AGV scheduling system optimization method based on dynamic priority adjustment and real-time load balance

Abstract

The invention discloses an AGV scheduling system optimization method based on dynamic priority adjustment and real-time load balance, and the method comprises the steps: dynamically monitoring the emergency degree and arrival time of cargos, enabling a system to adjust the task priority in real time, guaranteeing the priority processing of emergency cargos, and giving consideration to the scheduling efficiency of common cargos; information such as current positions, loads and electric quantity of the AGVs is collected, weighted load values are calculated, weight factors are dynamically adjusted according to task types, load balance is achieved, and the situation that part of the AGVs are overloaded and part of the AGVs are insufficient in load is avoided; and finally, scheduling optimization is carried out by comprehensively considering factors such as the adaptation degree between the task and the AGV, the path distance, the predicted completion time and the load balance, and the overall performance of the AGV system is fully exerted. The flexibility, efficiency and resource utilization rate of the AGV scheduling system are effectively improved, and the AGV scheduling system is suitable for various scenes such as logistics storage and production and manufacturing.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent warehousing scheduling technology, and particularly relates to an optimization method for an AGV scheduling system based on dynamic priority adjustment and real-time load balancing. Background Art

[0002] In modern logistics and manufacturing, automated guided vehicles (AGVs) have become key equipment for realizing automated transportation. With the advancement of intelligent manufacturing, the application scope and complexity of AGVs have been continuously increasing. However, existing AGV scheduling systems still have many deficiencies in practical applications and are difficult to meet the complex and changing production logistics requirements.

[0003] Traditional AGV scheduling systems usually adopt fixed priorities or the simple first-come, first-served principle for task allocation. When faced with high loads or changes in task urgency, the flexibility and response speed of this method are significantly insufficient. Fixed priorities cannot dynamically adapt to real-time changing task requirements, resulting in the situation that urgent tasks may not be processed in a timely manner, affecting the overall system efficiency. At the same time, for low-priority tasks, the traditional system fails to effectively identify and handle delays, resulting in some tasks waiting for a long time, affecting the fairness and efficiency of the overall scheduling.

[0004] The problem of load balancing has not been fully solved in the prior art. When AGVs execute tasks, due to uneven load, some AGVs are overused while others are idle, resulting in a decrease in the overall resource utilization rate. In addition, the existing system is not fine enough in managing and scheduling the battery power of AGVs, resulting in the situation that low-battery AGVs may run out of power during the execution of tasks, affecting the continuity and reliability of tasks.

[0005] There is also a problem of insufficient optimization space in the task allocation decision of existing automated guided vehicle (AGV) scheduling systems. When selecting an AGV to execute a task, multiple factors such as the compatibility between the task and the AGV, the path distance, the estimated completion time, and load balancing are not comprehensively considered, resulting in unreasonable task allocation and the inability to fully exert the overall performance of the AGV system. The existence of these problems makes it difficult for existing AGV scheduling systems to achieve efficient, flexible, and stable scheduling goals in the face of complex logistics and production environments.

[0006] For example, Patent CN2018114748444 discloses an AGV dynamic scheduling method, system, device, and storage medium. This technical solution schedules AGVs depending on the on / off status of workstations and the number of queuing positions, without fully considering the urgency of tasks and the arrival time of goods; it inadequately considers the load balance of AGVs, without real-time monitoring of the load and power of AGVs, which may lead to some AGVs being overloaded while others are idle, affecting the overall resource utilization rate; it lacks a mechanism for handling delayed tasks and does not dynamically monitor and increase the priority of low-priority tasks, which may result in long-term backlogs of ordinary goods.

[0007] For example, Patent CN202410173041.4 discloses a method, device, and system for controlling the priority of multi-system AGV scheduling tasks. When controlling priorities, this technical solution mainly relies on pre-set task priority factors and has limited real-time dynamic adjustment capabilities; it does not comprehensively consider the load balance and power management of AGVs, lacking real-time monitoring and dynamic adjustment mechanisms, which will result in uneven AGV loads; task execution is rigid, and the lower-level execution system executes tasks in the order of task generation time, without dynamically optimizing the task allocation path in combination with the real-time status of AGVs. Summary of the Invention

[0008] Object of the Invention: The object of the present invention is to solve the deficiencies existing in the prior art and provide an optimization method for an AGV scheduling system based on dynamic priority adjustment and real-time load balance.

[0009] Technical Solution: An optimization method for an AGV scheduling system based on dynamic priority adjustment and real-time load balance of the present invention includes the following steps:

[0010] Step 1. In the dynamic priority adjustment stage, by continuously monitoring the upper-level system (here, the upper-level system includes a platform management system, a WMS system, an MES system, etc., and the upper-level system submits task orders to the AGV scheduling system, and the AGV scheduling is responsible for allocation), perform dynamic scheduling on ordinary goods based on the first-come-first-allocate principle, while for urgent goods, adjust the priority according to the urgency and arrival time (urgent goods are sent by the upper-level system to issue an urgent instruction to the AGV scheduling system); the specific method is as follows:

[0011] Step 1.1. The scheduling system detects whether the upper-level system sends an urgent goods instruction.

[0012] Step 1.2. If no urgent goods instruction is detected, perform ordinary goods allocation, that is, adopt a queue data structure and implement the first-come-first-allocate principle. The task at the head of the queue has the highest priority level, and so on, and queue priorities allow for queue jumping; the priority allocation formula:

[0013]

[0014] Where P is the priority of the goods, the larger P is, the higher the priority is. E is the quantified value of the urgency level and is given by the upper-level system. T is the arrival time of the goods, and T max is the maximum allowable waiting time, ω 1 , ω 2 is the weight coefficient and satisfies ω 1 + ω 2 = 1, ensuring that the value of the priority P is between 0 and 1;

[0015] For ordinary goods, ω 1 = ω 2 , and E = 0. The ordinary priority is determined according to the size of the arrival time T, and the smallest T has the highest priority; ω 2 is related to the ordinary priority of the goods (related to the size of the arrival time T). For the same batch of ordinary goods, ω 2 is a fixed value; ω 1 is related to the urgency level of the goods. Values of ω 1 are assigned to three types of urgent goods;

[0016] Step 1.3: If an urgent goods instruction sent by the upper-level system is detected, first assign ω 1 , ω 2 to it according to the urgency level of the goods to improve the priority P of the goods: If multiple urgent goods instructions are detected at the same time, then according to the arrival time T of the goods and E provided by the upper-level system, uniformly adjust ω 1 , ω 2 , and make the priorities P of urgent goods be reasonably assigned according to the formula;

[0017] Finally, insert the adjusted urgent goods priority sequence in front of the original queue to implement goods allocation;

[0018] Step 2: In the stage of reallocating low-priority tasks, adjust the low-priority queue threshold according to the goods allocation situation, and improve the priorities of the goods with delay situations; The specific method is:

[0019] First, collect the waiting time data of all ordinary goods within a certain period, calculate the coefficients of all waiting times, then judge the low-priority goods sequence in the current task processing queue, and then dynamically calculate the response ratios of all low-priority tasks. When the response ratio of a certain low-priority task reaches or exceeds the preset threshold, automatically promote the priority of this task to the urgent task queue, but do not preempt the priority of actual urgent goods; Then, reorder the task queue according to the adjusted priority queue;

[0020] Step 3: In the load balancing stage, collect the current positions of each AGV, calculate the weighted load value of each AGV according to factors such as the maximum load capacity, current load, and battery power of the AGV, and at the same time perform special instruction processing on the AGVs with low battery power;

[0021] Step 4: In the dispatching and general control stage, combine the results of the real-time load balancing module and the dynamic priority adjustment module to generate the final task allocation plan, and control the AGV to execute the corresponding task. The specific method is as follows:

[0022] Step 4.1: Allocate AGVs according to the task queue sorted in the above steps, and adopt a greedy selection strategy in combination with the weighted load value of the AGV calculated in Step 3;

[0023] Step 4.2: Each decision is based on the following principles: Principle 1): The current task preferentially selects the AGV that can complete it the fastest; Principle 2): The current AGV preferentially selects the goods with the shortest path or the lowest cost;

[0024] Adopt the following formula:

[0025]

[0026] Where Score is the fitness score of the task and the AGV (the higher the better), t represents the time required for the AGV to complete the current task (the smaller the better), WeightedLoad represents the weighted load value of the AGV (i.e., AGV load balancing), ω 3 , ω 4 , ω 5 Weight coefficients, satisfying ω 3 + ω 4 + ω 5 = 1;

[0027] Step 4.3: When an emergency priority cut-in occurs during the AGV's journey to the goods, if the dispatching system considers that the current task has not started, it interrupts the ordinary task and returns to Step 4.1 to match the optimal AGV for the emergency task. If the current task has started, the dispatching system reallocates the AGV at the starting point.

[0028] Furthermore, after detecting the emergency goods instruction in Step 1, the allocation rules of ω 1 , ω 2 are as follows:

[0029] If the emergency level E is 1, the weight coefficient ω 1 is 0.25, and the weight coefficient ω 2 is 0.75;

[0030] If the emergency level E is 2, the weight coefficient ω 1 is 0.5, and the weight coefficient ω 2 is 0.5;

[0031] If the urgency level E is 3, the weight coefficient ω 1 is 0.75, and the weight coefficient ω 2 is 0.25.

[0032] Furthermore, during the reallocation process of low-priority tasks in step 2, collect the waiting time data of all general goods within a certain period, calculate the average value μ and standard deviation σ of these waiting times, and select the coefficient k according to the actual situation; k is used to adjust the contribution weight of the standard deviation to the threshold and determine the sensitivity of the system. If the business requires quick response to delayed tasks (such as an e-commerce warehouse), the initial k can be set to a lower value (such as 0.5 - 1); if it is necessary to prioritize high-priority tasks (such as medical supplies scheduling), the initial k can be set to a higher value (such as 1.5 - 2). When the load is high (such as AGV utilization rate > 80%): reduce k to allow more low-priority tasks to increase their priorities and relieve the backlog; when the load is low (such as AGV utilization rate < 50%): increase k to reduce the frequency of priority increases and avoid resource waste;

[0033] The threshold determination formula for the low-priority queue is:

[0034] T threshold = μ + kσ

[0035] Calculate the response ratio for the delayed goods of all low-priority tasks. The calculation formula for the response ratio RR is:

[0036]

[0037] where T w is the waiting time, and T s is the expected service time of the AGV, which is updated according to the real-time position. The calculation formula is:

[0038] T s = h j + r j + L s (S 1 + S 2 )

[0039] h j is the loading and carrying time of the AGV for the jth task, r j is the empty running time of the AGV for the jth task, S 1 , S 2 respectively represent the number of stop-and-wait times of the AGV in the empty and loaded states, and L s is the single stop-and-avoidance time:

[0040] During the task execution process, the response ratio of low-priority tasks is updated in real time to ensure that delayed cargo tasks are scheduled in a timely manner while meeting the response ratio priority rule.

[0041] Furthermore, the specific process of calculating the weighted load value in step 3 is as follows:

[0042] Step 3.1: Collect AGV status: information such as the current location, current load, battery level, maximum load capacity, full battery, etc. The AGV scheduling system background monitors these status information at all times, and the upper-level system mainly provides cargo orders and issues emergency cargo instructions;

[0043] Step 3.2: Traverse the task target point locations and task priorities;

[0044] Step 3.3: Initialize the weight factors (α, β, γ) and dynamically set the initial values according to the scenario, calculate the weighted load values of each AGV and record them. The specific implementation is as follows:

[0045]

[0046] D is the distance between the current location of the AGV and the target task point, WL n represents the current load of the AGV, WL max represents the maximum load capacity of the AGV, Q n is the current battery level of the AGV, and Q is the full battery level;

[0047] Then, calculate the weighted load value for each candidate AGV in turn, where the weights (α, β, γ) are dynamically adjusted according to the task type and task priority:

[0048] For emergency tasks: reduce the distance weight (α) and increase the load weight (β); for ordinary tasks:

[0049] Increase the battery level weight (γ) and preferentially allocate AGVs with high battery levels;

[0050] Step 3.4: Record the weighted load values of each AGV in real time for subsequent task allocation decisions.

[0051] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0052] (1) Dynamically adjust the task priority according to the urgency and arrival time of the goods to ensure that emergency goods can be processed in a timely manner, while taking into account the scheduling efficiency of ordinary goods, avoiding task backlogs and delays. Optimize the scheduling mechanism for low-priority tasks, which can dynamically adjust their priorities according to the delay situation of the tasks to ensure that delayed tasks are processed in a timely manner, enhancing the stability and reliability of the system.

[0053] (2) By monitoring various factors such as the current load, battery level, and maximum load capacity of AGVs in real time, situations where some AGVs are overloaded while others are underloaded are avoided, improving the overall resource utilization rate and achieving load balance. In terms of task allocation decisions, the present invention comprehensively considers various factors such as the compatibility between tasks and AGVs, path distance, estimated completion time, and load balance, optimizing the scheduling strategy to ensure more reasonable task allocation and giving full play to the overall performance of the AGV system.

[0054] In summary, the present invention has made remarkable progress in improving the flexibility, efficiency, and resource utilization rate of the AGV scheduling system, solving problems existing in the prior art such as unreasonable priority allocation, insufficient load balancing ability, inflexible scheduling of low-priority tasks, and suboptimal task allocation decisions. The technical solution proposed by the present invention is more conducive to use in complex industrial environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a schematic framework diagram of the present invention;

[0056] Figure 2 is a schematic overall process diagram of the present invention;

[0057] Figure 3 is a schematic diagram of the application scenario simulation in the embodiment;

[0058] Figure 4 is a schematic diagram of the corresponding scenario and partial operation results in the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0059] The technical solution of the present invention will be described in detail below, but the protection scope of the present invention is not limited to the described embodiments.

[0060] As Figure 1 and Figure 2 shown, an optimization method for an AGV scheduling system based on dynamic priority adjustment and real-time load balancing of the present invention includes the following steps:

[0061] Step 1. In the dynamic priority adjustment stage, by continuously monitoring the upper-level system, dynamic scheduling is performed on ordinary goods based on the principle of first-come, first-served allocation, while the priority of emergency goods is adjusted according to the urgency level and arrival time; the specific method is as follows:

[0062] Step 1.1. The scheduling system detects whether the upper-level system sends an emergency goods instruction;

[0063] Step 1.2. If no emergency goods instruction is detected, ordinary goods allocation is implemented, that is, a queue data structure is adopted, and the principle of first-come, first-served allocation is implemented. The task at the head of the queue has the highest priority level, and so on, and queue priority allows for queue jumping; the priority allocation formula:

[0064]

[0065] Where P is the priority of the goods, the larger P is, the higher the priority. E is the quantified value of the urgency level, T is the arrival time of the goods, and T max is the maximum allowable waiting time, ω 1 , ω 2 is the weight coefficient and satisfies ω 1 + ω 2 = 1, ensuring that the value of the priority P is between 0 and 1;

[0066] For ordinary goods, ω 1 = ω 2 , and E = 0. The ordinary priority is determined according to the size of the arrival time T, and the smallest T has the highest priority;

[0067] Step 1.3: If an emergency goods instruction sent by the upper system is detected, first assign ω 1 , ω 2 for it according to the urgency level of the goods to improve the priority P of the goods: If multiple emergency goods instructions are detected at the same time, then according to the arrival time T of the goods and E provided by the upper system, uniformly adjust ω 1 , ω 2 , and then reasonably allocate the priority P among the emergency goods;

[0068] Finally, insert the adjusted emergency goods priority sequence in front of the original queue to implement goods allocation;

[0069] Step 2: In the stage of reallocating low-priority tasks, adjust the low-priority queue threshold according to the goods allocation situation, and improve the priority of the goods with delay; The specific method is:

[0070] First, collect the waiting time data of all ordinary goods within a certain period, calculate the coefficient of all waiting times, then judge the low-priority goods sequence in the current task processing queue, and then dynamically calculate the response ratio of all low-priority tasks. When the response ratio of a certain low-priority task reaches or exceeds the preset threshold, automatically promote the priority of this task to the emergency task queue, but do not preempt the priority of actual emergency goods; Then reorder the task queue according to the adjusted priority queue;

[0071] Step 3: In the load balancing stage, collect the current positions of each automatic guided vehicle (AGV), calculate the weighted load value of each AGV according to factors such as the maximum load capacity, current load, and battery power of the AGV, and at the same time perform special instruction processing on the AGV with low battery;

[0072] Step 4. In the overall scheduling and control phase, combine the results of the real-time load balancing module and the dynamic priority adjustment module to generate the final task allocation plan, and control the AGV to execute the corresponding tasks. The specific method is as follows:

[0073] Step 4.1. Allocate AGVs according to the task queue sorted in the above steps, and adopt a greedy selection strategy in combination with the weighted load values of AGVs calculated in Step 3.

[0074] Step 4.2. Each decision is based on the following principles: Principle 1: The current task preferentially selects the AGV that can complete it fastest; Principle 2: The current AGV preferentially selects the goods with the shortest path or the lowest cost.

[0075] The following formula is adopted:

[0076]

[0077] where Score is the fitness score between the task and the AGV, t represents the time required for the AGV to complete the current task, WeightedLoad represents the weighted load value of the AGV, ω 3 , ω 4 , ω 5 is the weight coefficient, satisfying ω 3 + ω 4 + ω 5 = 1;

[0078] Step 4.3. When an emergency priority queue jumps in during the process of the AGV going to pick up goods, if the scheduling system considers that the current task has not started, it interrupts the normal task and returns to Step 4.1 to match the optimal AGV for the emergency task. If the current task has started, the scheduling system reallocates the AGV at the starting point.

[0079] Embodiment

[0080] To verify the technical effects and feasibility of the present invention, in this embodiment, the technical solution is applied to a medical e-commerce warehouse that needs to sort 500 ordinary goods and 15 emergency goods, and the number of AGVs is 10. Figure 3 It is a schematic diagram of the simulation of the shipping effect for the scenario.

[0081] Step (1). In the dynamic priority adjustment phase, by continuously monitoring the upper system, perform dynamic scheduling on ordinary goods based on the principle of first come, first served.

[0082] Step (1-1). The scheduling system detects the upper system and detects 15 emergency task instructions (quantify their emergency level E = 3).

[0083] Step (1-2): For the general cargo queuing area 1, a queue data structure is adopted, and the principle of first come, first served is implemented. The task at the head of the queue has the highest priority, and so on. And queue priority allows queue jumping;

[0084] Step (1-3): Allocate ω for the goods according to the urgency of the goods 1 , ω 2 , and increase the priority P of the goods. The allocation rules are as follows:

[0085] Emergency level E <![CDATA[Weight coefficient ω 1 > <![CDATA[Weight coefficient ω 2 > 1 0.25 0.75 2 0.5 0.5 3 0.75 0.25

[0086] After the injection of urgent tasks, it is dynamically adjusted according to the priority allocation formula:

[0087]

[0088] The priority of the urgent task P = 0.9, insert it at the front of the queue and place it in the urgent cargo queuing area 3;

[0089] Step (1-4): Multiple urgent tasks are sorted in ascending order according to the arrival time to ensure that the earliest arrival is given priority for allocation;

[0090] Step (2): In the stage of reallocating low-priority tasks, adjust the low-priority queue threshold according to the cargo allocation situation, and increase the priority of the goods with delay;

[0091] Step (3): In the load balancing stage, collect the current positions of each AGV, calculate the weighted load value of each AGV according to factors such as the maximum load capacity, current load, and battery power of the AGV, and at the same time perform special instruction processing on the AGV with low battery;

[0092] Step (4): In the overall scheduling control stage, combine the results of the real-time load balancing module and the dynamic priority adjustment module to generate the final task allocation plan and control the AGV to execute the corresponding tasks.

[0093] The specific process of scheduling the delayed goods of the above low-priority tasks is as follows:

[0094] Step (2-1): Collect the waiting time data of all general goods within a certain period, calculate the average value μ = 12 and the standard deviation σ = 3 of the waiting time, and select the coefficient k = 2 according to the actual situation;

[0095] Step (2-2): Judge the low-priority cargo sequence in the current task processing queue, and calculate the result according to the low-priority queue threshold judgment formula:

[0096] T threshold = μ + kσ = 18min

[0097] Step (2-3): Calculate the response ratio for the delayed goods of all low-priority tasks. The calculation formula for the response ratio RR is as follows:

[0098]

[0099] where T w is the waiting time, and T s is the estimated service time of the AGV, which is updated according to the real-time position. The calculation formula is:

[0100] T s = h j + r j + L s (S 1 + S 2 )

[0101] h j is the load-carrying time of the AGV for executing the jth task, r j is the no-load running time of the AGV for executing the jth task, S 1 , S 2 respectively represent the number of stop waiting times of the AGV in no-load and load-carrying states, and L s is the single stop avoidance time;

[0102] Step (2-4): Dynamically calculate the response ratio of all low-priority tasks. When RR≥2.0, promote the task to the end of the emergency queue;

[0103] Step (2-5): At this time, it is detected that the response ratio of 1 ordinary task meets the standard, which is the low-priority delayed goods 2. Promote it to the end of the emergency queue and reorder the task queue;

[0104] Step (2-6): During the task execution process, update the response ratio of the low-priority tasks in real time.

[0105] The specific process of the above weighted load value calculation is as follows:

[0106] Step (3-1): Collect the AGV status every 10 seconds: current position, current load, power, maximum load capacity, full power, etc. information;

[0107] Step (3-2): Traverse the task target point positions and task priorities;

[0108] Step (3-3): Initialize the weight factors (α, β, γ) and dynamically set the initial values according to the scenario;

[0109] Step (3-4): Calculate and record the weighted load values of each AGV. The specific implementation is:

[0110]

[0111] D is the distance between the current position of the AGV and the target task point, WL n represents the current load of the AGV, WL max represents the maximum load capacity of the AGV, Q n is the current battery level of the AGV, and Q is the full battery level;

[0112] Calculate the comprehensive load value for each candidate AGV in sequence. When Q n < 25%, it is marked as a low-battery AGV, the new task is suspended and it is guided to the charging station;

[0113] a. Emergency task: Reduce the distance weight (α = 0.2), increase the load weight (β = 0.6);

[0114] b. Ordinary task: Increase the battery level weight (γ = 0.5), and preferentially allocate high-battery-level AGVs;

[0115] In step (3-5), record the weighted load value of each AGV in real time for subsequent task allocation decisions.

[0116] In the scheduling and overall control stage of this embodiment, the task allocation decision method combining dynamic priority adjustment and real-time load balancing is as follows:

[0117] Allocate AGVs from the task queue sorted by step (1) and step (2), and adopt the greedy selection strategy in combination with the weighted load value of the AGV calculated in step (3);

[0118] Adopt the following formula:

[0119]

[0120] Calculate the scores of each AGV, arrange tasks for it and plan the path. As Figure 3 shown, when it is detected that an emergency task cuts in line, AGV-4 (Score = 1.2) has the highest score and the current task has not started, so it is preferentially allocated to the emergency task.

[0121] From the above experimental results, it can be seen that in terms of the dynamic priority adjustment mechanism of the present invention, it monitors the urgency of goods in real time (such as the priority of medical supplies is higher than that of ordinary goods) and the arrival time, and dynamically adjusts the task queue, rather than relying on fixed rules; in terms of AGV load balancing, it calculates the weighted load value of the AGV, comprehensively considers parameters such as distance, current load, and power, and dynamically allocates tasks. For example, AGVs with high power give priority to executing ordinary tasks, and AGVs with low power are guided to charge, while the other two patents do not manage the power finely; in terms of delaying the processing of low-priority tasks, it dynamically adjusts their priorities according to the delay situation of the tasks to ensure that the delayed tasks are processed in a timely manner. When the waiting time of a certain task exceeds the threshold, it is automatically upgraded to the emergency queue to avoid long-term backlog; at the same time, the present invention updates the AGV status (such as location, power) and task response ratio in real time during the task execution process, supporting dynamic reordering of the queue. For example, when an emergency task jumps the queue, the unstarted ordinary tasks are immediately interrupted, while in the other two patents, it is necessary to wait for the current task to be completed.

Claims

1. An AGV scheduling system optimization method based on dynamic priority adjustment and real-time load balancing, characterized in that: The following steps are involved: Step 1: In the dynamic priority adjustment stage, by continuously monitoring the upper system, ordinary goods are dynamically scheduled based on the first-come-first-served principle, while urgent goods are prioritized according to the urgency and arrival time; the specific method is as follows: Step 1.1, the dispatching system detects whether the upper system sends an emergency cargo instruction; Step 1.2: If no urgent cargo instruction is detected, ordinary cargo distribution is implemented, that is, the queue data structure is adopted, and the first-come-first-served principle is implemented. The task at the head of the queue has the highest priority, and so on, and the queue priority allows queue jumping; priority distribution formula: Where P is the priority of the goods. The larger the P is, the higher the priority is. E is the quantitative value of the urgency. T is the arrival time of the goods. max is the maximum allowed waiting time, ω1, ω2 are weight coefficients and satisfy ω1+ω2=1, ensuring that the value of priority P is between 0 and 1; For ordinary goods, ω1=ω2, and E=0, the ordinary priority is determined according to the arrival time T, and the highest priority has the smallest T; Step 1.3: If an urgent cargo instruction sent by the upper system is detected, ω1 and ω2 are first assigned to the cargo according to the urgency of the cargo, and the priority P of the cargo is increased; if multiple urgent cargo instructions are detected at the same time, ω1 and ω2 are uniformly adjusted according to the cargo arrival time T and E provided by the upper system, and then the priority P between the urgent cargoes is reasonably assigned; Finally, the adjusted priority sequence of urgent goods is inserted in front of the original queue to implement goods distribution; Step 2: During the low-priority task reallocation phase, adjust the low-priority queue threshold according to the cargo allocation situation, and increase the priority of delayed cargo. The specific method is as follows: First, the waiting time data of all ordinary goods in a certain period is collected, and the coefficients of all waiting times are calculated. Then, the sequence of low-priority goods in the current task processing queue is determined, and then the response ratio of all low-priority tasks is dynamically calculated. When the response ratio of a low-priority task reaches or exceeds the preset threshold, the priority of the task is automatically raised to the emergency task queue, but the priority of the actual emergency goods is not preempted; then, the task queue is reordered according to the adjusted priority queue; Step 3: In the load balancing stage, the current position of each AGV is collected, and the weighted load value of each AGV is calculated based on factors such as the maximum carrying capacity, current load, and power of the AGV. At the same time, special instructions are processed for the AGV with low power; Step 4: In the overall scheduling control stage, the results of the real-time load balancing module and the dynamic priority adjustment module are combined to generate the final task allocation plan, and the AGV is controlled to perform the corresponding tasks. The specific method is as follows: Step 4.1, assign AGVs to the task queues sorted in the above steps, and adopt a greedy selection strategy based on the obtained AGV weighted load value; Step 4.2, each decision is based on the following principles: Principle 1), the current task is given priority to the AGV that can complete the task the fastest; Principle 2), the current AGV gives priority to the goods with the shortest path or the lowest cost; Use the following formula: Where Score is the fit score between the task and the AGV, t is the time the AGV is expected to take to complete the current task, WeightedLoad is the weighted load value of the AGV, and the weight coefficients of ω3, ω4, and ω5 satisfy ω3+ω4+ω5=1; Step 4.3: When an emergency priority queue jump occurs when the AGV is heading to the cargo, the scheduling system considers that the current task has not started, so it interrupts the ordinary task and returns to step 4.1 to match the optimal AGV for the emergency task. If the current task has started, the scheduling system reallocates the AGV at the starting point.

2. The AGV scheduling system optimization method based on dynamic priority adjustment and real-time load balancing according to claim 1 is characterized in that: After the urgent cargo order is detected in step 1, the allocation rule of ω1 and ω2 is as follows: If the urgency E is 1, the weight coefficient ω1 is 0.25 and the weight coefficient ω2 is 0.75; If the urgency E is 2, the weight coefficient ω1 is 0.5 and the weight coefficient ω2 is 0.5; If the urgency level E is 3, the weight coefficient ω1 is 0.75 and the weight coefficient ω2 is 0.

25.

3. The AGV scheduling system optimization method based on dynamic priority adjustment and real-time load balancing according to claim 1 is characterized in that: In the process of low priority task reallocation in step 2, the low priority queue threshold determination formula is: T threshold =μ+kσ Calculate the response ratio of all delayed goods of low priority tasks. The calculation formula of response ratio RR is: Where T w is the waiting time, T s is the estimated service time of the AGV, which is updated according to the real-time location. s The calculation formula is: T s =h j +r j +L s (S1+S2) h j is the time it takes for the AGV to perform the jth task of loading and unloading goods, r j is the idle running time of the AGV when executing the jth task, S1 and S2 represent the number of parking waiting times of the AGV when it is empty and loaded, respectively, and L s It is the single stop avoidance time; During the task execution process, the response ratio of low-priority tasks is updated in real time to ensure that delayed cargo tasks are scheduled in a timely manner while meeting the response ratio priority rules.

4. The AGV scheduling system optimization method based on dynamic priority adjustment and real-time load balancing according to claim 1 is characterized in that: The specific process of calculating the weighted load value in step 3 is as follows: Step 3.1, collect AGV status, including current position, current load, power, maximum load capacity and full power; Step 3.2, traverse the task target point location and task priority; Step 3.3: Initialize the weight factors (α, β, γ) and dynamically set the initial values ​​according to the situation. Calculate and record the weighted load values ​​of each AGV. The specific implementation is as follows: D is the distance between the current position of AGV and the target task point, WL n Indicates the current load of AGV, WL max Indicates the maximum carrying capacity of AGV, Q n is the current power of AGV, Q is the full power; Then the weighted load value is calculated for each candidate AGV in turn, where the weights (α, β, γ) are dynamically adjusted according to the task type and task priority: For urgent tasks: reduce the distance weight α and increase the load weight β; For common tasks: increase the power weight γ and give priority to high-power AGVs; Step 3.4: Record the weighted load value of each AGV in real time for subsequent task allocation decisions.

Images