Load distribution intelligent optimization method for prefabricated part concrete distribution robot

Abstract

The invention discloses a load distribution intelligent optimization method for a prefabricated part concrete distribution robot, which is used in the field of robot intelligent control, and comprises the following steps: collecting structural parameters of the concrete distribution robot; determining an optimization variable vector of each joint of the concrete distribution robot in the movement process; establishing an objective function model of the concrete distribution robot; establishing a constraint condition model of the concrete distribution robot; according to the optimization variable vector, a niche grey wolf optimization algorithm is selected to solve the target function, and a joint motion planning scheme is obtained; the execution system receives and executes the joint control signal to drive the concrete distribution robot to move; feeding back and adjusting the target function model and the constraint condition model; training a state prediction model of the concrete distribution robot; and recording a fault reason. The method does not depend on complex parameter adjustment and control, and the global optimization capability is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot control, and more specifically, to an intelligent optimization method for load distribution of a precast concrete placing robot. Background Technology

[0002] With the development of the construction industry, traditional construction methods, due to their labor intensity and severe pollution, are no longer suitable for the needs of industrial development. Prefabricated buildings, because they enable industrialized production, are gradually becoming the mainstream development direction of the construction industry. Driven by policy support and market demand, my country's construction industry is undergoing a major transformation—specialized and industrialized industrialized construction production methods are replacing inefficient manual production methods.

[0003] The production line for precast concrete components is crucial for realizing industrialized construction. Concrete pouring is one of the most important production stages. The pouring process requires the coordinated operation of a concrete placing robot's placing and walking systems to evenly pour concrete into the mold area according to specified weight and thickness. Therefore, the coordination of the various systems within the placing robot directly affects product quality. If the walking speed is too fast while maintaining a constant spiral placing system rotation speed, it will result in insufficient and uneven pour thickness; if the walking speed is too slow, it will not only reduce production efficiency but also increase material waste.

[0004] However, in existing technologies, the models are overly idealized and do not account for dynamic changes in parameters. For example, due to factors such as machine wear and power fluctuations, the spiral rotation speed and walking speed of the fabric robot may fluctuate. This can cause the optimization results to deviate from reality, and the response mechanism for unexpected situations could be strengthened. When the production line malfunctions or raw material supply problems occur, the solution lacks means for rapid response and production recovery.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To overcome the above problems, this invention aims to propose an intelligent optimization method for load distribution of a precast concrete placing robot. The purpose is to address the issue that fluctuations in the robot's screw speed and travel speed due to factors such as machine wear and power supply fluctuations can cause the optimization results to deviate from actual conditions.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows: A method for intelligent optimization of load distribution for a precast concrete placing robot includes the following steps: S1. Collect the structural parameters of the concrete placing robot, establish its kinematic model, dynamic model and raw material supply chain model, and analyze and obtain the kinematic constraints and dynamic constraints during the robot's motion process. S2. Based on the kinematic model, determine the optimized variable vector for each joint of the concrete placing robot during its movement; S3. Based on the dynamic model, establish the objective function model of the concrete placing robot; S4. Based on the kinematic and dynamic constraints, establish the constraint model for the concrete placing robot. S5. The objective function model and constraint model are used as the objective function and constraint conditions for optimization. At the same time, based on the optimization variable vector, the niche gray wolf optimization algorithm is selected to solve the objective function and obtain a joint motion planning scheme that satisfies the constraint conditions. S6. Convert the joint motion planning scheme into joint control signals, and receive and execute the joint control signals through the execution system to drive the concrete placing robot to move. S7. By setting up sensors to collect the motion status of the concrete placing robot in real time, feedback and adjustment of the objective function model and constraint condition model are provided, while the raw material supply chain model is monitored to achieve closed-loop control. S8. Collect a large amount of historical operating data of the concrete placing robot, and train a state prediction model of the concrete placing robot based on the kinematic model and dynamic model of the concrete placing robot. S9. Set up redundant modules for the drive joints and sensors of the concrete placing robot, monitor the raw material supply status in real time based on the raw material supply chain model, and automatically activate the corresponding backup module or supplier when a fault is detected in the concrete placing robot or an abnormality in the raw material supply to ensure the stable operation of the execution system and record the cause of the fault. At the same time, expand the backup supplier resources to reduce the risk of single point of failure in the supply chain.

[0008] Optionally, the structural parameters of the concrete placing robot are collected, and its kinematic model, dynamic model, and raw material supply chain model are established. The kinematic and dynamic constraints during the robot's movement are analyzed, including the following steps: S11. Review the design drawings of the concrete placing robot and obtain the structural parameters of the design drawings. The structural parameters should include at least the number of joints, joint type, link length and mass distribution. S12. Perform static and dynamic measurements on the concrete placing robot on site to obtain the joint range of motion, the overall mass of the robot and the position of its center of mass. S13. Based on the robot's structural parameters, establish a kinematic model of the concrete placing robot, including forward and backward kinematic equations. S14. Based on the mass distribution and centroid position parameters, establish a dynamic model of the concrete placing robot, including the robot's dynamic equations; S15. Analyze the kinematic model and determine the position and velocity constraints during the robot's motion. S16. Analyze the dynamic model and determine the force and torque constraints during the robot's motion. S17. Review the company's raw material supply information, determine the suppliers, supply chain routes, and backup plans, and establish a status monitoring model for the raw material supply chain to describe the normal supply status. At the same time, obtain the motion constraints and dynamic constraints of the concrete placing robot.

[0009] Optionally, the objective function model and constraint model are used as the objective function and constraint conditions for optimization. Based on the optimization variable vector, and using the niche gray wolf optimization algorithm, the objective function is solved to obtain a joint motion planning scheme that satisfies the constraints. This includes the following steps: S51. Initialize the population for the niche gray wolf optimization algorithm. Generate the initial population based on the optimization variable vector. Each individual represents a set of joint motion parameters. S52. Set a fitness function to evaluate the merits of each individual gray wolf, calculate the fitness value of each individual gray wolf, and evaluate the energy consumption of the corresponding combination of movement parameters. S53. Select the three gray wolf individuals with the highest fitness values ​​(α, β, δ) as the current optimal solution. S54. Based on the positions of individual gray wolves α, β, and δ, and the algorithm's preset movement step size and distance adjustment parameters, calculate the new position of the non-optimal individual. S55. Determine whether the new position satisfies the kinematic and dynamic constraints. If it does, map the new position to a new combination of joint motion parameters. S56. Repeat steps S52 to S55 until the maximum number of iterations is reached, and then output the final joint motion planning scheme.

[0010] Optionally, setting a fitness function to evaluate the quality of each individual gray wolf, calculating the fitness value of each individual gray wolf, and evaluating the energy consumption of the corresponding combination of movement parameters includes the following steps: S521. Based on the robot motion planning problem, define the fitness function as the motion energy consumption corresponding to the combination of motion parameters, and ensure the computability of the fitness function. S522. Initialize the gray wolf pack, determine the combination of motion parameters represented by each individual gray wolf, and obtain accurate combination data; S523. Calculate and record the fitness value of each individual gray wolf, i.e., the corresponding energy consumption, through the defined fitness function; S524. Track changes in individual gray wolves in real time, and automatically recalculate and update the fitness value of each individual gray wolf. S525. Regularly check the rationality of the fitness function to ensure that it can effectively evaluate the quality of individual gray wolves.

[0011] Optionally, calculating the new position of a non-optimal individual based on the positions of gray wolf individuals α, β, and δ, and the algorithm's preset movement step size and distance adjustment parameters, includes the following steps: S541. Initialize the key parameters of the niche gray wolf optimization algorithm. The key parameters include the step size and the distance adjustment factor. S542. In each iteration, identify the three individuals with the highest fitness values ​​in the current population, and denote them as α, β, and δ, respectively. S543. For each non-optimal individual other than gray wolf individual α, gray wolf individual β, and gray wolf individual δ, calculate its distance from gray wolf individual α, gray wolf individual β, and gray wolf individual δ respectively. S544. Based on the movement step size, calculate the movement distance of the non-optimal individual relative to the optimal gray wolf individual α, gray wolf individual β, and gray wolf individual δ, and combine the three movement distances to calculate the new position of the gray wolf individual. S545. Based on the distance adjustment factor, fine-tune the calculation of the new position to avoid excessive clustering of individuals in the group.

[0012] Optionally, calculating the new location of an individual gray wolf includes the following steps: Calculate the distance D between the non-optimal gray wolf and the optimal gray wolf individual α. α The distance D between the non-optimal gray wolf and the optimal gray wolf individual β β The distance D between the non-optimal gray wolf and the optimal gray wolf individual δ δ ; According to distance D α Distance D β and distance D δ Calculate the new position of the non-optimal gray wolf individual after it moves according to the behaviors of the optimal gray wolf individual α, δ, and δ. Where, distance D α The calculation formula is: ; Distance D β The calculation formula is: ; Distance D δ The calculation formula is: ; In the formula, This represents the difference between non-optimal and optimal gray wolf individuals. The distance between them; Represents the optimal gray wolf individual and the optimal gray wolf individual The distance between them; Represents the optimal gray wolf individual and the optimal gray wolf individual The distance between them; , and These represent randomly generated distance adjustment factors; , and These represent the calculated movement step size; This indicates the current position of a non-optimal gray wolf individual.

[0013] Optionally, determining whether the new position satisfies kinematic and dynamic constraints, and if so, mapping the new position to a new combination of joint motion parameters, includes the following steps: S551. Calculate the rotational speed and acceleration motion parameters of each joint of the concrete placing robot based on the new position; S552. Substitute the motion parameters into the kinematic model, calculate the kinematic performance, and determine whether the predetermined kinematic constraints are met. S553. Substitute the motion parameters into the dynamic model, calculate the force and torque during the motion process, and determine whether the predetermined dynamic constraints are met. S554. Determine whether the motion parameters corresponding to the new position will cause the joint to exceed the rotational range of motion constraint. S555. Determine whether the results of steps S552-S554 all satisfy the corresponding constraints. S556. If all the corresponding constraints are met, then the new position is determined to satisfy the constraints. If the corresponding constraints are not met, the solution is set as infeasible.

[0014] Optionally, by setting sensors to collect the motion state of the concrete placing robot in real time, feeding back and adjusting the objective function model and constraint model, and simultaneously monitoring the raw material supply chain model, closed-loop control is achieved, including the following steps: S71. Set encoders at the joint positions of the concrete placing robot to obtain the real-time angle values ​​of each joint. S72. Install accelerometers on the robot body to obtain the real-time acceleration of each joint; S73. Input the data from the encoder and accelerometer into the control execution system; S74. Develop a closed-loop control program to compare sensor data with the output of the objective function model; S75. If there is a deviation, the control quantity is calculated by the PID controller. S76. Feedback the calculated control quantity to adjust the parameters in the objective function model; S77. Repeat steps S71-S76 until the error between the model output and the actual state is minimized; S78. Simultaneously, the constraint model is adjusted using sensor data to ensure it reflects the current constraints, thereby achieving coordinated optimization between the objective function model, the constraint model, and the execution system.

[0015] Optionally, collecting a large amount of historical operational data from the concrete placing robot, and training a state prediction model for the concrete placing robot based on its kinematic and dynamic models, includes the following steps: S81. Collect the angle, speed, acceleration and torque data of each joint during the operation of the concrete placing robot; S82. Collect energy consumption data and operation data of concrete placing robot when performing different movements, as well as operation data under different loads and material supply, and collect component usage time and wear data. S83. Extract the feature parameters of the kinematic and dynamic models, and use neural network algorithms to construct a state prediction model; S84. The state prediction model is trained using a large amount of collected historical operating data, and the state prediction model is continuously optimized through online learning. S85. Establish an error assessment mechanism between the state prediction model and the actual data, output the future state of the concrete robot, and assess the robot's health.

[0016] Optionally, an error assessment mechanism is established between the state prediction model and actual data to output the future state of the concrete robot and assess the robot's health, including the following steps: S851. On the test dataset of the state prediction model, calculate the prediction error between the model's prediction output and the actual data. S852. Set error thresholds for various predictions, including angle prediction error threshold and velocity prediction error threshold. S853. Statistically analyze the error distribution of different predictions and determine the normal error range; S854. During real-time prediction, determine whether the prediction error exceeds the set threshold. S855. If the prediction error exceeds the threshold, a health decline warning will be generated, and the wear condition and remaining life of key components and the energy consumption of robot movement will be assessed at the same time. S856. Based on the evaluation results, output the robot's health status for the current and future periods; If the health level is below the threshold, maintenance or replacement of parts is required.

[0017] Compared with the prior art, this application has the following beneficial effects: 1. This invention allows for the acquisition of relatively accurate structural parameters by consulting design drawings, laying the foundation for modeling. Static and dynamic tests can obtain dynamic performance parameters, better describing the motion process. Establishing a kinematic model allows for the analysis of kinematic performance and the acquisition of motion constraints. Establishing a dynamic model allows for the analysis of dynamic performance and the acquisition of dynamic constraints. Analyzing the model yields constraint conditions, providing constraints for subsequent motion planning. Establishing a supply chain model allows for the monitoring of raw material supply status and the acquisition of robot motion constraints, providing a basis for optimizing model establishment. The collaborative establishment of multiple models makes the description of the robot more comprehensive and accurate.

[0018] 2. The niche gray wolf optimization algorithm of this invention simulates the collective intelligent behavior of organisms in nature, follows the evolutionary laws of biological populations, and has the advantage of strong global search capability. The niche gray wolf optimization algorithm is simple in concept and easy to implement, has high computational efficiency, and is easy to combine with other niche gray wolf optimization algorithms. The niche gray wolf optimization algorithm inherently balances global search and local search, has a fast convergence speed, and achieves adaptive adjustment of individual positions through a position update mechanism. It can adaptively adjust the search space and avoid getting trapped in local optima. The niche gray wolf optimization algorithm introduces randomness, which can escape local optima. It does not rely on complex parameter adjustment and control, and can be widely used to solve practical engineering optimization problems and improve global optimization capability.

[0019] 3. The state prediction model of this invention can achieve high-precision state prediction by using machine learning algorithms. Incremental learning enables the prediction model to be dynamically updated. The error evaluation mechanism can verify the prediction. The prediction results can evaluate the robot's health. The prediction trend can be used for maintenance decisions. The modular design of the fault management module of this invention makes faulty parts easy to replace. The cascading use of backup suppliers improves the resilience of the supply chain. Machine learning can assist in fault prediction and identification. Digital twins can simulate and test fault conditions. The automatic processing mechanism reduces reliance on manual labor. The online knowledge base accumulates fault handling experience. The risk warning system realizes fault prevention, improves the ability to respond to supply chain emergencies, and enhances the system's fault tolerance and recovery capabilities. Attached Figure Description

[0020] The above-mentioned features, characteristics, and advantages of the present invention, as well as their implementation methods, will become clearer and more readily understood in conjunction with the following description of the embodiments, which are illustrated in detail with reference to the accompanying drawings. Schematic diagrams are shown here: Figure 1This is a flowchart of an intelligent optimization method for load distribution of a precast concrete placing robot according to an embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0022] According to an embodiment of the present invention, a method for intelligent optimization of load distribution for a precast concrete placing robot is provided.

[0023] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the intelligent optimization method for load distribution of a precast concrete placing robot according to an embodiment of the present invention includes the following steps: S1. Collect the structural parameters of the concrete placing robot, establish its kinematic model, dynamic model and raw material supply chain model, and analyze and obtain the kinematic constraints and dynamic constraints during the robot's movement.

[0024] Preferably, the process of collecting structural parameters of the concrete placing robot, establishing its kinematic model, dynamic model, and raw material supply chain model, and analyzing the kinematic and dynamic constraints during the robot's movement includes the following steps: S11. Review the design drawings of the concrete placing robot and obtain the structural parameters of the design drawings. The structural parameters should include at least the number of joints, joint type, link length and mass distribution. S12. Perform static and dynamic measurements on the concrete placing robot on site to obtain the joint range of motion, the overall mass of the robot and the position of its center of mass. S13. Based on the robot's structural parameters, establish a kinematic model of the concrete placing robot, including forward and backward kinematic equations. S14. Based on the mass distribution and centroid position parameters, establish a dynamic model of the concrete placing robot, including the robot's dynamic equations; S15. Analyze the kinematic model and determine the position and velocity constraints during the robot's motion. S16. Analyze the dynamic model and determine the force and torque constraints during the robot's motion. S17. Review the company's raw material supply information, determine the suppliers, supply chain routes, and backup plans, and establish a status monitoring model for the raw material supply chain to describe the normal supply status. At the same time, obtain the motion constraints and dynamic constraints of the concrete placing robot.

[0025] It should be explained that in the process of establishing the kinematic model, dynamic model and supply chain model of the concrete placing robot, it is necessary to consult the parameters of the design drawings, obtain parameters by static and dynamic measurements, establish the model using matrix and Lagrange methods, analyze and obtain the constraints, and use CAD, sensor and simulation technologies to assist in modeling, while verifying the correctness of the model.

[0026] S2. Based on the kinematic model, determine the optimal variable vector for each joint of the concrete placing robot during its movement.

[0027] It should be explained that when determining the optimization variables in the movement process of the concrete placing robot, the base position and posture variables can be added. Relevant feature variables can be extracted from historical data using machine learning, and the variables can be divided into mandatory and optional variables to form a variable library. The variables can be ranked through sensitivity analysis, the rationality of the variables can be evaluated on the simulation platform, and the variable selection can be dynamically updated using an incremental learning algorithm.

[0028] S3. Based on the dynamic model, establish the objective function model of the concrete placing robot.

[0029] It should be explained that when determining the objective function based on the established dynamic model of the concrete placing robot, a multi-objective candidate function library can be constructed. Deep reinforcement learning and incremental learning can be used to automatically learn and dynamically adjust the objective function, and its optimization effect can be tested on a simulation platform.

[0030] S4. Based on the kinematic and dynamic constraints, establish the constraint model for the concrete placing robot.

[0031] It should be explained that when establishing the constraint model of the concrete placing robot, more data can be collected to learn the implicit constraints using a data-driven approach, simulation can be used to explore the constraints under extreme conditions, a configurable constraint model library can be established, online learning algorithms can be used to dynamically update the model, and the optimization effect of different constraint models can be tested in a simulation environment to obtain a better constraint model.

[0032] S5. The objective function model and constraint model are used as the objective function and constraint conditions for optimization. At the same time, based on the optimization variable vector, the niche gray wolf optimization algorithm is selected to solve the objective function and obtain a joint motion planning scheme that satisfies the constraint conditions.

[0033] Preferably, the objective function model and constraint model are used as the objective function and constraint conditions for optimization. Simultaneously, based on the optimization variable vector, the niche gray wolf optimization algorithm is selected to solve the objective function and obtain a joint motion planning scheme that satisfies the constraints. This includes the following steps: S51. Initialize the population for the niche gray wolf optimization algorithm. Generate the initial population based on the optimization variable vector. Each individual represents a set of joint motion parameters. S52. Set a fitness function to evaluate the merits of each individual gray wolf, calculate the fitness value of each individual gray wolf, and evaluate the energy consumption of the corresponding combination of movement parameters. S53. Select the three gray wolf individuals with the highest fitness values ​​(α, β, δ) as the current optimal solution. S54. Based on the positions of individual gray wolves α, β, and δ, and the algorithm's preset movement step size and distance adjustment parameters, calculate the new position of the non-optimal individual. S55. Determine whether the new position satisfies the kinematic and dynamic constraints. If it does, map the new position to a new combination of joint motion parameters. S56. Repeat steps S52 to S55 until the maximum number of iterations is reached, and then output the final joint motion planning scheme.

[0034] Preferably, the following steps are taken to set a fitness function, evaluate the quality of each individual gray wolf, calculate the fitness value of each individual gray wolf, and evaluate the energy consumption of the corresponding combination of movement parameters: S521. Based on the robot motion planning problem, define the fitness function as the motion energy consumption corresponding to the combination of motion parameters, and ensure the computability of the fitness function. S522. Initialize the gray wolf pack, determine the combination of motion parameters represented by each individual gray wolf, and obtain accurate combination data; S523. Calculate and record the fitness value of each individual gray wolf, i.e., the corresponding energy consumption, through the defined fitness function; S524. Track changes in individual gray wolves in real time, and automatically recalculate and update the fitness value of each individual gray wolf. S525. Regularly check the rationality of the fitness function to ensure that it can effectively evaluate the quality of individual gray wolves.

[0035] Preferably, calculating the new position of a non-optimal individual based on the positions of gray wolf individuals α, β, and δ, and the algorithm's preset movement step size and distance adjustment parameters includes the following steps: S541. Initialize the key parameters of the niche gray wolf optimization algorithm. The key parameters include the step size and the distance adjustment factor. S542. In each iteration, identify the three individuals with the highest fitness values ​​in the current population, and denote them as α, β, and δ, respectively. S543. For each non-optimal individual other than gray wolf individual α, gray wolf individual β, and gray wolf individual δ, calculate its distance from gray wolf individual α, gray wolf individual β, and gray wolf individual δ respectively. S544. Based on the movement step size, calculate the movement distance of the non-optimal individual relative to the optimal gray wolf individual α, gray wolf individual β, and gray wolf individual δ, and combine the three movement distances to calculate the new position of the gray wolf individual. S545. Based on the distance adjustment factor, fine-tune the calculation of the new position to avoid excessive clustering of individuals in the group.

[0036] Preferably, calculating the new location of an individual gray wolf includes the following steps: Calculate the distance D between the non-optimal gray wolf and the optimal gray wolf individual α. α The distance D between the non-optimal gray wolf and the optimal gray wolf individual β β The distance D between the non-optimal gray wolf and the optimal gray wolf individual δ δ ; According to distance D α Distance D β and distance D δ Calculate the new position of the non-optimal gray wolf individual after it moves according to the behaviors of the optimal gray wolf individual α, δ, and δ. Where, distance D α The calculation formula is: ; Distance D β The calculation formula is: ; Distance D δ The calculation formula is: ; In the formula, This represents the difference between non-optimal and optimal gray wolf individuals. The distance between them; Represents the optimal gray wolf individual and the optimal gray wolf individual The distance between them; Represents the optimal gray wolf individual and the optimal gray wolf individual The distance between them; , and These represent randomly generated distance adjustment factors; , and These represent the calculated movement step size; This indicates the current position of a non-optimal gray wolf individual.

[0037] Preferably, determining whether the new position satisfies kinematic and dynamic constraints, and if so, mapping the new position to a new combination of joint motion parameters, includes the following steps: S551. Calculate the rotational speed and acceleration motion parameters of each joint of the concrete placing robot based on the new position; S552. Substitute the motion parameters into the kinematic model, calculate the kinematic performance, and determine whether the predetermined kinematic constraints are met. S553. Substitute the motion parameters into the dynamic model, calculate the force and torque during the motion process, and determine whether the predetermined dynamic constraints are met. S554. Determine whether the motion parameters corresponding to the new position will cause the joint to exceed the rotational range of motion constraint. S555. Determine whether the results of steps S552-S554 all satisfy the corresponding constraints. S556. If all the corresponding constraints are met, then the new position is determined to satisfy the constraints. If the corresponding constraints are not met, the solution is set as infeasible.

[0038] It should be explained that the fitness function typically considers energy consumption, but can also incorporate evaluation metrics such as motion smoothness and trajectory accuracy; when initializing the population, a uniform or Gaussian distribution can be used, or historical data can be employed; α, β, and δ can be chosen as global optimum, local optimum, or random individuals, with different choices affecting the convergence speed; the step size and distance adjustment parameters when calculating new positions affect the iteration process; when determining whether a new position satisfies the constraints, boundary condition handling also needs to be considered; the termination condition can also be set as the number of iterations and the fitness threshold; an adaptive parameter adjustment mechanism can be introduced to improve the algorithm's convergence; the algorithm framework is flexible and can be combined with other algorithms to improve global optimization capabilities; a matching rule base for algorithm parameter selection and optimization scenarios can be established.

[0039] The extended explanation is as follows: The α wolf corresponds to the optimal combination of joint motion parameters, which is the globally optimal solution throughout the optimization process. The β and δ wolves correspond to the second-best and third-best combinations of joint motion parameters, respectively; they can be considered locally optimal solutions, assisting the α wolf in approaching the optimal goal. The remaining ω wolves correspond to non-optimal combinations of joint motion parameters; they adjust their own combination parameters based on the positions of the α, β, and δ wolves to approximate the optimal solution.

[0040] Analogously, in this method, α wolf corresponds to the optimal joint motion planning scheme obtained through optimization algorithms. β wolf and δ wolf correspond to the second and third best planning schemes, respectively. The remaining ω wolves correspond to the non-optimal planning schemes after iterative updates. Together, they form the initial population, and by iteratively approximating the optimal solution α wolf, the search for and solution of the optimal planning scheme is achieved.

[0041] The GreyWolf Optimizer is a natural heuristic optimization algorithm that simulates the predation behavior of grey wolves. Its basic idea is as follows: The algorithm assumes that the wolf pack is divided into ranks: α (alpha wolf), β (secondary wolves), δ (ordinary wolves), and ω (candidate wolves). The alpha wolf is the leader of the pack and needs to determine the hunting location; β and δ assist ω, while the remaining candidate wolves ω follow their actions. During the hunt, α, β, and δ surround the prey, and the remaining candidate wolves ω adjust their positions according to their positions. Gradually approaching the prey, the wolf pack completes the hunt when the prey stops moving. Analogizing wolf pack behavior to an optimization process: Initialize the group, each wolf corresponds to a possible solution, design a fitness function to evaluate the quality of each wolf (solution), α, β, and δ correspond to the three wolves with the highest fitness, and ω is the remaining wolf. ω adjusts its position according to the positions of α, β, and δ to approach the optimal solution, iterating this process until the termination condition is met.

[0042] S6. Convert the joint motion planning scheme into joint control signals, and receive and execute the joint control signals through the execution system to drive the concrete placing robot to move.

[0043] It needs to be explained that a mapping model between the planning scheme and the control signal is established to directly generate the voltage / current signals required for control execution. A smoothing filter is added to smooth the converted control signal to reduce execution oscillations. A dynamic model of the execution system is constructed, and the effectiveness of the control signal is tested through simulation. By adding an observer to the execution system, the actual rotational speed and torque of each joint are estimated to achieve closed-loop control. For different execution systems, an adaptive conversion module for the control signal is constructed, and a monitoring module is set up to monitor motion stability in real time during execution and respond quickly when necessary. A digital twin of the execution system is constructed, and the simulation results are used to verify the correctness of the actual execution.

[0044] S7. By setting up sensors to collect the motion status of the concrete placing robot in real time, feedback and adjustment of the objective function model and constraint condition model are provided, while the raw material supply chain model is monitored to achieve closed-loop control.

[0045] Preferably, closed-loop control is achieved by setting sensors to collect the motion state of the concrete placing robot in real time, feeding back and adjusting the objective function model and constraint model, and monitoring the raw material supply chain model, including the following steps: S71. Set encoders at the joint positions of the concrete placing robot to obtain the real-time angle values ​​of each joint. S72. Install accelerometers on the robot body to obtain the real-time acceleration of each joint; S73. Input the data from the encoder and accelerometer into the control execution system; S74. Develop a closed-loop control program to compare sensor data with the output of the objective function model; S75. If there is a deviation, the control quantity is calculated by the PID controller. S76. Feedback the calculated control quantity to adjust the parameters in the objective function model; S77. Repeat steps S71-S76 until the error between the model output and the actual state is minimized; S78. Simultaneously, the constraint model is adjusted using sensor data to ensure it reflects the current constraints, thereby achieving coordinated optimization between the objective function model, the constraint model, and the execution system.

[0046] It should be explained that encoders typically use photoelectric encoders, which can acquire joint angle and rotation speed information with high precision; accelerometers generally use MEMS technology to obtain joint acceleration by detecting inertial forces; PID controllers adjust the control input through proportional, integral, and derivative actions to make the system output reach the target; the objective function model usually includes the robot's dynamic model, and the actual motion is approximated by adjusting the model parameters; the constraints reflected in the constraint model can come from joint physical limitations, environmental limitations, etc.; closed-loop control can improve the robustness of control and adapt to model errors through feedback adjustment; compared with open-loop control, closed-loop control can achieve adaptive control and has a certain ability to adapt to abnormal situations; the effectiveness of closed-loop control depends on factors such as the real-time performance of feedback, controller design, and model accuracy.

[0047] S8. Collect a large amount of historical operating data of the concrete placing robot, and train a state prediction model for the concrete placing robot based on the kinematic and dynamic models of the concrete placing robot.

[0048] Preferably, collecting a large amount of historical operational data of the concrete placing robot and training a state prediction model for the concrete placing robot based on its kinematic and dynamic models includes the following steps: S81. Collect the angle, speed, acceleration and torque data of each joint during the operation of the concrete placing robot; S82. Collect energy consumption data and operation data of concrete placing robot when performing different movements, as well as operation data under different loads and material supply, and collect component usage time and wear data. S83. Extract the feature parameters of the kinematic and dynamic models, and use neural network algorithms to construct a state prediction model; S84. The state prediction model is trained using a large amount of collected historical operating data, and the state prediction model is continuously optimized through online learning. S85. Establish an error assessment mechanism between the state prediction model and the actual data, output the future state of the concrete robot, and assess the robot's health.

[0049] Preferably, establishing an error assessment mechanism between the state prediction model and actual data, outputting the future state of the concrete robot, and assessing the robot's health includes the following steps: S851. On the test dataset of the state prediction model, calculate the prediction error between the model's prediction output and the actual data. S852. Set error thresholds for various predictions, including angle prediction error threshold and velocity prediction error threshold. S853. Statistically analyze the error distribution of different predictions and determine the normal error range; S854. During real-time prediction, determine whether the prediction error exceeds the set threshold. S855. If the prediction error exceeds the threshold, a health decline warning will be generated, and the wear condition and remaining life of key components and the energy consumption of robot movement will be assessed at the same time. S856. Based on the evaluation results, output the robot's health status for the current and future periods; If the health level is below the threshold, maintenance or replacement of parts is required.

[0050] It should be explained that neural network algorithms such as RNN (Recurrent Neural Network) and LSTM (Long Short-Term Memory Network) can be used for time-series data modeling and state prediction. Online learning methods can use incremental training and transfer learning to update models. Error assessment mechanisms can use indicators such as MSE (Mean Squared Error), MAE (Mean Absolute Error), and MAPE (Mean Absolute Percentage Error) to quantify prediction deviations. Setting reasonable thresholds is crucial for providing early warnings of prediction results. Component wear can be assessed using wear models or life estimation methods. Energy consumption assessment requires the dynamic model to calculate motor output power. Health assessment can use a cumulative scoring method, weighted according to various indicators. The health assessment indicator system will differ for different types of robots. The trend of health changes can also serve as a basis for maintenance decisions. Wear models and life estimation can assess the health status of key robot components. The motor power calculated by the dynamic model can reflect the robot's energy consumption. A weighted scoring method can assess the overall health of the robot. The health assessment system will differ for different robots. Assessing the trend of health changes can guide the maintenance and upkeep of the robot.

[0051] S9. Set up redundant modules for the drive joints and sensors of the concrete placing robot, monitor the raw material supply status in real time based on the raw material supply chain model, and automatically activate the corresponding backup module or supplier when a fault is detected in the concrete placing robot or an abnormality in the raw material supply to ensure the stable operation of the execution system and record the cause of the fault. At the same time, expand the backup supplier resources to reduce the risk of single point of failure in the supply chain.

[0052] It needs to be explained that the modular design allows for the rapid replacement of faulty components, rather than entire modules. A cascaded multi-level backup supplier system is implemented, with different levels activated based on the fault situation. Machine learning algorithms are used to analyze historical fault causes, establishing fault diagnosis and prediction models. A digital twin system is built to simulate various fault scenarios in the digital space. An automatic fault identification and handling mechanism is developed to reduce reliance on manual analysis. An online fault knowledge base is established to collect handling experience and assist manual analysis. A risk warning system is set up to take proactive measures when the risk of a single point of failure increases. By simulating different supply chain emergencies, responsiveness is improved.

[0053] In summary, by utilizing the above-mentioned technical solutions of this invention, relatively accurate structural parameters can be obtained by consulting design drawings, laying the foundation for modeling. Static and dynamic tests can obtain dynamic performance parameters, better describing the motion process. Establishing a kinematic model allows for the analysis of kinematic performance and the acquisition of motion constraints. Establishing a dynamic model allows for the analysis of dynamic performance and the acquisition of dynamic constraints. Analyzing the model yields constraint conditions, providing constraints for subsequent motion planning. Establishing a supply chain model allows for the monitoring of raw material supply status and the acquisition of robot motion constraints, providing a basis for optimizing the model. The collaborative establishment of multiple models makes the description of the robot more comprehensive and accurate. The niche gray wolf optimization algorithm of this invention simulates the collective intelligent behavior of natural populations, follows the evolutionary laws of biological populations, and has the advantage of strong global search capabilities. The niche gray wolf optimization algorithm is simple in concept and easy to implement, has high computational efficiency, and is easy to combine with other niche gray wolf optimization algorithms. The niche gray wolf optimization algorithm inherently balances global search and local search, has a fast convergence speed, and achieves this through position... The update mechanism enables adaptive adjustment of individual positions, allowing for adaptive adjustment of the search space and avoiding getting trapped in local optima. The niche gray wolf optimization algorithm introduces randomness, enabling it to escape local optima without relying on complex parameter adjustments and control. It can be widely used to solve practical engineering optimization problems and improve global optimization capabilities. The state prediction model of this invention can achieve high-precision state prediction by using machine learning algorithms. Incremental learning enables the prediction model to be dynamically updated. The error evaluation mechanism can verify the predictions. The prediction results can assess the robot's health, and the predicted trends can be used for maintenance decisions. The modular design of the fault management module of this invention makes faulty components easy to replace. The cascading use of backup suppliers improves supply chain resilience. Machine learning can assist in fault prediction and identification. Digital twins can simulate and test fault conditions. The automatic processing mechanism reduces reliance on manual labor. The online knowledge base accumulates fault handling experience. The risk warning system enables fault prevention, improves the ability to respond to supply chain emergencies, and enhances the system's fault tolerance and recovery capabilities.

[0054] Although the present invention has been disclosed above with reference to preferred embodiments, the embodiments are merely examples for illustrative purposes and are not intended to limit the present invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention should be determined by the claims.

Claims

1. A method for intelligent optimization of load distribution in a precast concrete placing robot, characterized in that, This intelligent load allocation optimization method includes the following steps: S1. Collect the structural parameters of the concrete placing robot, establish its kinematic model, dynamic model and raw material supply chain model, and analyze and obtain the kinematic constraints and dynamic constraints during the robot's motion process. S2. Based on the kinematic model, determine the optimized variable vector for each joint of the concrete placing robot during its movement; S3. Based on the dynamic model, establish the objective function model of the concrete placing robot; S4. Based on the kinematic and dynamic constraints, establish the constraint model for the concrete placing robot. S5. The objective function model and the constraint model are used as the objective function and constraint conditions for optimization. At the same time, based on the optimization variable vector, the niche gray wolf optimization algorithm is selected to solve the objective function and obtain a joint motion planning scheme that satisfies the constraint conditions. S6. Convert the joint motion planning scheme into joint control signals, and receive and execute the joint control signals through the execution system to drive the concrete placing robot to move. S7. By setting sensors to collect the motion state of the concrete placing robot in real time, the objective function model and the constraint condition model are fed back and adjusted, and the raw material supply chain model is monitored to achieve closed-loop control. S8. Collect a large amount of historical operating data of the concrete placing robot, and train a state prediction model of the concrete placing robot based on the kinematic model and dynamic model of the concrete placing robot. S9. Set up redundant modules for the drive joints and sensors of the concrete placing robot, monitor the raw material supply status in real time based on the raw material supply chain model, and automatically activate the corresponding backup module or supplier when a fault is detected in the concrete placing robot or an abnormality in the raw material supply to ensure the stable operation of the execution system and record the cause of the fault. At the same time, expand the backup supplier resources to reduce the risk of single point of failure in the supply chain.

2. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 1, characterized in that, The process of collecting structural parameters of the concrete placing robot, establishing its kinematic model, dynamic model, and raw material supply chain model, and analyzing the kinematic and dynamic constraints during the robot's movement includes the following steps: S11. Review the design drawings of the concrete placing robot and obtain the structural parameters of the design drawings. The structural parameters include at least the number of joints, joint type, link length and mass distribution. S12. Perform static and dynamic measurements on the concrete placing robot on site to obtain the joint range of motion, the overall mass of the robot and the position of its center of mass. S13. Based on the robot's structural parameters, establish a kinematic model of the concrete placing robot, including forward and backward kinematic equations. S14. Based on the mass distribution and centroid position parameters, establish a dynamic model of the concrete placing robot, including the robot's dynamic equations; S15. Analyze the kinematic model and determine the position and velocity constraints during the robot's motion. S16. Analyze the dynamic model and determine the force and torque constraints during the robot's motion. S17. Review the company's raw material supply information, determine the suppliers, supply chain routes, and backup plans, and establish a status monitoring model for the raw material supply chain to describe the normal supply status. At the same time, obtain the motion constraints and dynamic constraints of the concrete placing robot.

3. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 1, characterized in that, The process of using the objective function model and the constraint model as the objective function and constraint conditions for optimization, and simultaneously using the optimization variable vector and the niche gray wolf optimization algorithm to solve the objective function to obtain a joint motion planning scheme that satisfies the constraint conditions includes the following steps: S51. Initialize the population for the niche gray wolf optimization algorithm. Generate the initial population based on the optimization variable vector. Each individual represents a set of joint motion parameters. S52. Set a fitness function to evaluate the merits of each individual gray wolf, calculate the fitness value of each individual gray wolf, and evaluate the energy consumption of the corresponding combination of movement parameters. S53. Select the three gray wolf individuals with the highest fitness values ​​(α, β, δ) as the current optimal solution. S54. Based on the positions of individual gray wolves α, β, and δ, and the algorithm's preset movement step size and distance adjustment parameters, calculate the new position of the non-optimal individual. S55. Determine whether the new position satisfies the kinematic and dynamic constraints. If it does, map the new position to a new combination of joint motion parameters. S56. Repeat steps S52 to S55 until the maximum number of iterations is reached, and then output the final joint motion planning scheme.

4. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 3, characterized in that, The process of setting a fitness function to evaluate the quality of each individual gray wolf, calculating the fitness value of each individual gray wolf, and evaluating the energy consumption of the corresponding combination of movement parameters includes the following steps: S521. Based on the robot motion planning problem, define the fitness function as the motion energy consumption corresponding to the combination of motion parameters, and ensure the computability of the fitness function. S522. Initialize the gray wolf pack, determine the combination of motion parameters represented by each individual gray wolf, and obtain accurate combination data; S523. Calculate and record the fitness value of each individual gray wolf, i.e., the corresponding energy consumption, through the defined fitness function; S524. Track changes in individual gray wolves in real time, and automatically recalculate and update the fitness value of each individual gray wolf. S525. Regularly check the rationality of the fitness function to ensure that it can effectively evaluate the quality of individual gray wolves.

5. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 3, characterized in that, The step of calculating the new position of a non-optimal individual based on the positions of individual gray wolves α, β, and δ, and the algorithm's preset movement step size and distance adjustment parameters, includes the following steps: S541. Initialize the key parameters of the niche gray wolf optimization algorithm, including the step size and distance adjustment factor; S542. In each iteration, identify the three individuals with the highest fitness values ​​in the current population, and denote them as α, β, and δ, respectively. S543. For each non-optimal individual other than gray wolf individual α, gray wolf individual β, and gray wolf individual δ, calculate its distance from gray wolf individual α, gray wolf individual β, and gray wolf individual δ respectively. S544. Based on the movement step size, calculate the movement distance of the non-optimal individual relative to the optimal gray wolf individual α, gray wolf individual β, and gray wolf individual δ, and combine the three movement distances to calculate the new position of the gray wolf individual. S545. Based on the distance adjustment factor, fine-tune the calculation of the new position to avoid excessive clustering of individuals in the group.

6. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 5, characterized in that, The calculation of the new location of the individual gray wolf includes the following steps: Calculate the distance D between the non-optimal gray wolf and the optimal gray wolf individual α. α The distance D between the non-optimal gray wolf and the optimal gray wolf individual β β The distance D between the non-optimal gray wolf and the optimal gray wolf individual δ δ ; According to distance D α Distance D β and distance D δ Calculate the new position of the non-optimal gray wolf individual after it moves according to the behaviors of the optimal gray wolf individual α, δ, and δ. Wherein, the distance D α The calculation formula is: ； The distance D β The calculation formula is: ； The distance D δ The calculation formula is: ； In the formula, This represents the difference between non-optimal and optimal gray wolf individuals. The distance between them; Represents the optimal gray wolf individual and the optimal gray wolf individual The distance between them; Represents the optimal gray wolf individual and the optimal gray wolf individual The distance between them; , and These represent randomly generated distance adjustment factors; , and These represent the calculated movement step size; This indicates the current position of a non-optimal gray wolf individual.

7. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 6, characterized in that, The process of determining whether the new position satisfies kinematic and dynamic constraints, and mapping the new position to a new combination of joint motion parameters if it does, includes the following steps: S551. Calculate the rotational speed and acceleration motion parameters of each joint of the concrete placing robot based on the new position; S552. Substitute the motion parameters into the kinematic model, calculate the kinematic performance, and determine whether the predetermined kinematic constraints are met. S553. Substitute the motion parameters into the dynamic model, calculate the force and torque during the motion process, and determine whether the predetermined dynamic constraints are met. S554. Determine whether the motion parameters corresponding to the new position will cause the joint to exceed the rotational range of motion constraint. S555. Determine whether the results of steps S552-S554 all satisfy the corresponding constraints. S556. If all the corresponding constraints are met, then the new position is determined to satisfy the constraints. If the corresponding constraints are not met, the solution is set as infeasible.

8. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 1, characterized in that, The process of using sensors to collect real-time data on the motion state of the concrete placing robot, feeding back and adjusting the objective function model and the constraint model, and simultaneously monitoring the raw material supply chain model to achieve closed-loop control includes the following steps: S71. Set encoders at the joint positions of the concrete placing robot to obtain the real-time angle values ​​of each joint. S72. Install accelerometers on the robot body to obtain the real-time acceleration of each joint; S73. Input the data from the encoder and accelerometer into the control execution system; S74. Develop a closed-loop control program to compare sensor data with the output of the objective function model; S75. If there is a deviation, the control quantity is calculated by the PID controller. S76. Feedback the calculated control quantity to adjust the parameters in the objective function model; S77. Repeat steps S71-S76 until the error between the model output and the actual state is minimized; S78. Simultaneously, the constraint model is adjusted using sensor data to ensure it reflects the current constraints, thereby achieving coordinated optimization between the objective function model, the constraint model, and the execution system.

9. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 8, characterized in that, The process of collecting a large amount of historical operational data from the concrete placing robot and training a state prediction model for the robot based on its kinematic and dynamic models includes the following steps: S81. Collect the angle, speed, acceleration and torque data of each joint during the operation of the concrete placing robot; S82. Collect energy consumption data and operation data of concrete placing robot when performing different movements, as well as operation data under different loads and material supply, and collect component usage time and wear data. S83. Extract the feature parameters of the kinematic model and the dynamic model, and construct a state prediction model using a neural network algorithm; S84. The state prediction model is trained using a large amount of collected historical operating data, and the state prediction model is continuously optimized through online learning. S85. Establish an error assessment mechanism between the state prediction model and the actual data, output the future state of the concrete robot, and assess the robot's health.

10. The intelligent optimization method for load distribution of a precast concrete placing robot according to claim 9, characterized in that, The establishment of an error assessment mechanism between the state prediction model and actual data, which outputs the future state of the concrete robot and assesses the robot's health, includes the following steps: S851. On the test dataset of the state prediction model, calculate the prediction error between the model's prediction output and the actual data. S852. Set error thresholds for various predictions, including angle prediction error threshold and velocity prediction error threshold. S853. Statistically analyze the error distribution of different predictions and determine the normal error range; S854. During real-time prediction, determine whether the prediction error exceeds the set threshold. S855. If the prediction error exceeds the threshold, a health decline warning will be generated, and the wear condition and remaining life of key components and the energy consumption of robot movement will be assessed at the same time. S856. Based on the evaluation results, output the robot's health status for the current and future periods; If the health level is below the threshold, maintenance or replacement of parts is required.

Images