit.

[0034] In the embodiments of this application, the trajectory planning module is used to construct a two-dimensional position model of the elbow joint end effector based on the set starting position and target position. Specifically, the bionic robotic arm establishes a Cartesian two-dimensional position model with the elbow position as the origin (0,0); and obtains the starting position of the action. and the finish line Where N1=1,2,3...,100, continuous motion trajectory data is generated by simulating the trajectory change trend of the lower arm controlled by the elbow joint in a human body. The trajectory data consists of 100 discrete points in a two-dimensional plane. These points are connected to form a curved trajectory; the X-axis extends horizontally and the Y-axis extends vertically.

[0035] In the embodiments of this application, based on N points (x, y) in the motion trajectory, one point is selected as the target position, and the angle change sequence between the target position and the starting position is calculated; the rate of change of the angle difference with respect to time is calculated for adjacent position points in the angle change sequence to obtain angle change rate data, and angular velocity data at the target position is generated. At this time, the angular velocity data is the target angular velocity data. In this application, the expected angle sequence in the trajectory is calculated, and the expected elbow joint angle sequence is derived based on inverse kinematics during the implementation process. The mathematical model is: ; in, and Let k = 1, 2, 3, ..., N, be the horizontal and vertical position coordinates of a randomly selected bionic robotic arm end effector on a two-dimensional plane. and The coordinates of the starting position of the bionic robotic arm end effector in the horizontal and vertical directions on the two-dimensional plane; Based on the desired angle sequence, the corresponding angular velocity data sequence is derived, and the mathematical model is as follows: ; in, This represents the angular velocity corresponding to the target angle when the k-th position is the target's position coordinate. The angle between the starting position and the origin of the elbow joint is here. The time taken from the starting position to the target position. The target angle is the coordinate of the kth discrete point in the desired elbow joint angle sequence, which is the coordinate of the target position. In the embodiments of this application, the target position is a discrete point selected from continuous motion trajectory data (curved motion trajectory), denoted as... Set the time from the initial position to the target position. The data is then incorporated into the angular velocity data sequence. The target angular velocity value of this embodiment is obtained. .

[0036] The sensor acquisition module collects the current angular velocity data of the elbow joint in real time. The error calculation unit of the angular velocity controller module recalculates the error with the target angular velocity value based on the current data. The angular velocity controller module includes an error calculation unit, a position-based PID control algorithm unit, and an angular velocity control command output unit. It performs error calculation between the target angle and the real-time angle, and calculates the angular velocity control command through an improved position-based PID control algorithm. The control parameters in the position-based PID controller are dynamically and adaptively adjusted through an improved state optimization algorithm.

[0037] As mentioned above, PID parameters include proportional parameter Kp and integral parameter Ki. i The improved state optimization algorithm of this application, along with the differential parameter Kd, is an improvement on the standard state optimization algorithm. It mainly involves the improvement of the position update strategy. The control parameters of the PID controller described in this application are dynamically adjusted by the improved state optimization algorithm. The improved state optimization algorithm constructs an evaluation function based on the feedback state quantity and realizes online optimization of the proportional control parameters, integral control parameters, and differential control parameters through the adaptive adjustment strategy designed in this invention, so that the control parameters for calculating the angular velocity control command achieve the best accuracy.

[0038] Please see Figure 2 As shown, the process of online optimization and adjustment of the PID parameters of the speed PID controller by the improved state optimization algorithm is described in detail.

[0039] S31. Establish a mapping between the individual's position vector and the control parameter vector of the PID controller; wherein, the control parameters of the PID controller include proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd. Each set of control parameters constructs a three-dimensional vector. After mapping, the position vector of the three-dimensional vector is... The dimension value is the first of the individual position vectors. Dimension ,in .

[0040] S32. Set the initialization parameters for the improved state optimization algorithm. The initialization parameters include: maximum number of iterations T, problem dimension D, upper bound position vector ub and lower bound position vector lb of the solution space for algorithm optimization, and population size N. In this embodiment, the maximum number of iterations T=80, the problem dimension D=3, the upper limit position vector of the solution space of the algorithm optimization ub=[90 90 90], the limit position vector lb=[0.01 0.01 0.01], and the population size N=20; Step S32 is used for the initialization phase of the improved state optimization algorithm, which includes generating each individual position using a random method between the upper bound position vector ub and the lower bound position vector lb in the solution space, so as to provide an initial value for subsequent individual position updates; During the initialization phase, the positions of individuals in the population are randomly generated within the search space: ,in Let be the j-th dimension value of the i-th individual, and let the whole be the individual position vector. and Let be the upper bound position vector and lower bound position vector of the i-th individual in the solution space, and rand be a random number between 0 and 1.

[0041] S33. Calculate the fitness value of each agent's position vector using the evaluation function. Sort the fitness values ​​of each agent in order, and take the position vector value of the agent with the smallest fitness value as the global optimal solution. (The first two options in the sorted list are not directly related to the fitness value calculation and can be omitted.) The top 5 individuals are selected as the elite set; in this embodiment, the top 5 individuals are selected as the elite set. The evaluation function is used to evaluate the accuracy of the solution corresponding to each individual position and the rate of change of the evaluation function value to evaluate the optimization speed of the solution. The smaller the fitness value calculated by the evaluation function, the higher the accuracy of the solution. The faster the fitness value changes, the faster the optimization speed of the solution. As described above, this application provides an evaluation function model. The evaluation function adopts a global integral absolute error design. Based on the deviation between the actual elbow joint adjustment angular velocity and the target elbow joint adjustment angular velocity of the controlled object during the control cycle, the integral absolute error (IAE) in the target response process is calculated. The mathematical model is as follows: ,in The total runtime is 60 seconds in Siumulink.

[0042] S34. An improved state optimization algorithm is obtained by reconstructing the global search and local development stages of the state optimization algorithm based on the elite emergence jump mechanism and the dimensional perception perturbation update method based on the status fluctuation entropy and elite degeneration index, which updates the position vector of the individual. By constructing position update strategies for the elite contact stage and resource acquisition stage, a global search strategy and a local development strategy for optimizing the control parameters of the PID controller are realized. Based on the above, the elite emergence jump mechanism based on status fluctuation entropy and elite degradation index proposed in this application is used as the global search strategy in the elite contact stage, and the dimension-aware perturbation update method proposed in the resource acquisition stage is used as the local development strategy.

[0043] The global search strategy includes constructing a state fluctuation entropy index. Calculate the state difference factor Establish attraction weight factors And design indicators to determine whether a trend is deteriorating To enable individuals to make new leaps and advancements in the direction of the new elite.

[0044] In this embodiment, individual samples are extracted. In the near Local state change data sequence within the next iteration cycle Calculate the distance between adjacent states and construct weighting coefficients based on the relative fluctuation intensity of each iteration. , ; in, and Let i be the position vector of the i-th individual in the t-th and t-1-th iterations. To find the minimum value, avoid the denominator being zero; Using the distance and weighting coefficients as input, an individual is constructed. State fluctuation entropy index ; In some embodiments of this application, Set to 0.00001.

[0045] In this embodiment, an elite individual is randomly selected from the elite set, and based on the individual... The state fluctuation entropy index is used to calculate the elite individuals associated with the individual. State fluctuation entropy And calculate the state difference factor based on the state fluctuation entropy index of the individual and the associated elite individuals. .

[0046] In this embodiment, the current individual is obtained. Construct trajectory vectors from historical positions in the current and two previous iteration cycles. The current evolution curvature factor of the individual is calculated based on the directional angle of the trajectory vector and the path length. The individual's current evolutionary curvature factor is designed based on the discrete curvature principle; the mathematical model is as follows: ;in, Not equal to ; Based on the individual's current evolutionary curvature factor and combined with the curvature response adjustment factor, an attraction weighting factor for the individual to elite individuals is established using a nonlinear function. The attraction weighting factor is used to control the degree of guidance an individual receives at their current location, and whether it enhances their dependence on historical elites. The mathematical model is as follows: ; in, This is the curvature response adjustment factor. In this embodiment, its value must be greater than zero. The specific value needs to be adjusted based on the effect of the attraction weighting factor. The initial setting is 0.5. The curvature baseline is set and manually adjusted based on the individual's current evolutionary curvature factor. It serves as a stability boundary threshold, ranging from 0 to 1; in this embodiment, it is set to 0.2. Let be the evolution curvature factor of the i-th individual in the trajectory of three consecutive generations; A larger individual evolution curvature factor indicates that the trajectory direction of position updates is unstable, suggesting that individual position updates tend to depend on elite individuals. Then attract weighting factors The curvature factor tends towards 1. Conversely, the smaller the individual evolution curvature factor, the closer it is to 0, indicating that the evolutionary direction of position update is stable and that individual position update tends to depend on its own position, thus attracting weighting factors. It approaches 0.

[0047] In this embodiment, for each elite individual Perform state monitoring and count the number of consecutive iterations that fail to achieve positive evolutionary improvement. The number of iterations is compared with the preset maximum tolerance threshold. Perform ratio calculations to obtain trend indicators used to determine whether degradation has occurred. The number of consecutive times an elite individual's position remains unchanged is determined by comparing the individual's fitness value in each iteration. This is a fixed parameter, representing the maximum number of iterations allowed for an elite individual to maintain its current state; it is set to 10. If the trend indicator of any elite individual satisfies the degradation... Greater than condition; This embodiment introduces an elite jump mechanism to break local convergence. Based on the state difference factor of its associated individuals, and excluding itself, it selects the elite individual whose state is most different from the current elite individual as the jump target, and constructs a new jump elite index accordingly. The mathematical model is: .

[0048] In this embodiment, based on the elite index and attraction weight factor, combined with the standard perturbation structure, the current individual is... The position vector is updated to enable individuals to make updates and jumps under the new elite direction; ; in, This is the updated position vector of the i-th individual. It is a standard normal distribution with a mean of 0 and a variance of 1, used to perturb the individual position and reduce the probability of the individual position update getting trapped in a local optimum.

[0049] Local development strategies include dimension-aware perturbation update methods, which calculate dimensional contribution metrics. Constructing modulation factors Achieve dimension-aware perturbation updates by updating a specific dimension of the individual's position vector.

[0050] In this embodiment, based on the current individual The sensitivity of the evaluation function value to changes during each optimization process is assessed, and each variable dimension is constructed. Dimensional contribution metrics The dimensional contribution index is the contribution of each variable dimension value to the dynamic adjustment effect value of the joint angular velocity, where the effect value is calculated and fed back through an evaluation function; the mathematical model of the dimensional contribution index is: ; in, To determine the maximum number of iterations, Let be the fitness value of the individual in the t-th iteration. Let be the value of the j-th dimension of the individual in the t-th iteration.

[0051] In this embodiment, a modulation factor is constructed based on the dimensional contribution index. And for the current individual in each dimension The value is updated as follows to achieve dimension-aware perturbation update, wherein the modulation factor is proportional to the dimension contribution index and is used to adjust the direction of dimension value adjustment; wherein the mathematical model of dimension-aware perturbation update is: ; in, This represents the updated value of the j-th dimension at the position of the i-th individual. This represents the value of the j-th dimension at the current i-th individual position. This is the globally optimal solution for the j-th dimension in the current population.

[0052] The modulation factor is calculated for each dimension of each individual, representing the change in position of that dimension over M consecutive iterations. The unit direction of the dimension is defined based on the magnitude of the change over M consecutive steps. By evaluating the unit directional changes in this dimension during historical iterations, a directional consistency index is constructed based on the perturbation directional consistency integral. The mathematical model for the directional consistency index of each individual is as follows: ;in, Let be the directional consistency index for the i-th individual in the t-th iteration. Let be the unit direction of the dimension of the i-th individual in the t-th iteration. Let be the unit direction of the dimension of the i-th individual in the (t-1)-th iteration; the direction consistency index of each individual takes a value between [-1, 1]. Based on the dimensional contribution index of the target individual in this dimension, combined with the calculated direction consistency index, a regulation factor is generated. ;in, Let be the modulation factor for the i-th individual in the t-th iteration. The contribution index for the dimension of the i-th individual in the t-th iteration.

[0053] S35. After the location is updated, the fitness value is recalculated to determine the optimal individual location vector value. After the number of updates reaches the maximum number of iterations, the individual position vector corresponding to the minimum fitness value is output, and the globally optimal proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd are obtained analytically.

[0054] In this embodiment, the globally optimal values ​​of the proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd are used to calculate the error signal and generate angular velocity control commands. Specifically, the first dimension of the individual's location vector corresponding to the minimum fitness value. Assign the value to the proportional control parameter Kp, the second dimension value Assign the value to the proportional control parameter Ki, the third dimension. The mathematical model for generating angular velocity control commands using the position PID control algorithm, assigned to the proportional control parameter Kd, is as follows: ; in, The angular velocity control command for the current control cycle; This represents the angular velocity error value for the current control cycle. This is the angular velocity error value from the previous control cycle.

[0055] The elbow joint drive unit is used to receive angular velocity control commands, adjust the working state of the servo actuator, drive the elbow joint to perform corresponding movements at the desired angular velocity, and realize closed-loop tracking control. The physical model of the elbow joint is the servo actuator. The angular velocity control process of the bionic robotic arm elbow joint is based on the dynamic response model constructed by the servo motor drive mechanism. In this embodiment, the elbow joint of the bionic robotic arm is controlled by a small servo motor. The torque of the servo motor is controlled by the driver through the angular velocity control command, thereby achieving the control of the angular velocity. This embodiment establishes a servo drive model, that is, a drive model between the angular velocity of the elbow joint (servo motor) and the angular velocity control command.

[0056] The dynamic response model establishes a second-order dynamic response equation based on Kirchhoff's laws. The mathematical model is as follows: ; in, This is the angular velocity control command for the current control cycle. For armature inductance, For armature resistance, The back electromotive force constant is... The current angular velocity in rad / s during the current control cycle.

[0057] In this embodiment, the armature inductance L = 0.005H, the armature resistance R = 1.0Ω, and the back electromotive force constant is... It is 0.05 V·s / rad; Based on the moment of inertia and damping characteristics, a dynamic relationship between torque and angular velocity is constructed, and a dynamic relationship between torque and angular velocity is established. The mathematical model is as follows: ; in, For rotational inertia, The viscous damping coefficient is... The current output torque is given; in this embodiment, the moment of inertia J = 0.01 kg·m², and the viscous damping coefficient B = 0.02 N·m·s / rad; the linear relationship between current and torque is then established. By simultaneously substituting the equations, we can establish the second-order dynamic response equation from the control signal to the angular velocity. ; in, The torque constant is used in this embodiment. =0.05 N·m / A; The second-order dynamic response equation is obtained. .

[0058] In this embodiment, three methods are used to control the controlled function of the present invention: the pre-improved control algorithm, the post-improved control algorithm, and the positional PID control algorithm. The angular velocity is set to 9 rad / s, meaning the differential term is zero once the angular velocity reaches steady state. Therefore, based on the above, the angular velocity control command for the current control cycle should be 4V. With the angular velocity set to 31 rad / s, the angular velocity control command for the current control cycle should be 14V. The total running time is 60 seconds. Please refer to [link to relevant documentation]. Figure 3 As shown, the second-order dynamic response equation is expressed in complex frequency domain form. In the Siumulink simulation model of the elbow joint drive unit, which serves as the input of a controlled function to the angular velocity control system of the bionic robotic arm elbow joint, the system is run to obtain the following results: Figures 4 to 7 The effects of the method of the present invention.

[0059] Further, please see Figure 3 and Figure 4 As shown, Figure 3 In the fitness value monitoring section, the fitness value of the improved control algorithm of this invention for optimizing the control parameters of the elbow joint position PID control algorithm is 12.314; corresponding to Figure 4 The fitness values ​​of the control algorithm before and after the improvement are compared. The fitness value of the control algorithm before the improvement for the control parameters of the elbow joint position PID control algorithm is 81.7029. The smaller the fitness value, the higher the control accuracy, indicating that the improved control algorithm is better at optimizing the control parameters of the elbow joint position PID control algorithm.

[0060] Further, please see Figure 5 and Figure 6 As shown, in this embodiment, the improved control algorithm optimizes the control parameters of the elbow joint position PID control algorithm with Kp of 13.1784, Ki of 11.3693, and Kd of 0.01; the original control algorithm optimizes the control parameters of the elbow joint position PID control algorithm with Kp of 0.856797, Ki of 1.15773, and Kd of 0.274371.

[0061] Further, please see Figure 7As shown, the Siumulink simulation model was set to a total running time of 60 seconds, monitoring the angular velocity control commands. The control results clearly demonstrate that the improved control algorithm of this invention outperforms the previous control algorithm and the positional PID control algorithm in both accuracy and speed for controlling the angular velocity of the bionic robotic arm's elbow joint. The improved control algorithm did not exhibit overshoot in controlling the angular velocity of the bionic robotic arm's elbow joint, while the previous and positional PID control algorithms both showed significant overshoot. Furthermore, the improved control algorithm has a shorter steady-state time for controlling the angular velocity of the bionic robotic arm's elbow joint. This proves that this method can significantly improve the accuracy, robustness, and dynamic response capability of the bionic robotic arm's angular velocity control.

Claims

1. A method for controlling the angular velocity adjustment of the elbow joint of a bionic robotic arm, characterized in that, Using the elbow position of the bionic robotic arm as the origin, a two-dimensional position model of the end of the elbow joint of the bionic robotic arm is constructed to obtain the starting position and the ending position, and to generate the motion trajectory data of the bionic robotic arm; the motion trajectory is composed of N1 points (x,y) in the two-dimensional position model; Based on the principle of inverse kinematics, the expected angle sequence of the robotic arm's elbow joint in the trajectory is calculated, the corresponding angular velocity data is derived from the expected angle sequence, and the current angular velocity information of the elbow joint is collected in real time. An angular velocity error signal is constructed based on the current angular velocity and the target angular velocity. The error signal is then calculated by a positional PID controller to generate an angular velocity control command. The positional PID controller dynamically adjusts the control parameters using an improved state optimization algorithm. The angular velocity control command is input to the elbow joint drive unit of the bionic robotic arm to achieve dynamic adjustment and closed-loop control of the joint angular velocity.

2. The method for controlling the angular velocity of the elbow joint of a bionic robotic arm according to claim 1, characterized in that, The two-dimensional position model is a Cartesian two-dimensional position model. It simulates the trajectory change trend of the lower arm in the natural movement process controlled by the human elbow joint. The starting position and ending position of the target movement are the two endpoints, and the movement is radii of the arm length, forming a fan-shaped motion trajectory data. The arm length is the length from the elbow position of the bionic robotic arm to the end of the elbow joint of the robotic arm.

3. The method for controlling the angular velocity of the elbow joint of a bionic robotic arm according to claim 2, characterized in that, The desired angle sequence is the distance between each point (x, y) on the trajectory in the two-dimensional position model and the starting position. The included angle; The method for obtaining the corresponding angular velocity data is as follows: Based on N points (x, y) in the motion trajectory, select one point as the target position, and calculate the angle change sequence between the target position and the starting position according to the inverse kinematics principle; perform the rate of change of the angle difference with respect to time on adjacent position points in the angle change sequence to obtain the angle change rate data, and generate the angular velocity data at the target position. At this time, the angular velocity data is the target angular velocity data.

4. The method for controlling the angular velocity of the elbow joint of a bionic robotic arm according to claim 3, characterized in that, The positional PID includes a proportional parameter Kp and an integral parameter Ki. i The improved state optimization algorithm constructs an evaluation function based on the feedback state quantity and uses an adaptive adjustment strategy to achieve online optimization of the proportional control parameter, integral control parameter, and derivative control parameter; the feedback state quantity is the current angular velocity.

5. The method for controlling the angular velocity of the elbow joint of a bionic robotic arm according to claim 4, characterized in that, The evaluation function employs an overall integral absolute error design, adjusting the angular velocity of the elbow joint based on the actual angular velocity within the control cycle. Adjusting the angular velocity with the target elbow joint The deviation between the two values ​​is used to calculate the integral absolute error during the target response process.

6. The method for controlling the angular velocity of the elbow joint of a bionic robotic arm according to claim 4, characterized in that, The adaptive adjustment strategy improves the position update strategy for the global search and local development phases of the state optimization algorithm. The adaptive adjustment strategy includes an elite emergence jump mechanism based on status fluctuation entropy and elite degeneration index and a dimension-aware perturbation update method.

7. The method for controlling the angular velocity adjustment of the elbow joint of a bionic robotic arm according to claim 6, characterized in that, The improvement to the global search phase of the state optimization algorithm based on the elite emergence jump mechanism using status fluctuation entropy and elite degeneration index is as follows: Extract individuals In the near Local state change data sequence within the next iteration cycle The distance between adjacent states is calculated to construct the state fluctuation entropy index of the individual, and a weighting coefficient is constructed based on the relative fluctuation intensity of each iteration. Using the distance and weighting coefficients as input, an individual... State fluctuation entropy index ; Randomly select an elite individual from the elite set, and then... The state fluctuation entropy index is used to calculate the elite individuals associated with the individual. State fluctuation entropy And calculate the state difference factor based on the state fluctuation entropy index of the individual and the associated elite individuals. ; Get the current individual Construct trajectory vectors from historical positions in the current and two previous iteration cycles. The current evolution curvature factor of the individual is calculated based on the directional angle of the trajectory vector and the path length. Based on the individual's current evolutionary curvature factor and combined with the curvature response adjustment factor, an attraction weighting factor for the individual's attraction to elite individuals is established using a nonlinear function. The attraction weighting factor is used to control the degree of guidance an individual receives at their current location; For each elite individual Perform state monitoring and count the number of consecutive iterations that fail to achieve positive evolutionary improvement. The number of iterations is compared with the preset maximum tolerance threshold. Perform ratio calculations to obtain trend indicators used to determine whether degradation has occurred. If the trend index of any elite individual meets the degradation condition, based on the state difference factor of its associated individuals, and excluding itself, the elite individual whose state is most different from the current elite individual is selected as the jump target, and a new leap elite index is constructed accordingly. Based on the elite index and attraction weight factor, combined with the standard normal distribution, through elite individuals For the current individual The relative position of the position vector is updated to realize the position update of the individual under the new elite direction.

8. The method for controlling the angular velocity of the elbow joint of a bionic robotic arm according to claim 6, characterized in that, The dimension-aware perturbation update method improves the local development phase of the state optimization algorithm, including: Based on the current individual The sensitivity of the evaluation function value to changes during each optimization process is assessed, and each variable dimension is constructed. Dimensional contribution metrics The dimensional contribution index is the contribution of each variable dimension value to the dynamic adjustment effect value of the joint angular velocity, wherein the effect value is calculated and fed back through an evaluation function; Based on the aforementioned dimensional contribution index, a modulation factor is constructed. And for the current individual in each dimension The value is updated as follows to achieve dimension-aware perturbation update, wherein the modulation factor is proportional to the dimension contribution index and is used to adjust the direction of dimension value adjustment; wherein the mathematical model of dimension-aware perturbation update is: ； in, This represents the updated value of the j-th dimension at the position of the i-th individual. This represents the value of the j-th dimension at the current i-th individual position. This is the globally optimal solution for the j-th dimension in the current population.

9. A method for controlling the angular velocity of an elbow joint in a bionic robotic arm according to claim 8, characterized in that, Dynamically adjusting the control parameters includes: A mapping is established between the individual's position vector and the control parameter vector of the PID controller; where the control parameters of the PID controller include proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd. Each set of control parameters constructs a three-dimensional vector. After mapping, the position vector of the three-dimensional vector is... The dimension value is the first of the individual position vectors. Dimension ,in ; Set the initialization parameters for the improved state optimization algorithm; the initialization parameters include: maximum number of iterations T, problem dimension D, upper bound position vector ub and lower bound position vector lb of the solution space for the algorithm to find, and population size N; The fitness value of each agent's position vector is calculated using an evaluation function. The fitness values ​​of each agent are then sorted sequentially. The agent's position vector with the lowest fitness value is taken as the global optimum. The optimal individuals are selected as the elite set; An improved state optimization algorithm is obtained by reconstructing the global search and local development stages of the state optimization algorithm based on the elite emergence jump mechanism and the dimensional perception perturbation update method, which updates the position vector of individuals. After the maximum number of updates is reached, the individual position vector corresponding to the minimum fitness value is output, and the globally optimal proportional control parameter Kp, integral control parameter Ki, and derivative control parameter Kd are obtained analytically.

Images