An agv system hybrid task allocation method applied to an assembly line

Abstract

The application discloses an AGV system mixed task allocation method applied to an assembly line and belongs to the intelligent manufacturing field. When a task allocation trigger condition is met and there is a material distribution task to be allocated, firstly, the material distribution task to be allocated is divided according to a task combination forming rule, and multiple task combinations are formed; then, a task combination architecture composed of multiple task layers is formed according to a task combination hierarchical rule, and task combinations located in adjacent layers of the task combination architecture are bound according to a task combination binding rule; next, material distribution tasks between the bound task combinations are bound according to a task binding rule, and a task binding relationship is formed; finally, based on the priority of the determined task combination, a mixed task allocation scheme is generated according to an AGV task allocation rule, and the material distribution task to be allocated is allocated to the required number of AGVs. The method considers the correlation between tasks and the recombination time consumption when the required number of AGVs for executing the previous and subsequent tasks is different, and provides an AGV system mixed task allocation method applied to an assembly line, which can effectively reduce the task delivery delay caused by the recombination time consumption.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing, and more specifically to a hybrid task allocation method for AGV systems applied to assembly lines. Background Technology

[0002] Due to their high flexibility and intelligence, Automated Guided Vehicle (AGV) systems are widely used in material handling on assembly lines. Research on AGV systems mainly focuses on task allocation, path planning, and traffic control. Task allocation has been a particularly popular research area, with the key being determining the distribution relationship between material handling tasks and AGVs, as well as the order in which AGVs execute these tasks. While there is considerable research on task allocation, existing studies typically consider multiple material handling tasks within the same assembly operation separately, neglecting the interrelationships between these tasks. Furthermore, on assembly lines for large equipment, materials vary significantly in size and shape. Systems may simultaneously handle single-AGV material handling tasks for standard parts and multi-AGV collaborative handling tasks for large components. Compared to the traditional task allocation problem for single-AGV material handling, multi-AGV collaborative handling involves AGV assembly and disassembly, making the material allocation problem for complex systems with both single-AGV and multi-AGV collaborative handling tasks far more complex, and current research in this area is limited. Existing technologies lack research on task allocation for complex systems that consider the interrelationships between material handling tasks. Summary of the Invention

[0003] To address the technical deficiencies in existing technologies, this invention proposes a hybrid task allocation method for AGV systems applied to assembly lines. By considering the correlation between tasks, the method allocates the required number of AGVs to the material delivery tasks to be assigned based on task combinations. The proposed method is more in line with actual assembly scenarios, and the task allocation scheme generated by the method of this invention has less time spent on AGV combination and disassembly, effectively reducing the time spent on reassembly and thus reducing the delay in task delivery caused by reassembly time.

[0004] This invention is achieved through the following technical solution:

[0005] A method for hybrid task allocation in an AGV system applied to an assembly line includes the following steps:

[0006] Step 1: Determine if the task assignment trigger condition is met. If yes, proceed to Step 2; otherwise, proceed to Step 1.

[0007] Step 2: Determine if there are any material delivery tasks to be assigned. If yes, proceed to Step 3; otherwise, proceed to Step 1.

[0008] Step 3: Collect the material delivery tasks to be assigned in the system, and divide the material delivery tasks to be assigned into multiple task combinations according to the task combination rules.

[0009] The task combination formation rule is to group material delivery tasks with task relevance into the same task combination, and material delivery tasks without task relevance into different task combinations.

[0010] The task correlation refers to the task correlation between multiple material delivery tasks responsible for transporting various materials when an assembly operation requires multiple materials.

[0011] Step 4: Based on the task combination layering rules, form a task combination architecture consisting of multiple task layers. At the same time, based on the task combination binding rules, bind the task combinations located in adjacent layers of the task combination architecture.

[0012] Step 5: According to the task binding rules, bind the material delivery tasks between the bound task groups to form a task binding relationship, and obtain the bound material delivery task information.

[0013] Step 6: Determine the priority of task combinations based on the task combination priority determination rules.

[0014] Step 7: Based on the priority of the determined task combination, generate a mixed task allocation scheme according to the AGV task allocation rules, and assign the material delivery tasks to be allocated to the required number of AGVs for that task.

[0015] Furthermore, the task allocation trigger condition is periodic.

[0016] Furthermore, the material delivery tasks to be assigned are generated in advance by the system according to the material delivery task generation rules.

[0017] Furthermore, the material delivery task generation rules include rules for determining the task generation time and rules for determining the optimal delivery time period.

[0018] Furthermore, the rules for determining the task generation time are as follows:

[0019] For materials without a line-side buffer, the start time of the current assembly operation is the task generation time of the material delivery task corresponding to the materials required for the next assembly operation;

[0020] For materials in the wired buffer area, the time when the material in the wired buffer area is consumed to the set lower limit of the remaining quantity is the time when the material delivery task for the corresponding material is generated.

[0021] Furthermore, the lower limit of the set margin is the amount of material required for one assembly operation.

[0022] Furthermore, the optimal delivery time period determination rule for the task is to determine the optimal delivery time period required by the assembly operation served by the material delivery task for its required materials.

[0023] Furthermore, the optimal delivery time for the task is a period of time from the start time to the end time.

[0024] Furthermore, the start time of the optimal delivery time period is the earliest time when the assembly station allows the AGV to unload materials.

[0025] Furthermore, the end time of the optimal delivery time period is the latest time at which materials are allowed to be delivered during the assembly operation.

[0026] Furthermore, the rules for determining the optimal delivery time for the task are as follows:

[0027] For materials without a line-side buffer, the optimal delivery time for the material delivery task is [t]. p +t s -0.5△t, t p +t s -0.5△t],t p For the time when the task is generated, t s The predicted completion time for the current assembly operation is given by Δt, where Δt is the fault tolerance time.

[0028] For materials in the wired buffer area, the optimal delivery time for the material delivery task generated by the system to handle these materials is [t0, t...]. p +2t s ], where t0 is the current time.

[0029] Furthermore, the fault tolerance time Δt is a constant.

[0030] Furthermore, the task combination hierarchical rule includes the following steps:

[0031] Step 4.1.1: Sort the task combinations in ascending order according to the end time of the best delivery time period of the task combination to form a sequence of task combinations to be layered. Set the first layer of the task combination architecture as the target layer, take out the first task combination in the sequence and put it into the target layer.

[0032] Step 4.1.2: Take out the first task combination in the task combination sequence to be layered and set it as the current task combination. Determine whether there is a task combination in the target layer that has binding feasibility with the current task combination. If not, proceed to step 4.1.3. If it exists, proceed to step 4.1.4.

[0033] Step 4.1.3: According to the forward task combination determination rules, determine whether there is a forward task combination of the current task combination in the target layer. If it exists, proceed to step 4.1.4; if it does not exist, put the current task combination into the target layer and proceed to step 4.1.6.

[0034] Step 4.1.4: Determine if the target layer is the first layer of the task composition architecture. If yes, proceed to step 4.1.5; otherwise, invoke the task composition binding rule and proceed to step 4.1.5.

[0035] Step 4.1.5: Set the next layer of the target layer as the target layer, put the current task combination into the target layer, and proceed to step 4.1.6;

[0036] Step 4.1.6: Determine whether there are any task combinations in the task combination sequence to be layered. If yes, proceed to step 4.1.2; otherwise, end the call to the task combination layering rule.

[0037] Furthermore, the optimal delivery time period for the task combination is the same as the optimal delivery time period for each task in the task combination.

[0038] Furthermore, the task composition architecture consists of multiple task composition layers capable of storing task compositions.

[0039] Furthermore, the feasibility of the binding is determined by the following rules:

[0040] Assume two material delivery tasks T with the same number of AGVs required. i T i+1 They belong to the task combination TG. n TG n+1 If the AGV or AGV combination is executing T i Then execute T i+1 T can still be used i+1 If the task is delivered within its optimal delivery time, then the task combination TG is considered to be complete. n+1 Bound to TG n Binding is feasible.

[0041] Furthermore, the forward task combination determination rule is as follows:

[0042] Assume two material delivery tasks T with the same number of AGVs required. i T i+1 They belong to the task combination TG. n TG n+1 If AGV or AGV combination in T i Complete T at any time during the optimal delivery period i The execution of T, then the execution of T i+1 T i+1 The delivery times are all earlier than T.i+1 The optimal delivery time period starts at the time of task combination TG. n For TG n+1 Forward task combination.

[0043] Furthermore, the task combination binding rule includes the following steps:

[0044] Step 4.2.1: Generate a similarity matrix A between the target layer and the task combinations in the layer above the target layer. The number of rows in similarity matrix A is the number of task combinations in the target layer, and the number of columns in similarity matrix A is the number of task combinations in the layer above the target layer. The element in the i-th row and j-th column of the similarity matrix is ​​the task combination similarity between the i-th task combination in the target layer and the j-th task combination in the layer above the target layer. If the calculated similarity value of two task combinations is 0, or they cannot be bound, then the element at the corresponding position is 0.

[0045] Step 4.2.2: Determine whether there are non-zero elements in the similarity matrix A. If yes, proceed to step 4.2.3; otherwise, end the task and call the binding rules.

[0046] Step 4.2.3: Determine the row number x and column number y of the maximum element, set the xth task combination in the target layer as the follower task combination, set the yth task combination in the layer above the target layer as the target task combination, bind the follower task combination to the target task combination, set the elements in the xth row and yth column of the similarity matrix A to zero, and proceed to step 4.2.4.

[0047] Step 4.2.4: Invoke the task binding rules and proceed to step 4.2.2.

[0048] Furthermore, the calculation rules for the task combination similarity are as follows:

[0049] Assuming the task combination TG n+1 Bound to TG n With binding feasibility or task combination TG n Bound to TG n+1 Binding is feasible; obtain TG. n TG n+1 The task array is composed of the number of tasks in the task combination and the number of AGVs required to execute each task. Let's assume the task array is TG. n TG n+1 The constituent arrays are [abbc] and [bbcd], respectively. Elements with the same constituent array are taken to form a similarity array [bbc]. Task combination TG n With TG n+1 Task combination similarity f s The calculation formula is as follows:

[0050]

[0051] In the formula, w1 and w2 are weight coefficients, both w1 and w2 are positive numbers, and w1+w2=1.

[0052] If the task combination TG n+1 and TG n have no mutual binding feasibility, then the task combination TG n and TG n+1 The task combination similarity f s is 0.

[0053] Further, the weight coefficients w1 and w2 both take a value of 0.5.

[0054] Further, the task binding rule includes the following steps:

[0055] Step 5.1: Sort the material distribution tasks in the following task combination and the target task combination in descending order according to the number of AGVs required for the material distribution tasks, set the following task as the first material distribution task in the following task combination, and set the target task as the first material distribution task in the target task combination;

[0056] Step 5.2: Judge the magnitude relationship between the number m1 of AGVs required for the following task and the number m2 of AGVs required for the target task. If m1<m2, proceed to step 4.3.3; if m1=m2, bind the following task to the target task and proceed to step 5.4; if m1>m2, end the call of the task binding rule;

[0057] Step 5.3: Judge whether there is a next material distribution task in the target task combination. If yes, set the target task as the next material distribution task in the target task combination and proceed to step 5.2; if no, end the call of the task binding rule;

[0058] Step 5.4: Judge whether both the following task combination and the target task combination have a next material distribution task. If yes, set the following task as the next material distribution task in the bound task combination, set the target task as the next material distribution task in the target task combination, and proceed to step 5.2; if no, end the call of the task binding rule.

[0059] Further, the bound material distribution task information includes the free task combinations that are not bound to any task combination, the following task combinations bound after each task combination, the free tasks that are not bound to any task, and the situation of the following tasks bound after each task.

[0060] Furthermore, the task combination priority determination rule is one of the following principles: priority of task combinations with tight end time, priority of task combinations with many following task combinations, priority of task combinations with many following tasks, and priority of task combinations with a large total number of required AGVs.

[0061] Furthermore, the principle of prioritizing task combinations with tighter deadlines gives higher priority to task combinations with more urgent deadlines.

[0062] Furthermore, the principle of prioritizing task combinations with more follower task combinations means that task combinations with more follower task combinations after task combination have higher priority.

[0063] Furthermore, the principle of prioritizing task combinations with more bound tasks means that task combinations with a greater sum of the number of follow-up tasks after the task combination and the number of tasks in the task combination itself have higher priority.

[0064] Furthermore, the principle of prioritizing task combinations with a larger total number of AGVs means that when all tasks in a task combination are executed simultaneously, task combinations with a larger total number of AGVs have higher priority.

[0065] Furthermore, the specific steps of the AGV task allocation rule are as follows:

[0066] Step 7.1: Arrange the task combinations in descending order according to their priority to form a task combination sequence.

[0067] The task combination sequence does not include non-free task combinations.

[0068] Step 7.2: Take out the first task combination in the sequence and set it as the task combination to be assigned.

[0069] Step 7.3: According to the free task AGV allocation method, allocate the required number of AGVs to the free tasks in the task combination to be allocated in turn.

[0070] Step 7.4: Determine if there is a follower task combination after the task combination to be assigned. If not, proceed to step 7.6. If yes, according to the follower task AGV allocation method, allocate the required number of AGVs for the follower tasks in the follower task combination, and proceed to step 7.5.

[0071] Step 7.5: Set the follow task group as the task group to be assigned, and then assign the required number of AGVs to the free tasks in the task group to be assigned, and proceed to step 7.4.

[0072] Step 7.6: Determine if the task combination sequence is empty. If not, proceed to step 7.2. If yes, end the call to the AGV task allocation rule.

[0073] Furthermore, the free-task AGV allocation method is specifically as follows:

[0074] The estimated arrival time of all AGVs in the system to the loading point of the free task is calculated. Assuming that the number of AGVs required for the free task is 'a', the AGV with the earliest estimated arrival time is selected as the execution AGV for the free task, and it is added to the end of the list of pending tasks of the aforementioned AGVs.

[0075] Furthermore, the following task AGV allocation method involves obtaining the AGV to which the following task is bound to execute, assigning the following task to the same AGV, and in the list of tasks to be executed of the assigned AGV, the following task immediately follows the task to which it is bound.

[0076] Furthermore, the AGV's task list is a list of material delivery tasks that the AGV needs to perform, and the order of the material delivery tasks in the table is the AGV's task execution order.

[0077] Compared with the prior art, the present invention has at least the following beneficial effects or advantages:

[0078] This invention proposes a hybrid task allocation method for AGV systems applied to assembly lines. Based on the correlation between tasks, the method allocates material delivery tasks to be assigned in the system from the perspective of task combination. The proposed method is more in line with actual assembly scenarios. Moreover, the task allocation scheme generated by the method of this invention has less time spent on AGV combination and disassembly, effectively reducing the task delivery delays caused by reassembly time. Attached Figure Description

[0079] The present invention will now be described in further detail with reference to the accompanying drawings:

[0080] Figure 1 This is a flowchart of a hybrid task allocation method for an AGV system applied to an assembly line according to the present invention;

[0081] Figure 2 This is a flowchart of the task combination and hierarchical rules of the present invention;

[0082] Figure 3 This is a flowchart illustrating the AGV task allocation rules of the present invention. Detailed Implementation

[0083] The present invention will now be described in detail with reference to the accompanying drawings, but this is not intended to limit the scope of the invention. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0084] Since the allocation of mixed tasks is quite complex, the following assumptions are made to ensure the method's generality:

[0085] 1. All AGVs in the system are identical;

[0086] 2. Ignore the impact of motion conflicts, loop deadlocks, and AGV malfunctions on material delivery;

[0087] 3. An AGV can only handle one task at a time;

[0088] 4. When an AGV combination consisting of multiple AGVs completes a collaborative transport task, if the task lists of these AGVs are all empty, the AGV combination will not be disbanded.

[0089] Figure 1 This is a flowchart of a hybrid task allocation method for an AGV system applied to an assembly line, as described in this invention. The method specifically includes the following steps:

[0090] Step 1: Determine if the task assignment trigger condition is met. If yes, proceed to Step 2; otherwise, proceed to Step 1.

[0091] The task allocation triggering condition adopted in this invention is periodic triggering. The triggering period can be flexibly set according to the actual situation of the assembly line. For example, for an assembly line with a production cycle, the triggering period can be set to one production cycle.

[0092] Step 2: Determine if there are any material delivery tasks to be assigned. If yes, proceed to Step 3; otherwise, proceed to Step 1.

[0093] Since generating material delivery tasks only when assembly workstations require materials would inevitably lead to production stoppages due to material shortages, the system generates unassigned material delivery tasks in advance according to material delivery task generation rules. These rules include rules for determining the task generation time and the optimal delivery time period.

[0094] The task generation time determination rule is used to determine when the system generates tasks, as follows: For materials without a line-side buffer, the start time of the current assembly operation is the task generation time of the material delivery task corresponding to the materials required for the next assembly operation; For materials with a line-side buffer, the time when the material in the line-side buffer is consumed to the set lower limit of the remaining amount is the task generation time of the corresponding material delivery task.

[0095] For large and valuable materials, line-side buffer zones are typically not set up next to the workstation. For material delivery tasks of these materials, if the task is generated too early, it increases the likelihood that the target workstation will be busy when the AGV arrives, preventing the AGV from unloading and thus consuming its transport capacity, increasing the energy consumption of the handling system. If the task is generated too late, the AGV may not be able to deliver the materials to the target workstation on time, causing the workstation to be idle and production to stop. Therefore, for the above-mentioned materials, this invention sets the start time of the current assembly operation to the task generation time of the material delivery task corresponding to the materials required for the next assembly operation, ensuring the AGV transport time is taken without generating tasks too early. Considering lean manufacturing principles, this invention sets the lower limit of the allowance to be the amount of material required for one assembly operation, in order to reduce line-side inventory.

[0096] The optimal delivery time period for a material delivery task is determined by the following rules: For materials without a line-side buffer, the optimal delivery time period for the materials is [t]. p +t s -0.5△t, t p +t s -0.5△t],t p For the time when the task is generated, t s The predicted completion time for the current assembly operation is given by Δt, where Δt is the fault tolerance time. For materials in the wired buffer area, the optimal delivery time for the material delivery task generated by the system to handle these materials is [t0, t...]. p +2t s ], where t0 is the current time.

[0097] For materials without a line-side buffer, the theoretical optimal delivery time is the moment when the workstation actually needs the material. However, this exact moment is inconvenient for task scheduling. In actual assembly environments, to address this issue, a time period is typically set, which is the fault tolerance time in this invention. Tasks delivered within this time period are considered to have no impact on production. The predicted completion time of assembly operations can be obtained based on historical assembly operation times using neural network models, gray-scale theory prediction models, etc. In this implementation example, the predicted completion time of assembly operations uses the average time of assembly operations at this workstation.

[0098] Step 3: Collect the material delivery tasks to be assigned in the system, and divide the material delivery tasks to be assigned into multiple task combinations according to the task combination rules.

[0099] The task combination formation rule is to group material delivery tasks with task relevance into the same task combination, and material delivery tasks without task relevance into different task combinations.

[0100] The task correlation refers to the task correlation between multiple material delivery tasks responsible for transporting various materials when an assembly operation requires multiple materials.

[0101] In actual production assembly, an assembly operation usually requires multiple materials. The system will generate multiple material delivery tasks and assign them to multiple AGVs. The late arrival of any task will cause the assembly operation to be delayed. At the same time, when multiple tasks are late, the delay time of the assembly operation is the late arrival time of the task with the longest late arrival time, rather than the sum of the late arrival times of these late tasks. Therefore, this invention packages the multiple material delivery tasks generated by the system for an assembly operation into a task combination. Based on the task combination, AGVs are assigned to tasks in the system so that tasks in the same task combination are delivered to the target workstation as simultaneously as possible.

[0102] Step 4: Based on the task combination layering rules, form a task combination architecture consisting of multiple task layers. At the same time, based on the task combination binding rules, bind the task combinations located in adjacent layers of the task combination architecture.

[0103] The task combination layering rules and task combination binding rules will be described in detail later.

[0104] Step 5: According to the task binding rules, bind the material delivery tasks between the bound task groups to form a task binding relationship, and obtain the bound material delivery task information.

[0105] After the task combination binding operation in step 4, the tasks between the two task combinations need to be bound according to the task binding rules. The task binding rules include the following steps:

[0106] Step 5.1: According to the number of AGVs required for the material delivery task, sort the material delivery tasks in the following task group and the target task group in descending order, set the following task as the first material delivery task in the following task group, and set the target task as the first material delivery task in the target task group;

[0107] Step 5.2: Determine the magnitude relationship between the number of AGVs m1 required for the following task and the number of AGVs m2 required for the target task. If m1<m2, proceed to step 4.3.3; if m1=m2, bind the following task to the target task and proceed to step 5.4; if m1>m2, terminate the call of the task binding rule;

[0108] Step 5.3: Determine whether there is a next material distribution task in the target task combination. If yes, set the target task as the next material distribution task in the target task combination and proceed to step 5.2; if no, terminate the call of the task binding rule;

[0109] Step 5.4: Determine whether both the following task combination and the target task combination have a next material distribution task. If yes, set the following task as the next material distribution task in the bound task combination, set the target task as the next material distribution task in the target task combination, and proceed to step 5.2; if no, terminate the call of the task binding rule.

[0110] An example of the task binding rule is described as follows: Assume that the following task combination is TG1, which includes tasks T1, T2, T3, and T4, and the number of AGVs required for the tasks is 4, 4, 2, and 1 respectively; the target task combination is TG2, which includes tasks T5, T6, T7, and T8, and the number of AGVs required for the tasks is 4, 2, 2, and 1 respectively. The task binding result is that T1 is bound to T5, T3 is bound to T6, T4 is bound to T8, and other tasks are not bound.

[0111] After the task binding is completed, the material distribution task information is obtained, including the free task combination that is not bound to any task combination, the following task combination bound after each task combination, the free task that is not bound to any task, and the following task bound after each task.

[0112] Step 6: Determine the priority of the task combination according to the task combination priority determination rule.

[0113] The priority determination rules for task combinations can be one of the following: Prioritizing task combinations with urgent end times; prioritizing task combinations with many following task combinations; prioritizing task combinations with many following tasks; and prioritizing task combinations with a large total number of AGVs required. The "prioritizing task combinations with urgent end times" principle gives higher priority to task combinations with more urgent end times. The "prioritizing task combinations with many following task combinations" principle gives higher priority to task combinations with more following task combinations. The "prioritizing task combinations with many bound tasks" principle gives higher priority to task combinations where the sum of the number of following tasks and the number of tasks in the task combination itself is greater. The "prioritizing task combinations with a large total number of AGVs required" principle gives higher priority to task combinations where the total number of AGVs required when all tasks in the task combination are executed simultaneously is greater. It should be noted that for the "prioritizing task combinations with many following task combinations" principle, if no task combination is bound to task combination TG... n Then, the task combination TG n The number of follow-up task combinations is 0, if the task combination TG n+1 Bound to the TG task group n Afterwards, the task force TG n+2 Bound to the TG task group n+1 Afterwards, no task combination is bound to the task combination TG. n+2 Then, the task combination TG n The number of follower task combinations is 2; for task combinations with more follower tasks, the priority principle is applied, assuming task combination TG. n TG n+1 TG n+2 The number of tasks themselves are 2, 3, and 4 respectively. If there is no task combination bound to the task combination TG, n Then, the task combination TG n The sum of the numbers is 2, if the task combination TG n+1 Bound to the TG task group n Afterwards, the task force TG n+2 Bound to the TG task group n+1 Afterwards, no task combination is bound to the task combination TG. n+2 Then, the task combination TG n The sum of is 9.

[0114] Step 7: Based on the priority of the determined task combination, generate a mixed task allocation scheme according to the AGV task allocation rules, and assign the material delivery tasks to be allocated to the required number of AGVs for that task.

[0115] AGV task allocation rules are used to assign the required number of AGVs to material delivery tasks that exist in the system.

[0116] Figure 3 This is a flowchart of the AGV task allocation rule of the present invention. The specific steps of the AGV task allocation rule are as follows:

[0117] Step 7.1: Arrange the task combinations in descending order according to their priority to form a task combination sequence.

[0118] Based on the priority of the task combinations determined in step 6, the task combinations are arranged in descending order, and the task combination sequence does not include non-free task combinations.

[0119] Step 7.2: Take out the first task combination in the sequence and set it as the task combination to be assigned.

[0120] Step 7.3: According to the free task AGV allocation method, allocate the required number of AGVs to the free tasks in the task combination to be allocated in turn.

[0121] The free task AGV allocation method calculates the estimated arrival time of all AGVs in the system to the loading point of the free task. Assuming that the number of AGVs required for the free task is 'a', the AGV with the earliest estimated arrival time is selected as the execution AGV for the free task, and the free task is added to the end of the list of pending tasks of the aforementioned 'a' execution AGVs.

[0122] In this implementation case, the estimated arrival time of the AGV is calculated using the following formula:

[0123] t g，i =t g +d(p g L i ) / v+β*t c

[0124] In the formula, t g , i For the g-th AGV to reach task T i The estimated arrival time of the AGV, t g P represents the current available time of the g-th AGV. g Let L be the position of the g-th AGV at its current available moment. i For material distribution task T i The loading point is d(A,B), which is the shortest distance between positions A and B, v is the AGV's travel speed, and t is the shortest distance between positions A and B. c β represents the AGV reassembly time, and β is the AGV reassembly coefficient.

[0125] In this implementation case, the method for determining the current available time of the AGV is as follows: when the AGV is idle, the current available time of the AGV is the current time; when the AGV is performing a task, the current available time of the AGV is the time when it completes all the tasks it needs to perform.

[0126] In this implementation case, the AGV recombination coefficient is determined as follows: when the number of AGVs required for the previous task is the same as the current task, or when the previous task is a single-load task, the AGV recombination coefficient is 0; otherwise, the AGV recombination coefficient is 1.

[0127] Step 7.4: Determine if there is a follower task combination after the task combination to be assigned. If not, proceed to step 7.6. If yes, according to the follower task AGV allocation method, allocate the required number of AGVs for the follower tasks in the follower task combination, and proceed to step 7.5.

[0128] The AGV assignment method for follow tasks involves obtaining the AGV to which the follow task is bound to execute, assigning the follow task to the same AGV, and placing the follow task immediately after the task to be executed in the list of tasks to be executed of the assigned AGV.

[0129] Step 7.5: Set the follow task group as the task group to be assigned, and then assign the required number of AGVs to the free tasks in the task group to be assigned, and proceed to step 7.4.

[0130] Step 7.6: Determine if the task combination sequence is empty. If not, proceed to step 7.2. If yes, end the call to the AGV task allocation rule.

[0131] Figure 2 This is a flowchart of the task composition and layering rule of the present invention. The task composition and layering rule specifically includes the following steps:

[0132] Step 4.1.1: Sort the task combinations in ascending order according to the end time of the best delivery time period of the task combination to form a sequence of task combinations to be layered. Set the first layer of the task combination architecture as the target layer, take out the first task combination in the sequence and put it into the target layer.

[0133] All tasks in a task group serve the same assembly task, therefore, the tasks in a task group have the same optimal delivery time. The optimal delivery time of a task group is the same as the optimal delivery time of the tasks it contains. The task groups are sorted in ascending order according to the end time of their optimal delivery time to obtain a sequence of task groups to be layered. The first layer of the task group architecture is set as the target layer. The first task group in the sequence of task groups to be layered is placed in the target layer and then removed from the sequence. Proceed to step 4.1.2.

[0134] Step 4.1.2: Take out the first task combination in the task combination sequence to be layered and set it as the current task combination. Determine whether there is a task combination in the target layer that has binding feasibility with the current task combination. If not, proceed to step 4.1.3. If it exists, proceed to step 4.1.4.

[0135] The rule for determining the feasibility of binding is as follows: assuming material delivery tasks T with the same AGV demand. i T i+1 They belong to the task combination TG. n TG n+1 If the AGV or AGV combination is executing T i Then execute T i+1 T can still be used i+1 If the task is delivered within its optimal delivery time, then the task combination TG is considered to be complete. n+1 Bound to TG n It has binding feasibility, otherwise the task combination TG n+1 Bound to TG n Binding is not feasible.

[0136] Step 4.1.3: According to the forward task combination determination rules, determine whether there is a forward task combination of the current task combination in the target layer. If it exists, proceed to step 4.1.4; if it does not exist, put the current task combination into the target layer and proceed to step 4.1.6.

[0137] The forward task combination determination rule is as follows: Assume that there are material delivery tasks T with the same AGV demand. i T i+1 They belong to the task combination TG. n TG n+1 If AGV or AGV combination in T i Complete T at any time during the optimal delivery period i The execution of T, then the execution of T i+1 T i+1 The delivery times are all earlier than T. i+1 The optimal delivery time period starts at the time of task combination TG. n For TG n+1 Forward task combination.

[0138] Step 4.1.4: Determine if the target layer is the first layer of the task composition architecture. If yes, proceed to step 4.1.5; otherwise, invoke the task composition binding rule and proceed to step 4.1.5.

[0139] When the target layer has a task combination that is feasible to bind with the current task combination, or when the target layer has a forward task combination of the current task combination, determine whether the target layer is the first layer of the task combination architecture. If it is the first layer, proceed to step 4.1.5. If it is not the first layer, call the task combination binding rule and then proceed to step 4.1.5.

[0140] Task composition binding rules are used to bind task compositions in adjacent layers of a task composition architecture. They include the following steps:

[0141] Step 4.2.1: Generate a similarity matrix A between the target layer and the task combinations in the layer above the target layer. The number of rows in similarity matrix A is the number of task combinations in the target layer, and the number of columns in similarity matrix A is the number of task combinations in the layer above the target layer. The element in the i-th row and j-th column of the similarity matrix is ​​the task combination similarity between the i-th task combination in the target layer and the j-th task combination in the layer above the target layer. If the calculated similarity value of two task combinations is 0, or they cannot be bound, then the element at the corresponding position is 0.

[0142] Assuming there are 4 task combinations in the target layer and 3 task combinations in the layer above the target layer, the generated similarity matrix A will be a 4x3 matrix. The elements of the matrix represent the similarity between task combinations in the target layer and task combinations in the layer above the target layer.

[0143] The similarity of task combinations is calculated as follows: when binding two task combinations is not feasible, the similarity between the two tasks is 0; assuming task combination TG... n+1 Bound to TG n With binding feasibility or task combination TG n Bound to TG n+1 Binding is feasible; obtain TG. n TG n+1 The task array is composed of the number of tasks in the task combination and the number of AGVs required to execute each task. Let's assume the task array is TG. n TG n+1 The constituent arrays are [a bbc] and [bbcd]. Elements with the same constituent array are taken to form a similarity array [bbc]. Task combination TG n With TG n+1 Task combination similarity f s The calculation formula is as follows:

[0144]

[0145] In the formula, w1 and w2 are weighting coefficients, both of which are positive numbers, and w1 + w2 = 1.

[0146] In this implementation case, w1 = w2 = 0.5.

[0147] Step 4.2.2: Determine whether there are non-zero elements in the similarity matrix A. If yes, proceed to step 4.2.3; otherwise, end the task and call the binding rules.

[0148] If there are non-zero elements in the similarity matrix A, proceed to step 4.2.3 to perform task combination binding. If there are no non-zero elements, it means that there are no more task combinations to bind between the task combination of the target layer and the task combination of the layer above the target layer. Therefore, end the call to the task combination binding rule.

[0149] Step 4.2.3: Determine the row number x and column number y of the maximum element, set the xth task combination in the target layer as the follower task combination, set the yth task combination in the layer above the target layer as the target task combination, bind the follower task combination to the target task combination, set the elements in the xth row and yth column of the similarity matrix A to zero, and proceed to step 4.2.4.

[0150] Find the maximum element in the current similarity matrix A. The coordinates of this element correspond to two task combinations. Set the task combination in the target layer as the follower task combination, and set the task combination in the layer above the target layer as the target task combination. After binding the follower task combination to the target task combination, the binding of a set of task combinations in adjacent layers is completed. In this invention, the binding of task combinations is to facilitate the binding between subsequent tasks, and the essence of task binding is task T. i+1 It will inherit the task T it is bound to. i The execution of AGVs, that is, the same AGV after completing task T. i Then continue executing task T i+1 This reduces the time spent on combining and disassembling AGVs. Since a task can only be bound or bound once, a task combination can only be bound or bound once. Therefore, it is necessary to set all elements in the row and column containing the current largest element to zero to prevent a task combination from being bound to multiple task combinations or bound to multiple task combinations.

[0151] Step 4.2.4: Invoke the task binding rules and proceed to step 4.2.2.

[0152] Task binding rules are used to bind tasks between task groups bound in 4.2.3.

[0153] Step 4.1.5: Set the next layer of the target layer as the target layer, put the current task combination into the target layer, and proceed to step 4.1.6;

[0154] Step 4.1.6: Determine whether there are any task combinations in the task combination sequence to be layered. If yes, proceed to step 4.1.2; otherwise, end the call to the task combination layering rule.

[0155] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several improvements without departing from the principle of the present invention, and these improvements should also be considered within the scope of protection of the present invention.

Claims

1. A method for hybrid task allocation in an AGV system applied to an assembly line, characterized in that, Includes the following steps: Step 1: Determine if the task assignment trigger condition is met. If yes, proceed to Step 2; otherwise, proceed to Step 1. Step 2: Determine if there are any material delivery tasks to be assigned. If yes, proceed to Step 3; otherwise, proceed to Step 1. Step 3: Collect the material delivery tasks to be assigned in the system, and divide the material delivery tasks to be assigned into multiple task combinations according to the task combination rules; The task formation rule is to group material delivery tasks with task relevance into the same task group, and material delivery tasks without task relevance into different task groups. The task correlation refers to the task correlation between multiple material delivery tasks responsible for transporting these materials when an assembly operation requires multiple materials. Step 4: Based on the task combination layering rules, form a task combination architecture consisting of multiple task layers. At the same time, based on the task combination binding rules, bind the task combinations located in adjacent layers of the task combination architecture. Step 5: According to the task binding rules, bind the material delivery tasks between the bound task groups to form a task binding relationship, and obtain the bound material delivery task information; Step 6: Determine the priority of task combinations based on the task combination priority determination rules; Step 7: Based on the priority of the determined task combination, generate a mixed task allocation scheme according to the AGV task allocation rules, and allocate the material delivery tasks to be allocated to the required number of AGVs for the task. The task combination hierarchical rule includes the following steps: Step 4.1.1: Sort the task combinations in ascending order according to the end time of the best delivery time period of the task combination to form a sequence of task combinations to be layered. Set the first layer of the task combination architecture as the target layer, take out the first task combination in the sequence and put it into the target layer. Step 4.1.2: Take out the first task combination in the task combination sequence to be layered and set it as the current task combination. Determine whether there is a task combination in the target layer that has binding feasibility with the current task combination. If not, proceed to step 4.1.

3. If it exists, proceed to step 4.1.

4. Step 4.1.3: According to the forward task combination determination rules, determine whether there is a forward task combination of the current task combination in the target layer. If it exists, proceed to step 4.1.4; if it does not exist, put the current task combination into the target layer and proceed to step 4.1.

6. Step 4.1.4: Determine if the target layer is the first layer of the task composition architecture. If yes, proceed to step 4.1.5; otherwise, invoke the task composition binding rule and proceed to step 4.1.

5. Step 4.1.5: Set the next layer of the target layer as the target layer, put the current task combination into the target layer, and proceed to step 4.1.6; Step 4.1.6: Determine whether there are any task combinations in the task combination sequence to be layered. If yes, proceed to step 4.1.2; otherwise, end the call to the task combination layering rule.

2. The hybrid task allocation method for an AGV system applied to an assembly line according to claim 1, characterized in that, The task combination binding rule includes the following steps: Step 4.2.1: Generate a similarity matrix A for the task combinations of the target layer and the layer immediately above the target layer. The number of rows of the similarity matrix A is the number of task combinations in the target layer, and the number of columns of the similarity matrix A is the number of task combinations in the layer immediately above the target layer. The element at the i-th row and j-th column of the similarity matrix is the task combination similarity between the i-th task combination in the target layer and the j-th task combination in the layer immediately above the target layer. If the calculated similarity value of the two task combinations is 0 or binding cannot be performed, the element at the corresponding position is 0; Step 4.2.2: Determine whether there is a non-zero element in the similarity matrix A. If yes, proceed to step 4.2.3; if no, terminate the invocation of the task combination binding rule; Step 4.2.3: Determine the row number x and column number y where the maximum element is located, set the x-th task combination in the target layer as the following task combination, set the y-th task combination in the layer immediately above the target layer as the target task combination, bind the following task combination to the target task combination, then set all elements in the x-th row and the y-th column of the similarity matrix A to zero, and proceed to step 4.2.4; Step 4.2.4: Invoke the task binding rule, and proceed to step 4.2.

2.

3. The method for hybrid task allocation in an AGV system applied to an assembly line according to claim 1, characterized in that, The task binding rule includes the following steps: Step 5.1: Sort the material distribution tasks in the following task combination and the target task combination in descending order according to the number of AGVs required for the material distribution tasks, set the following task as the first material distribution task in the following task combination, and set the target task as the first material distribution task in the target task combination; Step 5.2: Determine the magnitude relationship between the number of required AGVs m1 of the following task and the number of required AGVs m2 of the target task. If m1<m2, proceed to step 5.3; if m1=m2, bind the following task to the target task, and proceed to step 5.4; if m1>m2, terminate the invocation of the task binding rule; Step 5.3: Determine whether there is a next material distribution task in the target task combination. If yes, set the target task as the next material distribution task in the target task combination, and proceed to step 5.2; if no, terminate the invocation of the task binding rule; Step 5.4: Determine whether both the following task combination and the target task combination have a next material distribution task. If yes, set the following task as the next material distribution task in the bound task combination, set the target task as the next material distribution task in the target task combination, and proceed to step 5.2; if no, terminate the invocation of the task binding rule.

4. The hybrid task allocation method for an AGV system applied to an assembly line according to claim 1, characterized in that, The task combination priority determination rule is a principle that task combinations with urgent end times are prioritized.

5. The method for hybrid task allocation in an AGV system applied to an assembly line according to claim 1, characterized in that, The AGV task allocation rule includes the following steps: Step 7.1: Sort the task combinations in descending order according to their priorities to form a task combination sequence; The task combination sequence does not include non-free task combinations; Step 7.2: Take out the first task combination in the sequence and set it as the to-be-allocated task combination; Step 7.3: According to the free task AGV allocation method, sequentially allocate the number of AGVs required by the task to the free tasks in the to-be-allocated task combination; Step 7.4: Determine if there is a follower task combination after the task combination to be assigned. If not, proceed to step 7.

6. If yes, according to the follower task AGV allocation method, allocate the required number of AGVs for the follower tasks in the follower task combination, and proceed to step 7.

5. Step 7.5: Set the follow task group as the task group to be assigned, and then assign the required number of AGVs to the free tasks in the task group to be assigned, and proceed to step 7.4; Step 7.6: Determine if the task combination sequence is empty. If not, proceed to step 7.

2. If yes, end the call to the AGV task allocation rule.

6. The method for hybrid task allocation in an AGV system applied to an assembly line according to claim 1, characterized in that, The rules for determining the feasibility of the binding are as follows: Assume two material delivery tasks T with the same number of AGVs required. i T i+1 They belong to the task combination TG. n TG n+1 If a single AGV or a combination of AGVs is executing T i Then execute T i+1 , will T i+1 If the task is delivered within its optimal delivery time, then the task combination TG is considered to be complete. n+1 Bound to TG n Binding is feasible.

7. The method for hybrid task allocation in an AGV system applied to an assembly line according to claim 1, characterized in that, The forward task combination determination rule is as follows: Assume two material delivery tasks T with the same number of AGVs required. i T i+1 They belong to the task combination TG. n TG n+1 If a single AGV or a combination of AGVs is in T i Complete T at any time during the optimal delivery period i The execution of T, then the execution of T i+1 T i+1 The delivery times are all earlier than T. i+1 The optimal delivery time period starts at the time of task combination TG. n For TG n+1 Forward task combination.

8. A hybrid task allocation method for an AGV system applied to an assembly line according to claim 5, characterized in that, The method for allocating free-task AGVs is as follows: The estimated arrival time of all AGVs in the system to the loading point of the free task is calculated. Assuming that the number of AGVs required for the free task is 'a', the AGV with the earliest estimated arrival time of 'a' is selected as the execution AGV for the free task, and the free task is added to the end of the list of pending tasks of the aforementioned AGV execution 'a'.

9. A hybrid task allocation method for an AGV system applied to an assembly line according to claim 5, characterized in that, The method for assigning AGVs to follow tasks is as follows: Get the AGV to be assigned to execute the task bound to the follow task, assign the follow task to the same AGV, and in the list of tasks to be executed of the assigned AGV, the follow task immediately follows the task bound to it.

Images