Multi-target processing time optimization method for industrial robot

Abstract

An industrial robot multi-target processing time optimization method comprises the following steps: S1, constructing a path optimization mathematical model based on total task execution time, and generating a path sequence; s2, generating an initial population of the path sequence by using a Logistic chaotic mapping method; generating a reverse symmetric solution of the initial population by using a dynamic reverse learning strategy, and adding the reverse symmetric solution into the initial population; s3, grouping path points of the path sequence in the initial population by inserting bridging points; generating a preliminary optimization path; and S4, further optimizing the preliminarily optimized path through a dynamic variation neighborhood search strategy to generate a better path. The method aims at solving the problems that an existing path planning technology is insufficient in real-time performance in a complex production environment, is prone to falling into local optimum and is limited in multi-target optimization capacity, and therefore the execution efficiency and path planning quality of an industrial robot in a multi-target task are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial robots, and particularly relates to an industrial robot multi-target processing time optimization method and application thereof. BACKGROUND

[0002] With the rapid development of intelligent manufacturing, intelligent industrial robots play an increasingly important role in modern manufacturing and are widely used in automobile manufacturing, electronic production, metal processing and other fields. In order to improve the versatility and work efficiency of robots, robots are usually equipped with multiple tools of different specifications. The tools are fixed at the end of the robot at a certain distance, and a single tool can be adapted to multiple specifications of targets. When performing tasks such as carrying that require the use of tools, the motion path of the robot is affected by the combination of multiple tools and multiple targets to be processed. Selecting a good combination will greatly improve the processing efficiency of the task. Industrial robots often face target task scheduling problems in this case.

[0003] Traditional path planning methods mostly rely on preset paths or manually designed path rules. These methods have problems such as local optimal solution, low computational efficiency, lack of real-time adaptability, etc. when dealing with complex and multi-target tasks. Therefore, how to optimize the robot path in multi-target tasks, reduce the total task execution time, and improve the energy efficiency and path smoothness is an important technical challenge currently faced. SUMMARY

[0004] The present application proposes an industrial robot multi-target processing time optimization method, aiming to solve the problems of real-time deficiency, easy to fall into local optimum and limited multi-target optimization capability of existing path planning technology in complex production environment, so as to improve the execution efficiency and path planning quality of industrial robots in multi-target tasks.

[0005] To achieve the above purpose, the technical scheme adopted by the present application is:

[0006] An industrial robot multi-target processing time optimization method, comprising the following steps:

[0007] S1, constructing a path optimization mathematical model based on total task execution time to generate a path sequence;

[0008] S2, generating an initial population of the path sequence by using a Logistic chaotic mapping method; generating a reverse symmetric solution of the initial population by using a dynamic inverse learning strategy, and adding it to the initial population;

[0009] S3, grouping the path points of the path sequence in the initial population by inserting a bridge point; generating a preliminary optimized path;

[0010] S4, further optimizing the preliminary optimized path through a dynamic variant neighborhood search strategy to generate a more optimal path.

[0011] As preferred, the step S1 comprises:

[0012] S11, task initialization phase;

[0013] Suppose the task point set is C={c1,c2,…,c n}, and the tool set is T={t1,t2,…,t m};

[0014] S12, constraint processing;

[0015] S13, generating the target function of the total task time:

[0016]

[0017] Wherein:

[0018] p is a single task point in the element robot path sequence of the robot path sequence PA; v represents the kinematics parameter set (such as speed, acceleration, etc.) of the axis, which is affected by the robot constraint condition; b is the motion connection mode setting parameter set, which is affected by the robot constraint condition; f(p,v,b) represents the execution time calculation function of executing a single task point once, which contains the influence of path distance and motion constraint.

[0019] As preferred, the constraint conditions in the constraint processing include:

[0020] The execution order of the robot task points must satisfy the tool compatibility:

[0021]

[0022] Wherein, f t (c i ,c j ) is the compatibility function of the tool t to the task points c i and c j , and a non-zero value indicates that the tool t allows to process c i after processing c j (or vice versa), ensuring that the task execution order satisfies the tool compatibility constraint;

[0023] The tool t can process the task point set S t There are the following constraint conditions:

[0024]

[0025] ∪S=C,S={S t |t∈T} (2)

[0026] Among them, S′ t Adapt the set of task points to tool t' (a tool in tool set T that is different from tool t);

[0027] Equation (1) indicates that there exist two tools t and t' in the tool set T, and their adaptation target set S t With S t' There must exist a non-empty intersection, meaning that the target c to be processed may have multiple tools that match it;

[0028] Equation (2) indicates that the union of the matching target sets of all tools in the tool set T is the target set C to be processed, which ensures that each target c to be processed has at least one tool to match it.

[0029] Preferably, step S3 includes:

[0030] S31. Calculate the maximum speed at the transition point of the path;

[0031] S32. According to the maximum speed limitation rule, insert the bridging point into the path point;

[0032] S33. Based on the bridging point information, update the robot's overall path trajectory; output a preliminary optimized path; the preliminary optimized path includes the path points, the bridging points, and speed information.

[0033] Preferably, the maximum speed at the path transition point is calculated using the following formula:

[0034]

[0035] Where: V max The maximum permissible speed at bridging points or waypoints, used for speed look-ahead optimization to ensure the trajectory meets task and dynamic constraints; d i-1,i The distance between the (i-1)th and i-th path points forms the spatial basis for determining the upper limit of velocity; T(c i ,t i ) is the task point c i Use tool t i Time constraints; DEC(v,b,d) i,i+1 ) represents the dynamically adjusted allowable speed function, where: v is the current speed; b is the constraint parameter; d i,i+1 This represents the distance between path points.

[0036] Preferably, step S4 includes:

[0037] S41. Evaluate the preliminary optimization path using the fitness function;

[0038] S42. Randomly generate neighborhood solutions for the preliminary optimized path;

[0039] S43, generating a new generated path according to the neighborhood solution; calculating the new generated path fitness;

[0040] S44, generating the better path according to the result of the new generated path fitness comparison.

[0041] As preferred, the dynamic adjustment formula of the mutation probability is:

[0042]

[0043] Wherein, P mutation is the dynamic mutation probability, which is adjusted with iterations; P min , P max are upper and lower limits of the mutation probability respectively; iter is the current iteration number; iter max is the total iteration number.

[0044] Compared with the prior art, the beneficial effects of the present application are embodied in:

[0045] 1. Different from the scheme of treating the path between target points as a constant value in the traditional technology, the present application substitutes the path points into the kinematics model for calculation, adopts the path optimization algorithm considering the optimization of the robot motion trajectory between path points, so that the path planning is more in line with the actual kinematics characteristics and dynamics constraints of the robot, effectively optimizes the total time consumption and total distance of each axis motion in task execution, and realizes more accurate path optimization in the dynamic multi-target processing scene.

[0046] 2. Different from the scheme that the traditional heuristic algorithm (such as genetic algorithm, ant colony algorithm, etc.) is easy to fall into local extremum or low efficiency in large-scale problems, the present application adopts the discretized solution space search scheme considering both local and global on the basis of the artificial bee colony algorithm, and introduces the dynamic mutation neighborhood search strategy, adjusts the mutation probability dynamically with the iteration number, realizes the adaptive switching of the early global search and the later local fine search, so that the algorithm can obtain high-quality solutions with higher efficiency in large-scale problems, compared with DABC, GA-FCS, etc., the solution quality is improved by 14%-31%, and the convergence speed and solution distribution concentration are significantly optimized. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 is the method flowchart of embodiment 1 of the present application;

[0048] Figure 2 is the path optimization mathematical model diagram based on the total execution time of the task of embodiment 1 of the present application;

[0049] Figure 3 is the initial population generation diagram of the Logistic chaotic mapping and dynamic reverse learning strategy of embodiment 1 of the present application;

[0050] Figure 4 Fig. 1 is a schematic diagram of optimizing trajectories between path points by using speed look-ahead optimization bridge strategy for embodiment 1 of the present application;

[0051] Figure 5 Fig. 2 is a schematic diagram of further optimizing path sequence by using dynamic variable neighborhood search strategy for embodiment 1 of the present application. DETAILED DESCRIPTION

[0052] In order to make the technical means, creative features, purposes and effects of the application easy to understand, the present application will be further described in conjunction with specific diagrams. However, the present application is not limited to the following embodiments.

[0053] It should be understood that the structure, proportion, size, etc. shown in the drawings attached to the present specification are only used to cooperate with the content disclosed in the specification, so that those skilled in the art can understand and read, and are not used to limit the implementation conditions of the present application. Therefore, any modification of structure, change of proportion relationship or adjustment of size, which does not affect the effects and purposes that can be achieved by the present application, should still fall within the scope of the technical content disclosed by the present application.

[0054] Embodiment 1

[0055] The multi-objective processing time optimization method of the industrial robot will be described in detail below in conjunction with the drawings:

[0056] As shown in Figure 1 , Figure 2 , a multi-objective processing time optimization method of an industrial robot includes the following steps:

[0057] Step S1, constructing a path optimization mathematical model based on total execution time of tasks, including task initialization, constraint processing, optimization target processing and finally task allocation.

[0058] S11, task initialization stage;

[0059] Suppose the task point set is C={c1, c2, …, c n}, and the tool set is T={t1, t2, …, t m}.

[0060] S12, constraint processing;

[0061] The constraint conditions in constraint processing include:

[0062] (1) The execution order of robot task points must satisfy tool compatibility:

[0063]

[0064] where f t (c i ,c j ) is the compatibility function of tool t for task points c i and c j , a non-zero value indicates that tool t allows c i to be processed after c j (or vice versa), ensuring that the task execution order meets the tool compatibility constraints.

[0065] (2) The set of task points S t that tool t can process exists as follows:

[0066]

[0067] ∪S=C,S={S t |t∈T} (2)

[0068] where S t ' is the set of task points that tool t' (a tool different from tool t in the tool set T) is adapted to;

[0069] Equation (1) indicates that there are two tools t and t' in the tool set T, which adapt to the target sets S t and S t' There must be a non-empty intersection, that is, there may be multiple tools that match the to-be-processed target c;

[0070] Equation (2) indicates that the union of the adaptation target sets of all tools in the tool set T is the to-be-processed target set C, ensuring that each to-be-processed target c is matched with at least one tool.

[0071] S13, optimize the total task execution time:

[0072] The objective function for optimizing the total task time in the target processing is defined as:

[0073]

[0074] where:

[0075] p is a single task point in the element robot path sequence of the robot path sequence PA;

[0076] v represents the kinematics parameter set (such as speed, acceleration, etc.) of the axis, which is affected by the robot constraint condition;

[0077] b is the set of motion connection mode setting parameters, which is affected by the robot constraint condition;

[0078] f(p,v,b) represents the execution time calculation function of executing a single task point once, which includes the influence of path distance and motion constraints.

[0079] MinimizeT total represents the total time consumption of task execution.

[0080] Step S2, generating an initial population by using Logistic chaotic mapping and dynamic reverse learning strategy (DOL strategy);

[0081] S21, generating an initial population by using Logistic chaotic mapping;

[0082] Logistic chaotic mapping is used to generate an initial population to improve the coverage of solution space and population diversity. The initial population is a set composed of multiple feasible solutions, each solution corresponding to a path sequence of industrial robot in a multi-objective task, i.e. the execution order of task points and tool combinations. It generates initial values through Logistic chaotic mapping to make solutions uniformly distributed in space, and then generates symmetric solutions by using dynamic reverse learning strategy, filters out better solutions to join in, while ensuring that each solution meets the constraint conditions such as tool compatibility, which covers a wide search space and contains high-quality initial solutions, and can improve the convergence speed and solution quality of the algorithm in the subsequent.

[0083] The quality of the initial population will greatly affect the convergence speed and solution quality of the algorithm. The ideal initial population should cover the solution space of the problem as much as possible to ensure the search breadth of the algorithm, and at the same time contain some high-quality initial solutions to ensure the convergence speed of the algorithm. Therefore, with reference to Figure 3 To improve the quality of solutions in the initial population while reducing the possibility of the algorithm falling into local extremal solutions, the TPIPO algorithm (Trajectory Planning Path Optimization, TP-IPO) proposed in the present application is based on a multi-objective task to construct a path optimization mathematical model with the total execution time of the task as the evaluation index, ensuring the constraints such as tool compatibility; it generates a high-quality initial population by using Logistic chaotic mapping and dynamic reverse learning strategy; then it bridges the path points by grouping and setting bridge points, dynamically adjusts the bridge point speed according to the speed limit to optimize the trajectory smoothness and task time consumption; finally, it realizes the balance between global search in the early stage and local optimization in the later stage by dynamically adjusting the mutation probability according to the number of iterations through the dynamic mutation neighborhood search strategy, and improves the path sequence quality and the convergence efficiency of the algorithm.

[0084] The chaotic mapping formula is:

[0085] X n+1 =μ·X n ·(1-X n ),μ∈(0,4),X n ∈(0,1)

[0086] wherein, X n+1X is the next value of the chaotic sequence; X is the current value of the chaotic sequence; μ is the branch parameter, when the value of μ is less than 3.5699456, the solution distribution is periodic, only when it is greater than 3.5699456, the Logistic mapping has chaotic properties. n X is the next value of the chaotic sequence; X is the current value of the chaotic sequence; μ is the branch parameter, when the value of μ is less than 3.5699456, the solution distribution is periodic, only when it is greater than 3.5699456, the Logistic mapping has chaotic properties.

[0087] S22, using dynamic inverse learning strategy, the quality of population is enhanced by generating inverse symmetric solution;

[0088] In the initialization of the population, the dynamic inverse learning strategy is introduced, and the quality of the population is enhanced by generating inverse symmetric solution. The calculation formula of the inverse symmetric solution is:

[0089] X reverse = X max -X

[0090] X max is the upper limit value of the solution space; X is the current solution; X reverse is the inverse symmetric solution of the current solution X.

[0091] The DOL strategy expands the symmetric space of the current point and its symmetric point in the entire search space, realizes a dynamically variable search space area, and reduces the possibility of the algorithm falling into a local extreme solution.

[0092] The population initialization stage is as follows: first, a chaotic parameter is obtained through the Logistic chaotic mapping to improve the uniformity of the solution distribution and enhance the diversity of the initial population. Then the dynamic reverse individual is generated by using the DOL strategy, and the individual is selected to join the initial population according to the greedy strategy, so as to improve the quality of the initial population. In the initial population generation and subsequent iteration, the value of each dimension of the individual is set with feasibility constraints to avoid the waste of computing resources caused by invalid individuals and ensure that each individual is a feasible solution.

[0093] In the initial population generation of industrial robot path optimization, the uniformity of the solution distribution is enhanced through chaotic mapping, and the search space is expanded by dynamic reverse learning, which effectively improves the population quality and algorithm performance

[0094] Step S3, the trajectory between the path points is optimized by using the speed look-ahead optimization bridging strategy to generate a preliminary optimized path;

[0095] Referring to Figure 4 , the speed look-ahead optimization bridging strategy is proposed, which optimizes the transition speed between path points by grouping the path points and introducing bridge points between the path groups to realize dynamic optimization of the path.

[0096] S31, the maximum speed of the path transition point is calculated;

[0097] The path transition point is an intermediate point set in path planning to realize smooth transition between adjacent path points. The path transition point is generated by a speed look-ahead optimization bridging strategy, used to connect path points in different groups, adjust the transition speed to optimize the smoothness of the overall trajectory and the task execution time under the premise of meeting the robot motion constraints, ensure smoother and smoother motion of the robot between different task points, and avoid the problem of incoherent trajectory caused by sudden speed change. The maximum speed of the path transition point is calculated by the following formula:

[0098]

[0099] Wherein:

[0100] V max is the maximum allowed speed of the bridging point or path point, used for speed look-ahead optimization to ensure that the trajectory meets the task and dynamic constraints.

[0101] d i-1,i is the distance between the i-1th and ith path points, which determines the spatial basis of the speed upper limit.

[0102] T(c i ,t i ) is the time constraint of task point c i using tool t i .

[0103] DEC(v,b,d i,i+1 ) is the dynamically adjusted allowed speed function, wherein: v is the current speed; b is the constraint parameter; d i,i+1 is the distance between path points.

[0104] S32, insert bridging points;

[0105] According to the maximum speed limit rule of the path transition point described above, one or more bridging points are inserted between each two groups of path points: the speed of each bridging point is calculated so that the transition trajectory is as smooth as possible.

[0106] S33, update the path trajectory;

[0107] Based on the speed adjusted by the bridging point and the bridging point information, the overall path trajectory of the robot is updated. The final output is the preliminary optimized path, including all path points, bridging points and speed information.

[0108] Step S4, further optimize the path sequence by using a dynamic variation neighborhood search strategy

[0109] Referring to Figure 5The application dynamically adjusts the mutation probability in the iteration process to realize the balance between early global search and late local optimization. The application combines dynamic mutation probability adjustment with fitness evaluation, and realizes the dynamic balance between global and local search for the multi-objective time optimization scene of industrial robots.

[0110] S41, calculating path fitness: using a fitness function to evaluate the quality of the current path. The fitness function and the fitness evaluation method are prior art.

[0111] S42, randomly generating a neighborhood solution:

[0112] The dynamic adjustment formula of the mutation probability is:

[0113]

[0114] Wherein: P mutation is the dynamic mutation probability, which is adjusted with iterations; P min , P max are the upper and lower limits of the mutation probability respectively; iter is the current iteration number; iter max is the total iteration number.

[0115] Based on the current path sequence P, that is, the preliminary optimized path, a plurality of new path solutions are generated in its neighborhood. The generation method of the neighborhood solution usually includes path exchange, inversion, insertion and other operations.

[0116] S43, calculating the fitness of the newly generated path.

[0117] S44, selecting a better path according to the result of fitness comparison.

[0118] Through dynamic mutation neighborhood search, the local area of the solution is focused on optimization in the later stage of the algorithm, and the path planning quality is improved.

[0119] The trajectory planning intelligent path optimization algorithm (TP-IPO) constructs the population through the initialization method combining the Logistic chaotic mapping strategy and the dynamic reverse learning strategy, realizes the search of the discretized solution space considering the local and global, and introduces the dynamic mutation neighborhood search strategy applied to the individual forced update operation, effectively enhances the population diversity, improves the ability to jump out of the local optimal solution, and has good solving effect in the large-scale industrial robot multi-objective processing path optimization problem.

Claims

1. A method for optimizing the multi-objective processing time of an industrial robot, characterized in that, Includes the following steps: S1. Construct a mathematical model for path optimization based on the total execution time of the task, and generate a path sequence; S2. Generate an initial population of the path sequence using the Logistic chaotic mapping method; generate an inverse symmetric solution of the initial population using a dynamic inverse learning strategy, and add it to the initial population; S3. By inserting bridging points, the path points of the path sequence in the initial population are grouped. Generate a preliminary optimized path; S4. Further optimize the preliminary optimized path through a dynamic mutation neighborhood search strategy to generate a better path.

2. The method for optimizing the multi-objective processing time of an industrial robot according to claim 1, characterized in that, Step S1 includes: S11, Task initialization phase; Assume the set of task points is C = {c1, c2, ..., c...} n The toolset is T = {t1, t2, ..., t}. m }; S12, Constraint handling; S13. Objective function for generating total task time: in: p is an element of the robot path sequence PA, representing a single task point in the robot path sequence; v represents the set of kinematic parameters of the axis (such as velocity, acceleration, etc.), which are affected by robot constraints; b is the set of motion connection mode setting parameters, which are affected by robot constraints; f(p,v,b) represents the function for calculating the execution time of executing a single task point, including the influence of path distance and motion constraints.

3. The method for optimizing the multi-objective processing time of an industrial robot according to claim 2, characterized in that, The constraints in the constraint processing include: The execution order of robot task points must satisfy tool compatibility: Among them, f t (c i ,c j ) is the tool t for task point c i and c j The compatibility function, with non-zero values ​​representing the tool t, allows for processing c. i Post-processing c j (Or vice versa), ensure that the task execution order meets tool compatibility constraints; The set of task points that tool t can handle is S t The following constraints exist: ∪S=C,S={S t |t∈T} (2) Among them, S t' Adapt the set of task points to tool t' (a tool in tool set T that is different from tool t); Equation (1) indicates that there exist two tools t and t' in the tool set T, and their adaptation target set S t With S t' There must exist a non-empty intersection, meaning that the target c to be processed may have multiple tools that match it; Equation (2) indicates that the union of the matching target sets of all tools in the tool set T is the set of target to be processed C, which ensures that each target to be processed c has at least one tool that matches it.

4. The method for optimizing the multi-objective processing time of an industrial robot according to claim 1, characterized in that, Step S3 includes: S31. Calculate the maximum speed at the transition point of the path; S32. According to the maximum speed limitation rule, insert the bridging point into the path point; S33. Based on the bridging point information, update the robot's overall path trajectory; output a preliminary optimized path; the preliminary optimized path includes the path points, the bridging points, and speed information.

5. The method for optimizing the multi-objective processing time of an industrial robot according to claim 4, characterized in that, The maximum speed at the path transition point is calculated using the following formula: Where: V max The maximum permissible speed at bridging points or waypoints, used for speed look-ahead optimization to ensure the trajectory meets task and dynamic constraints; d i-1,i The distance between the (i-1)th and i-th path points forms the spatial basis for determining the upper limit of velocity; T(c i ,t i ) is the task point c i Use tool t i Time constraints; DEC(v,b,d) i,i+1 ) represents the dynamically adjusted allowable speed function, where: v is the current speed; b is the constraint parameter; d i,i+1 This represents the distance between path points.

6. The method for optimizing the multi-objective processing time of an industrial robot according to claim 1, characterized in that, Step S4 includes: S41. Evaluate the preliminary optimization path using the fitness function; S42. Randomly generate neighborhood solutions for the preliminary optimized path; S43. Generate a new generated path based on the neighborhood solution; calculate the fitness of the new generated path; S44. Based on the results of the fitness comparison of the newly generated path, generate the better path.

7. The method for optimizing the multi-objective processing time of an industrial robot according to claim 6, characterized in that, The dynamic adjustment formula for the mutation probability is: Where: P mutation P represents the dynamic mutation probability, which adjusts with iteration. min P max These represent the upper and lower bounds of the mutation probability, respectively; iter is the current iteration number; iter max This represents the total number of iterations.

Images