PID control method for AGV wheel type driving system

Abstract

The invention provides a PID (Proportion Integration Differentiation) control method for an AGV (Automatic Guided Vehicle) wheel type driving system. The method comprises the following steps: acquiring a first parameter of an AGV; filtering the first parameter, and correcting error data in the first parameter to obtain a second parameter; setting a preset speed function based on a system error model of the AGV wheel type driving system and the second parameter, enabling the AGV to operate according to the preset speed function, and combining the error model to obtain a driving gain coefficient, a friction coefficient and a speedometer gain coefficient; substituting the plurality of coefficients into the fitness function, and carrying out optimal solution solving through a preset optimization algorithm to obtain an optimal solution; verifying the optimal solution to obtain an evaluation value, and when the evaluation value is smaller than a preset threshold value, outputting a target parameter corresponding to the optimal solution; and substituting the target parameters into an AGV wheel type driving system, and controlling a chassis motor servo driving device to cooperatively complete a specified action based on the control instruction. According to the invention, the adaptive capability of the AGV to various environments is improved.

Description

Technical Field

[0001] The present application relates to the technical field of automatic guided vehicles, and in particular to a PID control method for an AGV wheel drive system. Background Art

[0002] AGV (Automated Guided Vehicle), as an unmanned vehicle, is becoming more and more widely used in industrial scenarios. The large-scale application of AGV has significantly improved the automation level of production lines and logistics efficiency, mainly including efficient material handling in various links such as warehousing, production, and distribution, reducing dependence on manual labor, thereby reducing labor costs and error rates. As AGV application scenarios continue to expand and become more complex, the demand for AGV control accuracy is also increasing. The requirements for high precision and high reliability in industrial environments have challenged traditional control methods.

[0003] The factors that affect the control accuracy of AGV mainly include the accuracy of the sensor itself, environmental conditions, navigation algorithms, the accuracy of the wheel drive system, and dynamic response capabilities. The accuracy of the sensor directly determines the accuracy of position, posture estimation, and obstacle detection, while environmental conditions (such as changes in light and uneven ground) may interfere with the sensor data, thereby affecting the positioning accuracy. The degree of optimization of the navigation algorithm affects the accuracy of path planning, and the accuracy and dynamic response capabilities of the wheel drive system are directly related to the motion control accuracy and stability of the AGV. Among them, for the wheel drive system, the PID method is often used to complete the chassis motion control of the AGV. The straight and turning speeds of the current AGV are analyzed through path planning and navigation algorithms, and the theoretical running speed of the chassis drive wheel is solved using the inverse kinematics of the chassis model, and it is used as the input of the PID method to obtain the actual running speed of the AGV drive wheel.

[0004] In order to ensure that AGVs can operate safely and efficiently in complex environments and to ensure their adaptability to complex tasks and environments, it is necessary to accurately adjust the parameters of the wheel drive system. However, the parameter setting process of related technologies usually requires multiple experiments and adjustments, which is time-consuming. It cannot effectively respond to changes in complex or nonlinear dynamic environments. The PID controller cannot automatically adapt to different working environments and load changes, and usually requires manual parameter adjustment. This process is often cumbersome and consumes a lot of time and resources. Summary of the invention

[0005] The present application provides a PID control method for an AGV wheel drive system to solve the problem that the parameter setting process of the wheel drive system requires multiple experiments and adjustments, which is time-consuming, and cannot effectively cope with changes in complex or nonlinear dynamic environments. Manual adjustment of parameters is usually required, making the operation more cumbersome and consuming a lot of time and resources.

[0006] The present application provides a PID control method for an AGV wheel drive system, the method comprising:

[0007] Acquire first parameters of the AGV; the first parameters include chassis model parameters, structural design parameters, and servo drive system attribute parameters of the AGV;

[0008] Filtering the first parameter of the AGV, and correcting the erroneous data in the first parameter to obtain a second parameter;

[0009] A preset speed function is set based on a system error model of the AGV wheel drive system and a second parameter, and the AGV is operated according to the preset speed function and combined with the error model to obtain a drive gain coefficient, a friction coefficient, and an odometer gain coefficient;

[0010] Substituting the driving gain coefficient, friction coefficient and odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV wheel drive system, and solving the optimal solution through a preset optimization algorithm to obtain the optimal solution;

[0011] The optimal solution is verified to obtain an evaluation value. When the evaluation value is less than a preset threshold, the target parameter corresponding to the optimal solution is output; the target parameter includes a proportional coefficient K p , integration time constant K i and the differential time constant K d ;

[0012] The target parameters are substituted into the AGV wheel drive system, and the chassis motor servo drive device is controlled based on the control instructions to collaboratively complete the specified action.

[0013] In some possible implementations, the fitness function is:

[0014]

[0015] i is the transmission ratio of the chassis reduction mechanism, A and B are matrices, K is the target parameter, r is the wheel radius, and s is the safety factor.

[0016] In some possible implementations, substituting the drive gain coefficient, the friction coefficient, and the odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV wheel drive system includes:

[0017] is the proportional coefficient K in the adaptive PID control algorithm of the AGV wheel drive system p , integration time constant K i and the differential time constant K d Configure the corresponding particles;

[0018] The initial particle swarm is randomly generated according to the particle swarm selection scale S within the feasibility interval under the constraints;

[0019] Updating the position and velocity of the initial particle swarm based on the fitness function;

[0020] The global optimal position of the particle swarm is determined, and the optimal particle is output when a termination condition is met.

[0021] In some possible implementations, the constraint condition is:

[0022]

[0023] in,‖·‖ ∞ represents the infinite norm of a vector; τ i ∈{τ L ,τ R} represents the driving torque of the left and right wheels; n i ∈{n L ,n R} indicates the actual speed of the left and right wheels; K j ∈{K p ,K i ,K d}, To use the closed-loop regulator frequency response method, K p ,K i ,K d The optimization space expanded by parameter values.

[0024] In some possible implementations, updating the position and velocity of the initial particle swarm based on the fitness function includes:

[0025] After updating the initial particle group, determining the optimal particle position of each particle, and updating the global optimal position;

[0026] Update the position and velocity of the particle swarm and restrict each particle to be within the feasible range under the constraint conditions;

[0027] If the termination condition k>k is met max or F(g (k) )<F th , output the optimal particle;

[0028] Otherwise, the initial particle swarm is randomly generated according to the particle swarm selection scale S within the feasibility interval under the constraints.

[0029] In some possible implementations, after updating the initial particle group, determining the optimal particle position of each particle, and updating the global optimal position include:

[0030] The fitness value for each particle The fitness of the particle with the best historical position Compare and if the current position is better than the historical best position, update the historical best position;

[0031] Find the optimal particle position of the current particle swarm If it is better than the historical optimal position of the particle swarm Then it is updated to the global optimal position.

[0032] In some possible implementations, setting a preset speed function based on a system error model of the AGV wheel drive system and a second parameter includes:

[0033] Based on the system error model of the AGV wheel drive system and the second parameter estimation, and set the preset speed function And based on the initial PID control parameters K 0Run AGV;

[0034] The servo motor provides real-time feedback of the physical quantity parameters obtained by the AGV's own state sensor, and estimates its own system error and the environmental conditions of the AGV based on the slip condition of the AGV to determine the drive gain coefficient, friction coefficient, and odometer gain coefficient.

[0035] In some possible implementations, it is the proportional coefficient K in the adaptive PID control algorithm of the AGV wheel drive system p , integration time constant K i When the particle corresponding to the differential time constant Kd is configured, a single particle can be expressed as K h =[K ph K ih K dh ] T , select S for particle size, set the maximum number of iterations and the maximum search speed of particles to k max and v max , maximum inertia weight w max is 0.9, the minimum inertia weight w min is 0.4.

[0036] In some possible implementations, when the evaluation value is greater than or equal to a preset threshold, the second parameter is corrected, and the corrected second parameter is substituted into a preset optimization algorithm to find an optimal solution.

[0037] In some possible implementations, verifying the optimal solution to obtain an evaluation value includes:

[0038] Substitute the optimal solution into the evaluation function and output the evaluation value. The evaluation function is

[0039] If the evaluation value is greater than a preset threshold, the second parameter is corrected twice based on the evaluation value, and the corrected second parameter is substituted into a preset algorithm to find an optimal solution;

[0040] Otherwise, the target parameters corresponding to the optimal solution are output.

[0041] From the above content, it can be seen that the present application provides a PID control method for an AGV wheel drive system, the method comprising obtaining a first parameter of the AGV; filtering the first parameter of the AGV, correcting the erroneous data in the first parameter to obtain a second parameter; setting a preset speed function based on the system error model of the AGV wheel drive system and the second parameter, and making the AGV run according to the preset speed function and in combination with the error model to obtain a drive gain coefficient, a friction coefficient and an odometer gain coefficient; substituting the drive gain coefficient, the friction coefficient and the odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV, and solving the optimal solution through a preset optimization algorithm to obtain the optimal solution; verifying the optimal solution to obtain an evaluation value, and when the evaluation value is less than a preset threshold, outputting the target parameter corresponding to the optimal solution; substituting the target parameter into the AGV wheel drive system, and controlling the chassis motor servo drive device to collaboratively complete the specified action based on the control instruction. The present application improves the AGV's adaptability to various environments and chassis motion control performance, and can also avoid cumbersome parameter setting processes, and achieve high-precision and high-reliability control of the AGV. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0043] Figure 1 A schematic flow chart of a PID control method for an AGV wheel drive system provided in an embodiment of the present application;

[0044] Figure 2 A schematic diagram of a two-wheel differential chassis model provided in an embodiment of the present application;

[0045] Figure 3 Schematic diagram of the posture error model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The following embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following embodiments do not represent all implementations consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application as detailed in the claims.

[0047] AGV (Automated Guided Vehicle), as an unmanned vehicle, is becoming more and more widely used in industrial scenarios. The large-scale application of AGV has significantly improved the automation level of production lines and logistics efficiency, mainly including efficient material handling in various links such as warehousing, production, and distribution, reducing dependence on manual labor, thereby reducing labor costs and error rates. As AGV application scenarios continue to expand and become more complex, the demand for AGV control accuracy is also increasing. The requirements for high precision and high reliability in industrial environments have challenged traditional control methods.

[0048] The factors that affect the control accuracy of AGV mainly include the accuracy of the sensor itself, environmental conditions, navigation algorithms, the accuracy of the wheel drive system, and dynamic response capabilities. The accuracy of the sensor directly determines the accuracy of position, posture estimation, and obstacle detection, while environmental conditions (such as changes in light and uneven ground) may interfere with the sensor data, thereby affecting the positioning accuracy. The degree of optimization of the navigation algorithm affects the accuracy of path planning, and the accuracy and dynamic response capabilities of the wheel drive system are directly related to the motion control accuracy and stability of the AGV. Among them, for the wheel drive system, the PID method is often used to complete the chassis motion control of the AGV. The straight and turning speeds of the current AGV are analyzed through path planning and navigation algorithms, and the theoretical running speed of the chassis drive wheel is solved using the inverse kinematics of the chassis model, and it is used as the input of the PID method to obtain the actual running speed of the AGV drive wheel.

[0049] There are two methods for tuning PID control parameters: theoretical calculation method and engineering tuning method. The theoretical calculation method is based on the mathematical model of the system and determines the controller parameters by calculation, but these results usually need to be further adjusted in practical applications. Relatively speaking, the engineering tuning method relies on engineering experience and directly adjusts the parameters through experiments. It is simple to operate and widely used. Common engineering tuning methods include critical proportion method, reaction curve method and attenuation method. Regardless of which method is used, the final controller parameters need to be optimized through actual operation. This process usually requires a lot of human resources, including observing the operating status of the AGV and conducting multiple adjustment experiments to find the best PID parameters. Since each AGV has unique characteristics and the working environment (such as ground material and friction) is different, targeted PID parameter adjustment and setting are required.

[0050] In order to ensure that AGVs can operate safely and efficiently in complex environments and to ensure their adaptability to complex tasks and environments, it is necessary to accurately adjust the parameters of the wheel drive system. However, the parameter setting process of related technologies usually requires multiple experiments and adjustments, which is time-consuming. It cannot effectively respond to changes in complex or nonlinear dynamic environments. The PID controller cannot automatically adapt to different working environments and load changes, and usually requires manual parameter adjustment. This process is often cumbersome and consumes a lot of time and resources.

[0051] Based on this, the present application proposes a PID control method for an AGV wheel drive system. The optimized PID control method can not only improve the AGV's adaptability to various environments and chassis motion control performance, but also avoid the tedious parameter setting process, reduce human resource investment, and maximize its application value in industrial scenarios.

[0052] For the PID controller parameter selection problem of wheeled AGV motion control, due to the diversity of its application environment and the system errors existing in the installation and matching process of each AGV, the most effective approach is to tailor the PID control parameters for each AGV to optimize the PID control method of its wheeled chassis drive system. This process is undoubtedly quite cumbersome. However, the existing AGV adaptive PID parameter adjustment method does not perform well in dealing with complex and changing environments and system errors. The traditional adaptive PID parameter optimization method usually uses IAE (Integrated Absolute Error) integrated absolute error, ITAE (Integrated Time Absolute Error) integrated time absolute error, and ITSE (Integrated Time Square Error) integrated time square error as the objective function of the optimization method. These objective functions can only be calculated based on the statistical results of the actual operation of the AGV with given PID controller parameters for a period of time. Therefore, the usual practice is to establish an AGV simulation model to simulate the actual motion state and search for the optimal PID control parameters under this constraint based on modern optimization methods. The motion state simulated by simulation is usually different from the actual motion state. To this end, this application provides a PID control method for an AGV wheel drive system, combining theoretical analysis with the actual operating status of the AGV, and autonomously adjusting the K in the PID control method according to different usage environments. p , K i , K d Three controller parameters, proportional coefficient K p , integration time constant K i and the differential time constant K d .

[0053] In some embodiments, Figure 1 As shown, the present application provides a PID control method for an AGV wheel drive system, the method comprising:

[0054] Acquire first parameters of the AGV; the first parameters include chassis model parameters, structural design parameters, and servo drive system attribute parameters of the AGV;

[0055] Filtering the first parameter of the AGV, and correcting the erroneous data in the first parameter to obtain a second parameter;

[0056] A preset speed function is set based on a system error model of the AGV wheel drive system and a second parameter, and the AGV is operated according to the preset speed function and combined with the error model to obtain a drive gain coefficient, a friction coefficient, and an odometer gain coefficient;

[0057] Substituting the driving gain coefficient, friction coefficient and odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV wheel drive system, and solving the optimal solution through a preset optimization algorithm to obtain the optimal solution;

[0058] The optimal solution is verified to obtain an evaluation value. When the evaluation value is less than a preset threshold, the target parameter corresponding to the optimal solution is output; the target parameter includes a proportional coefficient K p , integration time constant K i and the differential time constant K d ;

[0059] The target parameters are substituted into the AGV wheel drive system, and the chassis motor servo drive device is controlled based on the control instructions to collaboratively complete the specified action.

[0060] Among them, the chassis model parameters of AGV include chassis type, size, wheelbase, and wheel diameter, which can be obtained through actual measurement. Structural design parameters include vehicle mass, moment of inertia around the center of mass, moment of inertia around the wheel, and transmission ratio. These parameters are determined by component design and selection. Servo drive system attribute parameters include the electrical properties of the servo drive system (position loop, speed loop), motor speed-torque characteristic curve, and these parameters are inherent attribute parameters of the servo drive system. Real-time acquisition and calculation of data obtained after filtering processing include wheel odometer data (this data is accumulated by combining motor encoder data with chassis kinematic model) and IMU data, laser radar point cloud data, QR code positioning data (these data are filtered and multi-sensor information fusion is used to estimate the actual position and actual motion state of AGV) and the real-time motor speed and real-time output torque fed back by the servo drive system. Environmental parameters, including ground slope, static friction coefficient, dynamic friction coefficient, wheel rolling friction coefficient and gain coefficient, can be estimated for some complex and changeable operating scenarios by running preset calibration functions (preset speed functions) combined with the real-time output torque and slip conditions of the motor.

[0061] In some embodiments, the fitness function is:

[0062]

[0063] i is the transmission ratio of the chassis reduction mechanism, A and B are matrices, K is the target parameter, r is the wheel radius, and s is the safety factor.

[0064] In some embodiments, substituting the drive gain coefficient, friction coefficient, and odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV wheel drive system includes:

[0065] is the proportional coefficient K in the adaptive PID control algorithm of the AGV wheel drive system p , integration time constant K i and the differential time constant K d Configure the corresponding particles;

[0066] The initial particle swarm is randomly generated according to the particle swarm selection scale S within the feasibility interval under the constraints;

[0067] Updating the position and velocity of the initial particle swarm based on the fitness function;

[0068] The global optimal position of the particle swarm is determined, and the optimal particle is output when a termination condition is met.

[0069] In some embodiments, the constraint condition is:

[0070]

[0071] in,‖·‖ ∞ represents the infinite norm of a vector; τ i ∈{τ L ,τ R} represents the driving torque of the left and right wheels; n i ∈{n L ,n R} indicates the actual speed of the left and right wheels; K j ∈{K p ,K i ,K d}, To use the closed-loop regulator frequency response method, K p ,K i ,K d The optimization space expanded by parameter values.

[0072] For any AGV chassis model, its motion can be decomposed into three directions of the global coordinate system: v x (along the X-axis), v y (along the Y-axis direction), ω (rotation along its own direction). These three parameters can describe all the actions of the AGV, that is, for any combination of three parameters, the steering angle and speed of each driving wheel can be analyzed through the AGV motion analysis method, so as to control the chassis motor servo drive system to collaboratively complete the specified action. The AGV mobile chassis motion analysis method includes forward kinematics solution and inverse kinematics solution. The forward kinematics solution calculates the speed and steering angle of each driving wheel based on the expected linear velocity and angular velocity in the global coordinate system. The inverse kinematics solution calculates the position change of the vehicle body in the global coordinate system based on the speed and steering angle of the driving wheel. Inverse kinematics analysis is to calculate the position change of the vehicle body in the global coordinate system based on the speed and steering angle of the driving wheel at the specified chassis driving speed v. r and radius R, solve the speed and angle of the driving wheel. Take the two-wheel differential chassis model as an example, Figure 2 shown.

[0073] The actual forward direction of the AGV is the positive direction of the x-axis in its own coordinate system. The angle between the actual forward direction of the AGV and the positive direction of the x-axis is θ. r .w R w L are the angular velocities of the right and left wheels respectively. Without considering slippage, the speed of the wheel center is equal to the linear speed of the wheel rotation, so v L =w L rv R =w R r, where v L 、v Rare the speeds of the left and right wheels respectively; r is the wheel radius. The inverse kinematics model of AGV in the plane global coordinate system can be expressed as:

[0074]

[0075] by v =[v L v R ] T Indicates the output left and right wheel speeds, r =[v r w r ] T Represents the input AGV overall driving speed and angular velocity, Represents the transformation matrix, and the inverse kinematics model of AGV is expressed as formula (2):

[0076] v =D r (2);

[0077] Usually, the AGV driving speed and angular velocity instructions are sent from the host computer to the servo control system controller. The theoretical linear speed of the two driving wheels is obtained by inverse kinematics, and then the actual running speed of each driving wheel is further obtained by the PID controller. The PID controller uses the proportional coefficient K p , integration time constant K i and the differential time constant K d The three controller parameters adjust the objective function characteristics. The PID controller controls the controlled object by adjusting the deviation e(t) between the predicted input value r(t) and the actual output value u(t). The deviation e(t) = r(t) - u(t) is formed by the system output value u(t) through system feedback. The PID control formula is as follows:

[0078]

[0079] For the motion control of the AGV chassis model, the host computer sends the expected linear velocity and angular velocity r(t) to the chassis controller, and the motion analysis module uses r (t) is used as the input of the PID controller and the actual real-time linear velocity and angular velocity output is obtained based on equation (3). Then, based on formula (2), the actual left and right wheel speeds are calculated as Finally, the actual speed of the servo motor is controlled as follows: Where i is the transmission ratio of the chassis reduction mechanism. For a given PID controller input r (t), if the control parameter K =[K p K i K d ] and output The relationship between is based on formula (3) as The actual speed of the motor can be expressed as:

[0080] n (t) = (2πri) -1 DΓ( K , r (t)) (4);

[0081] The force and torque on the wheels of the AGV during driving will affect the position and speed. At the same time, the rotational inertia of the AGV around its own center of mass, its own mass, the friction it experiences, and other factors will also affect the driving error. Therefore, it is necessary to analyze the AGV dynamics model and the relationship between the position, speed, and acceleration of the AGV. Based on the principle of torque balance and Newton's second law of motion, the following relationship exists for wheeled AGVs:

[0082]

[0083] Among them, I m is the moment of inertia of the AGV’s center of mass; F R is the driving force of the right wheel of AGV; F L is the driving force of the AGV left wheel; l is the distance from the AGV wheel to the center of mass; m represents the mass of the AGV. The following formula can be obtained from the wheel torque balance principle:

[0084]

[0085] Among them, I w is the moment of inertia around the robot wheel; c is the wheel rolling friction coefficient; k is the drive gain; τ L , τ R is the driving torque of the robot’s left and right wheels. If the real-time speed sent by the host computer to the mobile wheeled chassis is r (t), then the real-time driving torque required by the left and right wheels can be obtained from equations (2), (5) and (6): τ (t) = [τ L (t)τ R (t)] T for:

[0086]

[0087] In the formula, matrices A and B are:

[0088]

[0089] The torque-frequency characteristic curve of the servo motor driving the left and right wheels of the wheeled chassis can be expressed as τ=G(n), where n is the motor speed and τ is the torque that the motor can provide at this speed. Since the AGV chassis servo control system uses a PID controller to maximize the servo system response speed and reduce the steady-state error under the condition that the wheels do not slip, so as to reduce the AGV odometer error, the appropriate PID controller parameters are K =[K p K i K d ] T The selection can make full use of the driving torque provided by the servo motor. If the time function of the predetermined AGV running linear speed and angular speed is given r (t), taking the design safety factor as s (0<s<1), combining the above kinematic and dynamic analysis of AGV and the principle of PID controller, the PID control parameters for the wheeled chassis of AGV are proposed: K The fitness function of the optimization problem is defined as:

[0090]

[0091] Since the AGV should avoid slipping during operation, the driving force of the left and right wheels is required. F =[F L F R ] T Should be less than the maximum static friction between the wheel and the ground F m =μ s mgcosα, where μ s is the static friction coefficient, α is the ground slope, and the real-time driving torque required by the left and right wheels needs to be less than the maximum driving torque that the motor can provide, and considering that as the parameters K The optimization method with the expanded optimization space will make it more difficult to find the global optimal solution. Therefore, the constraints of the PID parameter optimization problem are as follows:

[0092]

[0093] in,‖·‖ ∞ represents the infinite norm of a vector; τ i ∈{τ L ,τ R} represents the driving torque of the left and right wheels; n i ∈{n L ,n R} indicates the actual speed of the left and right wheels; K j ∈{K p ,K i ,K d}, The ZN (Ziegler-Nichols, frequency response) method of the closed-loop regulator is used to set K p ,K i ,K d The parameter value expands a certain range of optimization space.

[0094] Even if the AGV does not slip during operation, the actual position and posture of the AGV may differ from the expected position and posture due to driving uncertainty, cumulative errors in the mileage calculation method, etc. Therefore, a simplified position and posture error model of the AGV is established. Figure 3 shown.

[0095] exist Figure 3 In the figure, the actual position of the AGV [x e y e θ e ] T and the expected pose [x r y r θ r ] T The error components along the x-axis, along the y-axis and at the angle in the local coordinate system xoy are e q =[x q y q θ q ] T , where θ q is the angle error component. Then according to Figure 3 The simplified model relationship between the actual posture and expected posture of the AGV can be obtained by:

[0096]

[0097] After the above formula is expanded and derivated, it can be written in matrix form:

[0098]

[0099] where v r 、w r The expected linear velocity and angular velocity of the AGV are the output values ​​of the PID controller. v e 、w e In this application, a multi-sensor information fusion positioning algorithm is used to approximately estimate the actual linear speed and angular speed of the AGV, using positioning algorithms other than wheel odometers, such as IMU (Inertial Measurement Unit), lidar, and visual QR code positioning methods.

[0100] In some embodiments, setting a preset speed function based on a system error model of the AGV wheel drive system and a second parameter includes:

[0101] Based on the system error model of the AGV wheel drive system and the second parameter estimation, and set the preset speed function And based on the initial PID control parameters K 0Run AGV;

[0102] The servo motor provides real-time feedback of the physical quantity parameters obtained by the AGV's own state sensor, and estimates its own system error and the environmental conditions of the AGV based on the slip condition of the AGV to determine the drive gain coefficient, friction coefficient, and odometer gain coefficient.

[0103] In some embodiments, the preset algorithm is one of a particle swarm algorithm, a genetic algorithm, an ant colony algorithm, a bee colony algorithm, a simulated degradation algorithm, and a neural network algorithm.

[0104] In some embodiments, updating the position and velocity of the initial particle swarm based on the fitness function includes:

[0105] After updating the initial particle group, determining the optimal particle position of each particle, and updating the global optimal position;

[0106] Update the position and velocity of the particle swarm and restrict each particle to be within the feasible range under the constraint conditions;

[0107] When the termination condition is met, the optimal particle is output.

[0108] In some embodiments, it is the proportional coefficient K in the adaptive PID control algorithm of the AGV wheel drive system p , integration time constant K i and the differential time constant K d When configuring the corresponding particles, a single particle can be represented as K h =[K ph K ih K dh ] T , select S for particle size, set the maximum number of iterations and the maximum search speed of particles to k max and v max , maximum inertia weight w max is 0.9, the minimum inertia weight w min is 0.4.

[0109] In some embodiments, after updating the initial particle group, determining the optimal particle position of each particle, and updating the global optimal position includes:

[0110] The fitness value for each particle The fitness of the particle with the best historical position Compare and if the current position is better than the historical best position, update the historical best position;

[0111] Find the optimal particle position of the current particle swarm If it is better than the historical optimal position of the particle swarm Then it is updated to the global optimal position.

[0112] In some embodiments, when the termination condition is met, outputting the optimal particle includes:

[0113] If the termination condition k>k is met max or F(g (k) )<F th , output the optimal particle;

[0114] Otherwise, the initial particle swarm is randomly generated according to the particle swarm selection scale S within the feasibility interval under the constraints.

[0115] This application uses particle swarm algorithm to find the optimal solution for PID controller. K Control parameters. Particle swarm algorithm is a swarm intelligence optimization algorithm based on a simplified biological social model. The algorithm originated from the study of bird flock predation behavior, and found that bird flocks capture more food by sharing information between each other. The core idea of ​​the particle swarm algorithm inspired by this is also information sharing between individuals and groups. The particle swarm algorithm can be described as: a group of n particles moves in a D-dimensional space, combining the best position experienced by each particle during its own movement with the best position experienced by all particles in the group, so as to determine its next movement state. The movement speed and position update of the particle are determined by equations (13) and (14).

[0116]

[0117] In the formula, is the moving speed of particle i in the kth step. is the best position experienced by particle i before k steps. 1 is the best position experienced by the entire particle swarm before the kth step. c1 and c2 are learning factors, which are generally 2. ξ and η are pseudo-random numbers in [0,1]. w is the inertia weight, which is generally a normal number. This application uses the linear transformation of the inertia weight to improve the search ability of the particle swarm algorithm, and takes w = w min +(w max -w min )×(1-k / k max ), where w max and w minare the maximum inertia weight and the minimum inertia weight, respectively, k max is the maximum number of iteration steps.

[0118] In some embodiments, when the evaluation value is greater than or equal to a preset threshold, the second parameter is corrected, and the corrected second parameter is substituted into a preset algorithm to obtain an optimal solution.

[0119] In some embodiments, verifying the optimal solution to obtain an evaluation value includes:

[0120] Substitute the optimal solution into the evaluation function and output the evaluation value. The evaluation function is

[0121] If the evaluation value is greater than a preset threshold, the second parameter is corrected twice based on the evaluation value, and the corrected second parameter is substituted into a preset algorithm to find an optimal solution;

[0122] Otherwise, the target parameters corresponding to the optimal solution are output.

[0123] The PID control method for the AGV wheel drive system provided in this application can not only realize accurate path planning, but also effectively deal with the obstacle avoidance problem in a dynamic environment, ensure the safe and efficient operation of the AGV in a complex environment, and improve the adaptability of the AGV in complex tasks and environments.

[0124] Example

[0125] Step 1: First, make a preliminary estimate of the AGV's own system error and the environmental conditions in which it is located, and set the operation calibration function (i.e., the preset speed function) determined by the AGV state on the AGV's host computer. And give the initial PID control parameters K 0 allows it to run according to the specified calibration function, and through the real-time feedback of the servo motor, the driving force distance combined with the AGV's own state sensors (laser positioning radar, IMU, wheel encoder, load quality feedback unit) and its own physical quantity parameters and based on the AGV's slip conditions, the host computer independently estimates its own system error model and the driving gain coefficient, friction coefficient and odometer gain coefficient in the environmental conditions adjustment model.

[0126] Among them, the initial parameters that need to be determined for the fitness function are:

[0127] 1. The chassis model parameters of AGV, including chassis type, size, wheelbase, and wheel diameter, can be obtained through actual measurement.

[0128] 2. Structural design parameters, including vehicle mass, moment of inertia about the center of mass, moment of inertia about the wheels, and transmission ratio. These parameters are determined by component design and selection.

[0129] 3. Servo drive system attribute parameters, including the electrical properties of the servo drive system (position loop, speed loop) and the motor speed-torque characteristic curve. These parameters are the inherent attribute parameters of the servo drive system.

[0130] 4. Real-time acquisition and calculation of filtered data, including wheel odometer data (the data is accumulated by combining motor encoder data with chassis kinematic model) and IMU data, lidar point cloud data, QR code positioning data (these data are filtered and multi-sensor information fusion is used to estimate the actual position and motion state of the AGV) and the real-time motor speed and real-time output torque fed back by the servo drive system.

[0131] 5. Environmental parameters, including ground slope, static friction coefficient, dynamic friction coefficient, wheel rolling friction coefficient and gain coefficient. For some complex and changeable operating scenarios, these parameters are described in this step. By letting the AGV run the preset calibration function r(t), combined with the real-time output torque of the motor and the AGV's own state based on the slip condition, the static friction coefficient, dynamic friction coefficient and other environmental parameters can be estimated.

[0132] Through the above steps, after directly or indirectly obtaining these environmental parameters and the physical parameters of the AGV itself, the fitness function determined by the adaptive PID algorithm parameters of the AGV wheeled chassis model of the present application is introduced, and the control parameters (K) of the PID control algorithm most suitable for the AGV in this environment can be obtained through the subsequent algorithm flow. p , K i , K d ).

[0133] Step 2: For the PID control parameter optimization problem of the AGV wheeled chassis, each particle in the particle swarm algorithm is selected as a PID control parameter, and a single particle can be expressed as K h =[K ph K ih K dh ] T , select S for particle size, set the maximum number of iterations and the maximum search speed of particles to k max and v max , comprehensively considering the algorithm's initial global search capability and later local search capability, select the maximum inertia weight w max is 0.9, the minimum inertia weight w min is 0.4.

[0134] Step 3: Randomly generate the initial particle swarm according to the particle swarm size S within the feasibility interval under the constraint condition s·t· And initialize the initial running speed of the particles

[0135] Step 4: Take the proposed optimization objective function (9) as the algorithm's fitness function F(K). According to the particle speed and position update formula (13) and formula (14), the position P of the particle group is respectively (k) With speed V (k) to update.

[0136] Step 5: Fitness value of each particle The fitness of its historical best position Compare and if the current position is better than the historical best position, update its historical best position.

[0137] Step 6: Find the optimal particle position of the current particle swarm If it is better than the historical optimal position of the particle swarm Then update it to the global optimal position.

[0138] Step 7: Update the position and velocity of the particle swarm according to equations (13) and (14), and limit each particle K h is within the feasible interval under the constraint s·t·.

[0139] Step 8: If the termination condition k>k is met max or F(g (k) )<F th The former represents the maximum number of iterative search steps, and the latter represents the fitness value of the acceptable optimal layout mode. The optimal particle is output. Otherwise return to step 3.

[0140] Step 9: Take the optimal solution found As the control parameter of the PID control method, let the AGV run the calibration function Run, at this time, the AGV posture error model (12) is used as the evaluation criterion to construct the evaluation function of the PID controller According to this formula, the evaluation value H is calculated. If H p -H<0(H p is the preset target evaluation value) and performs secondary correction on the AGV's own state parameters and operating environment condition parameters according to the size of the H value and jumps to step 2. If the preset target evaluation value is met, the parameter search of the adaptive PID control method is completed.

[0141] Among them, the usual optimization algorithm or adaptive PID algorithm will end at step 8. Based on actual field application experience, this application sets step 9 to perform secondary verification and revision of environmental parameters and AGV's own attribute parameters and optimize the fitness function after the modified parameters again. This can further ensure that the PID control parameters finally output by the algorithm have stronger adaptability to complex and changeable application environments.

[0142] It can be seen from the above embodiments that the present application provides a PID control method for an AGV wheel drive system, the method comprising obtaining a first parameter of the AGV; filtering the first parameter of the AGV, and correcting the erroneous data in the first parameter to obtain a second parameter; setting a preset speed function based on the system error model of the AGV wheel drive system and the second parameter, and making the AGV run according to the preset speed function and in combination with the error model to obtain a drive gain coefficient, a friction coefficient, and an odometer gain coefficient; substituting the drive gain coefficient, the friction coefficient, and the odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV, and solving the optimal solution through a preset optimization algorithm to obtain the optimal solution; verifying the optimal solution to obtain an evaluation value, and when the evaluation value is less than a preset threshold, outputting the target parameter corresponding to the optimal solution; substituting the target parameter into the AGV wheel drive system, and controlling the chassis motor servo drive device to collaboratively complete the specified action based on the control instruction. The present application improves the AGV's adaptability to various environments and chassis motion control performance, and can also avoid cumbersome parameter setting processes, and achieve high-precision and high-reliability control of the AGV.

[0143] Similar parts between the embodiments provided in this application can be referenced to each other. The specific implementation methods provided above are only a few examples under the general concept of this application and do not constitute a limitation on the protection scope of this application. For those skilled in the art, any other implementation methods expanded based on the scheme of this application without creative work belong to the protection scope of this application.

Claims

1. A PID control method for an AGV wheel drive system, characterized in that: The method comprises: Acquire first parameters of the AGV; the first parameters include chassis model parameters, structural design parameters, and servo drive system attribute parameters of the AGV; Filtering the first parameter of the AGV, and correcting the erroneous data in the first parameter to obtain a second parameter; A preset speed function is set based on a system error model of the AGV wheel drive system and a second parameter, and the AGV is operated according to the preset speed function and combined with the error model to obtain a drive gain coefficient, a friction coefficient, and an odometer gain coefficient; Substituting the driving gain coefficient, friction coefficient and odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV wheel drive system, and solving the optimal solution through a preset optimization algorithm to obtain the optimal solution; The optimal solution is verified to obtain an evaluation value. When the evaluation value is less than a preset threshold, the target parameter corresponding to the optimal solution is output; the target parameter includes a proportional coefficient K p , integration time constant K i and the differential time constant K d ; The target parameters are substituted into the AGV wheel drive system, and the chassis motor servo drive device is controlled based on the control instructions to collaboratively complete the specified action.

2. The PID control method for the AGV wheel drive system according to claim 1, characterized in that: The fitness function is: i is the transmission ratio of the chassis reduction mechanism, A and B are matrices, K is the target parameter, r is the wheel radius, and s is the safety factor.

3. The PID control method for the AGV wheel drive system according to claim 2, characterized in that: Substituting the drive gain coefficient, friction coefficient and odometer gain coefficient into the fitness function of the adaptive PID control algorithm of the AGV wheel drive system includes: is the proportional coefficient K in the adaptive PID control algorithm of the AGV wheel drive system p , integration time constant K i and the differential time constant K d Configure the corresponding particles; The initial particle swarm is randomly generated according to the particle swarm selection scale S within the feasibility interval under the constraints; Updating the position and velocity of the initial particle swarm based on the fitness function; The global optimal position of the particle swarm is determined, and the optimal particle is output when a termination condition is met.

4. The PID control method for the AGV wheel drive system according to claim 3, characterized in that: The constraints are: in,‖·‖ ∞ represents the infinite norm of a vector; τ i ∈{τ L ,τ R } represents the driving torque of the left and right wheels; n i ∈{n L ,n R } indicates the actual speed of the left and right wheels; K j ∈{K p ,K i ,K d }, To use the closed-loop regulator frequency response method, K p ,K i ,K d The optimization space expanded by parameter values.

5. The PID control method for the AGV wheel drive system according to claim 3, characterized in that: The updating of the position and velocity of the initial particle swarm based on the fitness function includes: After updating the initial particle group, determining the optimal particle position of each particle, and updating the global optimal position; Update the position and velocity of the particle swarm and restrict each particle to be within the feasible range under the constraint conditions; If the termination condition k>k is met max or F(g (k) )<F th , output the optimal particle; Otherwise, the initial particle swarm is randomly generated according to the particle swarm selection scale S within the feasibility interval under the constraints.

6. The PID control method for the AGV wheel drive system according to claim 5, characterized in that: After updating the initial particle swarm, determining the optimal particle position of each particle, and updating the global optimal position include: The fitness value for each particle The fitness of the particle with the best historical position Compare and if the current position is better than the historical best position, update the historical best position; Find the optimal particle position of the current particle swarm If it is better than the historical optimal position of the particle swarm Then it is updated to the global optimal position.

7. The PID control method for the AGV wheel drive system according to claim 1, characterized in that: The system error model based on the AGV wheel drive system and the second parameter setting preset speed function include: Based on the system error model of the AGV wheel drive system and the second parameter estimation, and set the preset speed function And run the AGV based on the initial PID control parameter K0; The servo motor provides real-time feedback of the physical quantity parameters obtained by the AGV's own state sensor, and estimates its own system error and the environmental conditions of the AGV based on the slip condition of the AGV to determine the drive gain coefficient, friction coefficient, and odometer gain coefficient.

8. The PID control method for the AGV wheel drive system according to claim 3, characterized in that: is the proportional coefficient K in the adaptive PID control algorithm of AGV p , integration time constant K i and the differential time constant K d When configuring the corresponding particles, a single particle can be represented as K h =[K ph K ih K dh ] T , select S for particle size, set the maximum number of iterations and the maximum search speed of particles to k max and v max , maximum inertia weight w max is 0.9, the minimum inertia weight w min is 0.

4.

9. The PID control method for the AGV wheel drive system according to claim 1, characterized in that: When the evaluation value is greater than or equal to a preset threshold, the second parameter is corrected, and the corrected second parameter is substituted into a preset optimization algorithm to obtain an optimal solution.

10. The PID control method for the AGV wheel drive system according to claim 1, characterized in that: Verifying the optimal solution to obtain an evaluation value includes: Substitute the optimal solution into the evaluation function and output the evaluation value. The evaluation function is If the evaluation value is greater than a preset threshold, the second parameter is corrected twice based on the evaluation value, and the corrected second parameter is substituted into a preset algorithm to find an optimal solution; Otherwise, the target parameters corresponding to the optimal solution are output.

Images