Multi-warehouse collaborative dynamic optimization method adaptive to multi-variety variable-batch mixed-line production

Abstract

The invention discloses a multi-warehouse collaborative dynamic optimization method adaptive to multi-variety variable-batch mixed line production, and relates to the technical field of logistics distribution optimization, and the method comprises the steps: carrying out the modeling of a multi-warehouse scene analysis problem, and carrying out the analysis of a multi-warehouse scene; performing multi-bin material state correlation analysis, and based on the multi-bin scene model, constructing a correlation relationship among multiple bins; dynamically optimizing a multi-bin collaborative seasoning threshold value, constructing an environment comprising a workshop, an inner bin and an outer bin, and solving a multi-bin collaborative optimal seasoning threshold value; a production logistics collaboration mechanism: constructing a production logistics collaboration double-layer optimization model, and realizing deep collaboration of production and logistics; and production logistics collaborative scheduling optimization is carried out, a multi-objective optimization algorithm and a production logistics collaborative mechanism are combined, minimization of maximum completion time and production waiting time, maximization of task income and minimization of material waiting time are taken as objectives, and through bidirectional feedback of production and logistics states, distribution priority and production scheduling are optimized. And the production income is evaluated from production and material levels.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of manufacturing service scheduling in a service-oriented manufacturing system, and particularly relates to a multi-bin collaborative dynamic optimization method suitable for multi-variety variable-batch mixed-line production. BACKGROUND

[0002] In the field of 3C manufacturing, the promotion and popularization of intelligent manufacturing put forward higher requirements on the production and logistics efficiency of the workshop. There are many problems in the production and logistics management of the current 3C manufacturing workshop that need to be solved.

[0003] In the multi-bin collaborative scheduling scenario, the products in the 3C workshop have the characteristics of multi-variety, small batch, and high frequency changeover. The traditional single-bin management mode and fixed threshold setting method have been difficult to adapt. In the production of the workshop, the multi-bin material adjustment is not collaborative, the information sharing between the inside and outside bins is insufficient, and the correlation is not clear, which leads to unreasonable call and adjustment threshold setting, and easily causes production line shutdown, production interruption or bin explosion, which seriously affects the production efficiency.

[0004] In terms of production and logistics collaboration, most existing researches model the production process and the logistics process independently, making it difficult for production and logistics to effectively collaborate. The 3C workshop production is dynamic and strongly correlated, and the material demand of each section is significantly different. The traditional scheduling strategy does not fully exploit the potential correlation of production and logistics data, making it difficult to cope with demand fluctuations caused by high-frequency changeover, often resulting in problems such as "line waiting for material" or "material accumulation", leading to low production efficiency and high inventory cost.

[0005] For high-frequency changeover scenarios, the state of the 3C workshop changes dynamically in real time, and resource constraints require some production tasks to change lines to improve efficiency. However, existing methods lack accurate prediction and active scheduling of changeover demand, making it difficult to reasonably arrange changeover scheduling and coordinate material transportation in a highly dynamic environment, and unable to meet complex production demands.

[0006] To solve the above problems, a multi-scenario fleet detection and scheduling algorithm testing and recommendation method and system are needed, which includes multi-bin scenario analysis problem modeling, multi-bin material state correlation analysis, multi-bin collaborative material threshold dynamic optimization, production and logistics collaboration mechanism, and production and logistics collaborative scheduling optimization. By analyzing the multi-bin collaborative scenario, constructing the multi-bin collaborative material correlation, and then realizing the multi-bin collaborative dynamic material threshold optimization, based on the threshold optimization of the multi-bin collaboration, further constructing the production and logistics collaboration mechanism, and combining with the multi-objective optimization algorithm, realizing the scheduling optimization of the production and logistics collaboration. SUMMARY

[0007] To solve the above technical problems, the application provides a multi-bin cooperative dynamic optimization method suitable for multi-variety variable batch mixed line production, which comprises five steps of multi-bin scene analysis problem modeling, multi-bin material state correlation analysis, multi-bin cooperative material adjustment threshold dynamic optimization, production logistics cooperative mechanism and production logistics cooperative scheduling optimization.

[0008] The application solves the technical problems by adopting the following technical solutions:

[0009] A multi-bin cooperative dynamic optimization method suitable for multi-variety variable batch mixed line production comprises the following steps:

[0010] Step 1. Abstract the workshop, inner bin and outer bin as a scene model, set the workshop capacity, task type and quantity, arrival and deadline time, the demand of each task for each type of material in the workshop model, and make the material remaining amount non-negative to establish the mapping relationship between the task quantity and the material demand; in the inner bin model, set the capacity, the quantity of each type of material, the consumption and the replenishment amount, and keep the material quantity non-negative; in the outer bin model, set the capacity, the quantity of each type of material, the consumption and the replenishment amount, and keep the material quantity non-negative; set the call threshold and the material adjustment threshold in real time, which are used to trigger the replenishment chain of the workshop, the inner bin and the outer bin;

[0011] Step 2. Multi-bin material state correlation analysis, construct the transportation state quantity of the workshop, the inner bin and the outer bin, and calculate the transportation coefficient according to the distance and the transportation speed; calculate the inner bin-outer bin correlation coefficient and the outer bin-outer bin correlation coefficient according to the storage of each type of material among the entities; fuse the transportation coefficient and the correlation coefficient to calculate the transportation correlation cooperative coefficient;

[0012] Step 3. Multi-bin cooperative material adjustment threshold dynamic optimization, build an environment including the workshop, the inner bin and the outer bin, based on the evaluation of the task demand state, the warehouse material state and the multi-bin cooperative state, configure the objective function, the constraint condition and the optimization target of the optimization problem, and solve the multi-bin cooperative optimal material adjustment threshold;

[0013] Step 4. Production logistics coordination mechanism, build production logistics coordination double-layer optimization model, in the production logistics coordination double-layer optimization model, the upper production scheduling optimization layer integrates the task and the section model to generate the production plan and receives the coordination layer constraint feedback, the coordination layer coordinates the task, the section, the logistics and the material, and the production waiting, the task benefit each coordination target is clear and the state update is transmitted, and the lower logistics distribution optimization layer realizes distribution optimization through the logistics and the material model, while setting the internal warehouse inventory, the section sequence each constraint condition, so that the total waiting time is minimum, and the task benefit is maximum, so that the production and the logistics are deeply coordinated;

[0014] Step 5. Production logistics coordination scheduling optimization, combining a multi-objective optimization algorithm and a production logistics coordination mechanism, the upper production scheduling optimization layer takes minimizing the maximum completion time, the production waiting time and maximizing the task benefit as the target, and the lower logistics distribution optimization layer takes minimizing the material waiting time as the target, and realizes the bidirectional feedback of the production and the logistics state through the coordination layer, so as to optimize the distribution priority and the production scheduling.

[0015] The present application has the following beneficial effects:

[0016] (1) The present application realizes the adaptive adjustment of the internal and external warehouse material adjustment threshold value through the construction of a multi-warehouse coordination multi-level dynamic model, the fusion of workshop and warehouse state evaluation and the reinforcement learning algorithm. Different from the traditional fixed threshold mode, the mechanism can dynamically optimize the material calling and replenishment threshold according to the real-time production state (such as material consumption rate, warehouse capacity), avoid material breakage or warehouse explosion, and realize the coordinated scheduling of multi-warehouse materials.

[0017] The present application proposes a production-logistics double-layer coordination optimization architecture: the upper layer constructs a task-section matching model to optimize scheduling, and the lower layer establishes a material-task matching model to dynamically schedule distribution. Through the improvement of a multi-objective algorithm, the multi-objective Pareto optimal solution search is realized, the limitation of traditional production and logistics independent modeling is broken through, and the coordination failure problem of "production line waiting for material" and "material accumulation" is solved. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is the flowchart of the present application. DETAILED DESCRIPTION

[0019] The present application will be further described in detail below in combination with the drawings.

[0020] The present application discloses a multi-warehouse coordination dynamic optimization method suitable for multi-variety variable batch mixed line production, such as Figure 1As shown, the method comprises five steps of multi-bin scene analysis problem modeling, multi-bin material state correlation analysis, multi-bin collaborative material adjustment threshold dynamic optimization, production logistics collaborative mechanism and production logistics collaborative scheduling optimization. The present application can analyze the multi-bin collaborative scene, build the multi-bin collaborative material correlation, and then realize the dynamic material threshold optimization of multi-bin collaboration. Based on the threshold optimization of multi-bin collaboration, the production logistics collaborative mechanism is further constructed, and the multi-objective optimization algorithm is combined to realize the scheduling optimization of production logistics collaboration. The specific implementation is as follows:

[0021] Step 1. The workshop, inner bin and outer bin are abstracted as a scene model, the workshop model is set to have a workshop capacity, a task type and quantity, an arrival and deadline time, a demand for each task for each type of material, and a non-negative material remaining amount, a mapping relationship between the task quantity and the material demand is established; in the inner bin model, the capacity, the quantity of each type of material, the consumption and the replenishment are set, and the material quantity is kept non-negative; in the outer bin model, the capacity, the quantity of each type of material, the consumption and the replenishment are set, and the material quantity is kept non-negative; the call threshold and the material adjustment threshold are set in real time, which are used to trigger the material replenishment chain of the workshop, the inner bin and the outer bin;

[0022] Step 2. Multi-bin material state correlation analysis, the transportation state quantity of the workshop, the inner bin and the outer bin is constructed, and the transportation coefficient is calculated according to the distance and the transportation speed; the inner bin-outer bin correlation coefficient and the outer bin-outer bin correlation coefficient are calculated according to the storage of each type of material in each entity; the transportation correlation coefficient is calculated by fusing the transportation coefficient and the correlation coefficient;

[0023] Step 3. Multi-bin collaborative material adjustment threshold dynamic optimization, an environment including the workshop, the inner bin and the outer bin is constructed, based on the evaluation of the task demand state, the bin material state and the multi-bin collaborative state, the objective function, the constraint condition and the optimization target of the optimization problem are configured, and the optimal multi-bin collaborative material adjustment threshold is solved;

[0024] Step 4. Production logistics collaborative mechanism, a production logistics collaborative double-layer optimization model is constructed, in the production logistics collaborative double-layer optimization model, the upper production scheduling optimization layer integrates the task and the section model to generate a production plan and receives the constraint feedback of the collaborative layer, the collaborative layer coordinates the task, the section, the logistics and the material, and clearly defines the production waiting, the task benefit and each collaborative target and transmits the state update, the lower logistics distribution optimization layer realizes the distribution optimization through the logistics and the material model, and sets the inner bin inventory, the section sequence and each constraint condition, so as to realize the deep collaboration of production and logistics with the optimization targets of minimum total waiting time and maximum task benefit;

[0025] Step 5. Production flow collaborative scheduling optimization, combined with multi-objective optimization algorithm and production flow collaborative mechanism, the upper production scheduling optimization layer aims to minimize the maximum completion time, production waiting time and maximize task revenue, the lower logistics distribution optimization layer aims to minimize the material waiting time, and through the collaborative layer to realize the bidirectional feedback of production and logistics state, to optimize the distribution priority and production scheduling.

[0026] Step 1 More specific implementation is as follows:

[0027] The step 1 specifically includes:

[0028] Multi-bin problem modeling. Production tasks can be produced by multiple workshops, and the materials needed for production are usually signaled by the production line to the inner bin, which directly transports materials to the production line. When the inner bin is out of stock, the inner bin sends a material adjustment signal to the outer bin, and the outer bin replenishes the material to the inner bin. A workshop usually corresponds to only one inner bin, while the material of an inner bin can be replenished by multiple outer bins. There are two collaborative modes in the multi-bin material distribution scenario: same-level collaboration between outer bins and multi-level collaboration between inner and outer bins. In the multi-bin material distribution scenario, scenario models, workshop models, inner bin models, outer bin models, and threshold parameters are set to model the degree of material shortage in the workshop, the material consumption rate in the workshop, the degree of material shortage in the warehouse, the material in-out rate of the warehouse, the call material threshold, and the material adjustment threshold.

[0029] ① Scenario model construction: The multi-bin collaborative scenario is composed of three parts: workshops, inner bins, and outer bins. The number of workshops is , the number of inner bins is , and the number of outer bins is . Usually, the number of workshops is equal to the number of inner bins, i.e. . The set of workshops is , where represents the th workshop; the set of inner bins is , where represents the th inner bin; the set of outer bins is , where represents the th outer bin, and there are kinds of materials needed for production.

[0030] ② Workshop model construction: The number of demand tasks in the workshop is , there are kinds of demand tasks, the capacity of workshop is , , the number of tasks in workshop at time is , where For the workshop middle Number of tasks, task set is ,in Indicates the first indivual Class of tasks, For the first indivual The arrival time of the task. For the first indivual Deadlines for different task types; corresponding deadlines for each type of task. The material requirements are ,in express Class of requirements task for the first The demand for this type of material. There is a certain amount of material available in the workshop for production. Time Workshop In The quantity of this type of material is , Time Workshop In The consumption of this type of material is , Time Workshop The material replenishment amount is , Time Workshop The remaining amount of material in the container is , Time Workshop right The required quantity of this material is .

[0031] (1)

[0032] That is, the remaining material quantity at the current moment is the sum of the remaining material quantity at the previous moment and the remaining material quantity after utilization at this moment; the remaining material quantity cannot be negative.

[0033] (2)

[0034] workshop right The demand for this type of material is in the workshop. middle Number of tasks and The product of the material requirements for each type of task.

[0035] ③ Internal warehouse model construction: Internal warehouse The capacity is The materials in the internal warehouse are used to supply the production lines in the corresponding workshops. In-time warehouse In The quantity of this type of material is , In-time warehouse In The consumption of this type of material is , In-time warehouse The material replenishment amount is , In-time warehouse The remaining amount of material in the container is .

[0036] (3)

[0037] That is, the amount of material in the warehouse at the current moment is the sum of the amount of material in the warehouse at the previous moment and the net increase in the amount of material in the warehouse at this moment. The amount of material in the warehouse cannot be negative.

[0038] ④ External Warehouse Model Construction: Raw materials from various locations are periodically transported to external warehouses as material storage points. Materials in the external warehouses are used to replenish materials in the internal warehouses. Since the types of materials in the external and internal warehouses may differ, sometimes multiple external warehouses need to coordinate to send materials to the internal warehouses for replenishment. External Warehouse The capacity is , Time Warehouse In The quantity of this type of material is , Time Warehouse In The consumption of this type of material is , Time Warehouse The material replenishment amount is , Time Warehouse The remaining amount of material in the container is .

[0039] (4)

[0040] That is, the amount of material in the outer warehouse at the current moment is the sum of the amount of material in the outer warehouse at the previous moment and the net increase in the amount of material in the outer warehouse at this moment. The amount of material in the outer warehouse cannot be negative.

[0041] ⑤ Threshold parameter: The threshold for the production line to call materials into the inner warehouse at any given time is That is, when the workshop middle Material quantity Less than or equal to When the material is replenished, a material call signal is sent to the inner warehouse. The replenishment threshold from the inner warehouse to the outer warehouse at the moment is , that is, when the inner warehouse The amount of similar materials is less than or equal to , a replenishment signal is sent to the outer warehouse, and the outer warehouse replenishes the materials.

[0042] Step 2 is more specifically implemented as follows:

[0043] Step 2.1. Transport coefficient construction. The state quantity of whether the workshop-inner warehouse-outer warehouse is associated in multi-warehouse cooperation is constructed, The state quantity of the workshop-inner warehouse calling is The moment the workshop If the inner warehouse calls for materials, then , otherwise ; The state quantity of the inner warehouse-outer warehouse adjusting is The moment the inner warehouse If the outer warehouse adjusts the materials, then , otherwise . The state quantity of the workshop feeding is The moment the workshop If there is material replenishment, then , otherwise ; The state quantity of the inner warehouse-outer warehouse replenishing is The moment the inner warehouse If there is material replenishment from the outer warehouse , then , otherwise . The distance parameters of the workshop-inner warehouse-outer warehouse in multi-warehouse cooperation are constructed, and then the multi-warehouse transport coefficient is constructed. The closer the distance between the warehouses, the larger the transport coefficient, and the lower the transport cost. The distance between the workshop and the inner warehouse , since the workshop and the inner warehouse are one-to-one corresponding, the workshop only feeds by its corresponding inner warehouse, so:

[0044] (5)

[0045] The distance between the inner warehouse and the outer warehouse , the inner warehouse and the outer warehouse are many-to-many, and there is a corresponding distance between each inner warehouse and each outer warehouse. The distance is not 0.

[0046] The transport material speed is , and the workshop-inner warehouse material transport time is ​, the transport time of the material between the inner warehouse and the outer warehouse is , the transport time calculation formula is shown in formula (6) and formula (7):

[0047] (6)

[0048] (7)

[0049] the inner warehouse and the outer warehouse , the transport coefficient between the inner warehouse and the outer warehouse is , the transport coefficient calculation formula is as follows:

[0050] (8)

[0051] wherein, is the attenuation factor.

[0052] Step 2.2. Multi-warehouse correlation coefficient construction. According to the material type correlation between the inner warehouse and the outer warehouse and the outer warehouse and the outer warehouse, a multi-warehouse correlation coefficient can be constructed, which represents the consistency or complementarity of the transported material. For the cooperation between the inner warehouse and the outer warehouse, the more similar the material types, the greater the synergy advantage, and the fewer the times required to meet the material supplement of the required material type; for the cooperation between the outer warehouse and the outer warehouse, the more complementary the material types, the greater the synergy advantage, and the more complementary multiple outer warehouses can supply material to a certain inner warehouse to meet the material demand.

[0053] Construct the material type state quantity, , which represents whether the workshop needs a material type, if the workshop needs a material type, then , otherwise, ; , which represents whether the inner warehouse stores a material type, if the inner warehouse stores a material type, then , otherwise, ; , which represents whether the outer warehouse stores a material type, if the outer warehouse stores a material type, then , otherwise, . Construct the inner warehouse-outer warehouse correlation coefficient and the outer warehouse correlation coefficient , , which represents the cooperative correlation between the inner warehouse and the outer warehouse ​ , represents the outer warehouse and the inner warehouse , and The calculation formula is shown in equation (9) and equation (10):

[0054] (9)

[0055] (10)

[0056] wherein, , is a weight coefficient, when the inner warehouse and the outer warehouse are in cooperation, the weight coefficient is approximately 1, at this time, the material similarity plays a leading role; when the outer warehouses are in cooperation, the weight coefficient is approximately 0, at this time, the material complementarity plays a leading role.

[0057] Step 2.3. Transportation-association synergy coefficient. According to the transportation coefficient and the inner-outer warehouse association coefficient, the synergy coefficient between the inner and outer warehouses can be constructed. When the inner warehouse is replenished, the replenishment warehouse can be selected according to the synergy coefficient and the material quantity of the outer warehouse. The inner warehouse and the outer warehouse between the inner and outer warehouses is , which represents the degree of fit of the outer warehouse to the inner warehouse , the greater the outer warehouse to the inner warehouse , The calculation formula is shown in equation (11):

[0058] (11)

[0059] Step 3 more specific implementation is as follows:

[0060] Step 3.1. Multi-warehouse synergy dosing threshold optimization algorithm based on reinforcement learning. By combining the reinforcement learning algorithm with the multi-warehouse synergy dosing scenario, the multi-warehouse synergy threshold optimization is realized. In multi-warehouse synergy optimization, the workshop, inner warehouse, outer warehouse, etc. are regarded as the environment, the action of reinforcement learning acts on the environment to change the state of the workshop, inner warehouse and outer warehouse, and then gives the evaluation of the action and the state at the next moment; the state space covers warehouse inventory level, order quantity, synergy demand, synergy cost, calling material threshold, replenishment threshold, etc. information, the reinforcement learning agent adjusts the calling material threshold and the replenishment threshold according to the state The action is generated through the policy network and the value network, that is, the calling material threshold and the replenishment threshold are adjusted, the adjustment of the calling material threshold is , and the adjustment of the dosing threshold is , , Positive value is increment, negative value is decrement; after the action acts on the environment, the environment state is updated as At the same time, the agent is rewarded according to the reward function composed of inventory cost, transportation cost, production efficiency, etc. The reward function includes task completion reward , inventory occupation cost , transportation cost and explosion / production penalty and The reward function calculation formula is shown in formula (12), and the strategy is improved through dynamic penalty term based on adaptive KL divergence, etc. to optimize the advantage function, so as to realize the close combination of multi-warehouse collaborative optimization and reinforcement learning algorithm, and continuously improve the system performance.

[0061] (12)

[0062] Step 3.2. Multi-warehouse collaborative threshold optimization evaluation index construction. The maximum task completion time, inner warehouse turnover rate, outer warehouse turnover rate, transportation cost and other indexes are constructed to evaluate the pros and cons of the results after the multi-warehouse collaborative threshold optimization. The maximum task completion time is the completion time of the last completed task in the time period, the inner warehouse turnover rate is the ratio of the inner warehouse material change amount to the inner warehouse capacity per unit time, the outer warehouse turnover rate is the ratio of the outer warehouse material change amount to the outer warehouse capacity per unit time, and the transportation cost is the transportation cost of manpower, vehicles and other transportation costs in the transportation process.

[0063] Step 4 more specific implementation is as follows:

[0064] Production multi-stage feature demand analysis. The production of 3C products has the characteristics of multi-variety, small batch and high frequency of production change. The production process usually includes SMT patch, plug-in, assembly, testing, packaging and other multiple stages. There are significant differences in material types, process requirements, etc. among production stages, and the process sequence is strict. Inner warehouse material distribution needs to be highly accurate feeding according to production rhythm and stage material demand, so it is necessary to clarify the correlation between production characteristics of each stage and material demand, which can provide basis for building production logistics collaborative double-layer optimization model and production logistics collaborative scheduling optimization method. The feature demand analysis of multi-stage production process analyzes the number of warehouse material types, the types and quantities of materials required by the stage, the material demand period and the material distribution mode, which can provide basis for subsequent production logistics collaboration. The production process of 3C workshop usually includes SMT patch, plug-in, assembly, testing, packaging and other core stages, each stage has the following characteristics:

[0065] ① SMT patch: high precision requirement, large equipment investment, multiple material types (such as resistor, capacitor, IC chip), high requirement for material supply timeliness, and quick switching of tray during production change.

[0066] ②Insertion section: Manual / semi-automatic operation is the main, the material volume is larger (such as connector, transformer), need to be batch feeding according to the process sequence, production line balancing difficulty.

[0067] ③Assembly section: Multi-product mixed production, complex matching of parts (such as shell, cable, mainboard), need to accurately control the material matching, avoid downtime.

[0068] ④Test section: Strong equipment specificity, test procedures change with product model, need to monitor material consumption in real time (such as test fixture, consumables), high data collection frequency.

[0069] ⑤Packaging section: High degree of order customization, many types of auxiliary materials (such as instruction manual, packaging box, damage prevention material), need to quickly respond to order changes, packaging line switching cost is low but frequent.

[0070] Because of the dynamic, correlation and timeliness of material demand between sections, production and logistics scheduling are relatively isolated, resulting in "line waiting for material" or "material accumulation" and other coordination failure problems, at the same time, due to the mismatch between distribution strategy and section characteristics, the first-in first-out task execution strategy is adopted, resulting in low logistics efficiency.

[0071] Because 3C products have the characteristics of fast updating and replacement (such as mobile phone models with an average life cycle of 6-12 months), the types and quantities of materials in each section change frequently with orders, making it difficult to plan production task scheduling. In addition, the material demand between sections has strong correlation, such as connectors that need to be inserted into the section after SMT patching, so it is necessary to establish a material-section association matrix to describe the association between materials and sections in detail.

[0072] A two-layer optimization model for production and logistics collaboration is constructed to enhance the correlation between production and logistics and improve collaborative efficiency. The architecture of this model comprises a production scheduling optimization layer, a collaboration layer, and a logistics distribution optimization layer. The production scheduling optimization layer corresponds to a production scheduling optimization model, the logistics distribution optimization layer corresponds to a logistics distribution optimization model, and the collaboration layer corresponds to the production-logistics collaborative coupling mechanism. The production scheduling optimization layer includes a task model and a work segment model. The task model includes elements such as task requirements, task time, and task status, while the work segment model includes information such as the number of work segments, work segment types, and work segment capacity. The production scheduling optimization layer generates production plans and receives constraint feedback from the collaboration layer to optimize task arrangement. The collaboration layer is the hub connecting the production scheduling optimization layer and the logistics distribution optimization layer. Based on collaboration assumptions and constraints, it coordinates tasks, work segments, logistics, and materials. The collaboration layer defines collaboration goals, such as production waiting time, task efficiency, material waiting time, and completion time, and transmits status updates to other layers while providing logistics solutions to the logistics distribution optimization layer. The logistics distribution optimization layer consists of a logistics model and a material model. The logistics model involves logistics carriers, delivery routes, and logistics status, while the material model includes material quantity, type, and update status, responsible for optimizing material distribution. The production-logistics collaborative two-tier optimization model achieves deep collaboration between production and logistics through layered cooperation, improving overall production efficiency and effectiveness.

[0073] (1) The production scheduling optimization model includes parameters related to the selection of multiple production sections and task execution. The total number of production tasks is... The number of production task types is Task set ,in Indicates the first There are [number] production tasks. The number of production section types is [number]. The work sections are set as follows ,in Indicates the first Work section. Task The corresponding processing section is ,in This is a state variable, representing the task. Do we need a work section? Processing, if Then the task No work section required Processing, or conversely, task Work section required Processing. The set of process segments is as follows: ,in In the workshop Number of work sections, total number of work sections The calculation formula is shown in equation (13):

[0074] (13)

[0075] Further construct production task related time state quantity, production task Arrival time is , production task The deadline is , production task The start time on the section Is , this is a decision variable, production task The end time on the section Is , the time of section Processing is , production task and section task execution time association as shown in equation (14), equation (15).

[0076] (14)

[0077] (15)

[0078] Where, The total processing time of production task .

[0079] On the basis of section and production task association, build section and material demand association, section Processing once the material quantity set , The number of material types, Indicates that the section Processing once the number of the first Class material, the number of each type of material in the warehouse , wherein The number of the first Class material in the warehouse, the upper limit of the production capacity of section Is , that is, the maximum number of workpieces that can be processed on section At the same time.

[0080] (2) In the logistics distribution optimization model, AGV car transports materials from the warehouse to the specific section of the production line. At present, the material distribution mode is mostly to transport materials at fixed time, without transporting materials in real time according to the production rhythm of the production line, which is easy to lead to material accumulation or production line material shortage and stop production, and reduce production efficiency. Therefore, the logistics distribution optimization model is constructed, and the material demand is calculated in real time according to the production state, and then the distribution optimization is carried out. The number of the first Class material in the warehouse at time , the number of the first The distance is The AGV cart's transport speed is constant. Then from the inner warehouse to the work section Delivery time The calculation formula is shown in equation (16):

[0081] (16)

[0082] (3) In the production logistics collaborative coupling mechanism, construct the production logistics collaborative state quantity. express Production tasks Is it in the work section? Upstream processing. Based on the production scheduling optimization model and the logistics distribution optimization model, the following production logistics coordination assumptions are set:

[0083] In the current scenario of joint scheduling of production logistics, only the multi-section production of a single workshop and the material distribution of the corresponding internal warehouse of a single workshop are considered; the transportation distance between sections is not considered, that is, it is assumed that the workpiece can immediately enter the next section for processing after it is processed in one section.

[0084] Assuming there are enough AGVs to transport materials to specific work sections, and without considering any AGV quantity limitations;

[0085] The materials in the inner warehouse are replenished regularly and will not exceed the maximum capacity of the inner warehouse;

[0086] The AGV transports materials at a constant speed.

[0087] Based on the above assumptions, the constraints of the production logistics collaboration scenario are extracted, and the constraint formulas are shown in equations (17)-(21):

[0088] (17)

[0089] (18)

[0090] (19)

[0091] (20)

[0092] (twenty one)

[0093] wherein, formula (17) is an inner warehouse material inventory constraint, the demand of a process for a material at the same time cannot exceed the inventory of the inner warehouse material; formula (18) is a process sequence constraint, the start time of a task in a previous process cannot be later than the start time in a subsequent process; formula (19) is a process capacity constraint, the processing task amount of a process at the same time cannot exceed the maximum capacity of the process; formula (20) is a process sequence constraint, the processing time of a same task in a previous process cannot be later than the processing time in a subsequent process, that is, the task must be processed in the previous process before entering the subsequent process for processing; formula (21) is a process quantity constraint, the number of processes occupied at the same time cannot exceed the total number of processes of the same type.

[0094] The final goal of production logistics coordination is to improve the adaptation degree of the production process and the logistics transportation process, thereby improving the production efficiency of the task and improving the efficiency of logistics distribution, and thus the optimization goal of the production logistics coordination double-layer optimization model is the total waiting time the minimum and maximum production time the minimum and total logistics waiting time the minimum and task benefit the maximum. The total waiting time is the sum of the start time and the arrival time of all production tasks, the maximum production time is the end time of the execution of all production tasks, the total logistics waiting time is the sum of the start time and the arrival time of all logistics tasks, and the task benefit is obtained by superimposing the benefit of each task (here, it can be regarded as having a benefit of 1 for completing a task).

[0095] Step 5 is more specifically implemented as follows:

[0096] 5.1. Production logistics coordination scheduling based on a multi-objective optimization algorithm. The production scheduling optimization model takes minimizing the maximum completion time, minimizing the production waiting time, and maximizing the task benefit as the goal, and optimizes the allocation and sequence of tasks in each process. Each process prioritizes high-priority tasks to avoid significant losses caused by delays in urgent tasks. Through process-machine double coding, task sequence constraints and equipment exclusivity constraints are ensured. The logistics distribution optimization model takes minimizing the material waiting time as the goal, and dynamically adjusts the coupling relationship between the material distribution time and the task start time. According to the AGV transportation speed and distance, the distribution time of the material from the inner warehouse to the process is calculated to ensure timely supply of materials to the task. Constraint conflicts are handled through repair operations, such as when the material replenishment time is later than the task start time, the subsequent task time is forced to be adjusted. The production and logistics states are transmitted through the coordination layer, such as the real-time feedback of process load to the logistics model to adjust the distribution priority; the logistics inventory state affects the production task scheduling to avoid idle processes due to lack of materials. The production status is fed back to the multi-warehouse coordination threshold optimization module, thereby realizing real-time threshold optimization.

[0097] 5.2. Production flow coordination optimization evaluation index construction. The evaluation indexes of material waiting time, production waiting time, maximum completion time, production benefit, etc. are constructed. The material waiting time is the time from production task to material matching, the production waiting time is the time from task to task execution, the maximum completion time is the last completion time of all production tasks, and the production benefit is the profit obtained by production. The indexes evaluate the production benefit from the production and material two levels.

[0098] In summary, the application discloses a multi-bin coordination dynamic optimization method suitable for multi-variety variable batch mixed-line production. The method comprises five steps of multi-bin scene analysis problem modeling, multi-bin material state correlation analysis, multi-bin coordination material adjustment threshold dynamic optimization, production logistics coordination mechanism and production logistics coordination scheduling optimization. The method can analyze the multi-bin coordination scene, construct the multi-bin coordination material correlation, realize the multi-bin coordination dynamic material threshold optimization, construct the production logistics coordination mechanism based on the multi-bin coordination threshold optimization, and realize the production logistics coordination scheduling optimization by combining the multi-objective optimization algorithm.

[0099] The contents not described in detail in the specification of the application belong to the prior art known to those skilled in the art.

[0100] The above is only the preferred embodiment of the application. It should be pointed out that those skilled in the art can make several improvements and refinements without departing from the principles of the application, and these improvements and refinements should also be regarded as the protection scope of the application.

Claims

1. A multi-warehouse collaborative dynamic optimization method adapted to multi-variety, variable-batch mixed-line production, characterized in that, Includes the following steps: Step 1. Abstract the workshop, inner warehouse, and outer warehouse into scenario models. In the workshop model, set the workshop capacity, task types and quantities, arrival and deadline times, and the demand for various materials for each task, ensuring that the remaining material quantity is non-negative, and establish a mapping relationship between task quantity and material demand. In the inner warehouse model, set the capacity, quantity of various materials, consumption and replenishment, and keep the material quantity non-negative. In the outer warehouse model, set the capacity, quantity of various materials, consumption and replenishment, and keep the material quantity non-negative. Set the material calling threshold and material adjustment threshold in real time to trigger the replenishment chain of the workshop, inner warehouse, and outer warehouse. Step 2. Multi-warehouse material status correlation analysis: Construct transportation status quantities for workshop, inner warehouse, and outer warehouse, and calculate transportation coefficients based on distance and transportation speed; calculate the correlation coefficients between inner and outer warehouses and between outer warehouses based on the storage status of material types in each entity; integrate the transportation coefficients and correlation coefficients to calculate the transportation correlation coordination coefficient. Step 3. Dynamic optimization of multi-warehouse collaborative seasoning threshold: Construct an environment including workshop, internal warehouse and external warehouse. Based on the evaluation of task requirement status, warehouse material status and multi-warehouse collaborative status, configure the objective function, constraints and optimization objectives of the optimization problem, and solve for the optimal seasoning threshold for multi-warehouse collaboration. Step 4. Production logistics collaboration mechanism: Construct a two-layer optimization model for production logistics collaboration. In this model, the upper production scheduling optimization layer integrates task and work section models to generate production plans and receives constraint feedback from the collaboration layer. The collaboration layer coordinates tasks, work sections, logistics, and materials, clarifies the collaboration goals of production waiting time and task benefits, and transmits status updates. The lower logistics distribution optimization layer optimizes distribution through logistics and material models, while setting constraints on internal warehouse inventory and work section sequence. The optimization goal is to minimize total waiting time and maximize task benefits, thereby achieving deep collaboration between production and logistics. Step 5. Production logistics collaborative scheduling optimization: Combining multi-objective optimization algorithms and production logistics collaborative mechanisms, the upper-level production scheduling optimization layer aims to minimize the maximum completion time, production waiting time and maximize task benefits, while the lower-level logistics distribution optimization layer aims to minimize material waiting time. Through the collaborative layer, bidirectional feedback between production and logistics status is achieved to optimize distribution priority and production scheduling.

2. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 1, characterized in that, Step 1 specifically includes: ① The scenario model includes: The multi-warehouse collaborative scenario consists of three parts: workshops, internal warehouses, and external warehouses. The number of workshops is... The number of internal warehouses is The number of external warehouses is The workshops are grouped as ,in Indicates the first One workshop; the inner warehouse is assembled as follows: ,in Indicates the first One internal warehouse; the external warehouse is a collection of... ,in Indicates the first There are one external warehouse, and the materials required for production total... kind; ② Workshop Model Construction: The number of required tasks in the workshop is The required tasks are as follows: Seeds, workshop The capacity is , Time Workshop middle The number of tasks is ,in For the workshop middle Number of tasks, task set is ,in Indicates the first indivual Class of tasks, For the first indivual The arrival time of the task. For the first indivual Deadlines for different task types; corresponding deadlines for each type of task. The material requirements are ,in express Class of requirements task for the first The demand for this type of material is such that there is a certain amount of material available in the workshop for production. Time Workshop In The quantity of this type of material is , Time Workshop In The consumption of this type of material is , Time Workshop The material replenishment amount is , Time Workshop The remaining amount of material in the container is , Time Workshop right The required quantity of this material is : （1） That is, the remaining amount of material at the current moment is the sum of the remaining amount of material at the previous moment and the remaining amount of material after utilization at this moment; the remaining amount of material cannot be negative. （2） workshop right The demand for this type of material is in the workshop. middle Number of tasks and The product of the material requirements for each type of task; ③ Internal warehouse model construction: Internal warehouse The capacity is The materials in the internal warehouse are used to supply the production lines in the corresponding workshops. In-time warehouse In The quantity of this type of material is , In-time warehouse In The consumption of this type of material is , In-time warehouse The material replenishment amount is , In-time warehouse The remaining amount of material in the container is : （3） That is, the amount of material in the warehouse at the current moment is the sum of the amount of material in the warehouse at the previous moment and the net increase in the amount of material in the warehouse at this moment. The amount of material in the warehouse cannot be negative. ④ External warehouse model construction: External warehouse The capacity is , Time Warehouse In The quantity of this type of material is , Time Warehouse In The consumption of this type of material is , Time Warehouse The material replenishment amount is , Time Warehouse The remaining amount of material in the container is : （4） That is, the current material quantity in the outer warehouse is the sum of the material quantity in the outer warehouse at the previous time and the net increase in the material quantity in the outer warehouse at this time; the material quantity in the outer warehouse cannot be negative. ⑤ Threshold parameter: The threshold for the production line to call materials into the inner warehouse at any given time is That is, when the workshop middle Material quantity Less than or equal to When the material is replenished, a material call signal is sent to the inner warehouse. The threshold for replenishing materials from the inner warehouse to the outer warehouse at any given time is That is, when the inner warehouse middle Material quantity Less than or equal to When the material replenishment signal is sent to the outer warehouse, the outer warehouse will replenish the material.

3. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 2, characterized in that, Step 2 specifically includes: Step 2.

1. Constructing Transportation Coefficients; Constructing the state variables to determine whether there is a correlation between workshops, internal warehouses, and external warehouses in multi-warehouse collaboration. This refers to the material requisition status in the workshop and internal warehouse. Time Workshop If to the inner warehouse Order materials, then ,on the contrary ; This refers to the quantity of seasonings in the inner and outer warehouses. In-time warehouse If to the outer warehouse Seasonings, ,on the contrary , This refers to the material loading status in the workshop. Time Workshop If materials are replenished, then ,on the contrary ; This refers to the replenishment status of materials between the inner and outer warehouses. If the time is inside the warehouse There are from external warehouses For material replenishment, ,on the contrary Construct distance parameters between workshop, inner warehouse, and outer warehouse in multi-warehouse collaboration, and then construct multi-warehouse transportation coefficients. For the workshop With the inner warehouse The distance between workshops and inner warehouses corresponds one-to-one. Each workshop only receives materials from its corresponding inner warehouse. Therefore: （5） For the inner warehouse With external warehouse The distance between the inner and outer warehouses is a many-to-many relationship, with each inner warehouse having a corresponding distance to each outer warehouse, and none of these distances are zero. The speed of transporting materials is The material transportation time between the workshop and the inner warehouse is The material transportation time between the inner and outer warehouses is The formulas for calculating transportation time are shown in equations (6) and (7): （6） （7） Inner warehouse With external warehouse The transport coefficient between them is The formula for calculating the transport coefficient is as follows: （8） in, It is the attenuation factor; Step 2.

2. Constructing multi-warehouse correlation coefficients; constructing material type status variables. Workshop Is it necessary? State quantities of similar materials, such as those in the workshop need For this type of material, ,on the contrary, ; Indicates internal warehouse Store State variables of similar materials, such as those in the inner warehouse. storage For this type of material, ,on the contrary, ; Indicates external warehouse Store State variables of similar materials, if external warehouse storage For this type of material, ,on the contrary, Construct the correlation coefficient between internal and external warehouses and the correlation coefficient between external warehouses. , Indicates internal warehouse With external warehouse Synergistic Relationship , Indicates external warehouse With external warehouse The synergistic relationship between them and The calculation formulas are shown in equations (9) and (10): （9） （10） in, , These are the weighting coefficients; Step 2.

3. Inner Warehouse With external warehouse The synergy coefficient between them is , indicating external warehouse For the inner warehouse The suitability of the supplementary materials, The larger the warehouse, the better. For the inner warehouse The lower the cost of replenishing materials, The calculation formula is shown in equation (11): （11）。 4. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 3, characterized in that, Step 3 specifically includes: Step 3.

1. In multi-warehouse collaborative optimization, the workshop, internal warehouse, and external warehouse are considered as the environment. Actions affect the environment, thereby changing the state of the workshop, internal warehouse, and external warehouse, and thus providing an evaluation of the action and the state at the next moment. The state space includes warehouse inventory level, order quantity, collaborative requirements, collaborative costs, material requisition threshold, and material replenishment threshold information, based on the state. Generate Actions This involves adjusting the material call threshold and the material replenishment threshold. The adjustment of the material call threshold is as follows: The adjustment of the seasoning threshold is as follows: , , Positive values ​​represent increments, and negative values ​​represent decrements; after an action is applied to the environment, the environment state is updated. The reward function that is formed at the same time gives a reward. The reward function includes a task completion reward. Inventory holding costs Transportation costs And penalties for margin calls / production shutdowns and The reward function is calculated as shown in equation (12): （12） Step 3.

2. Construction of evaluation indicators for multi-warehouse collaboration threshold optimization; construct indicators such as maximum task completion time, internal warehouse turnover rate, external warehouse turnover rate, and transportation cost to evaluate the merits of the results after multi-warehouse collaboration threshold optimization, and solve for the optimal seasoning threshold for multi-warehouse collaboration.

5. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 4, characterized in that, Step 4 specifically includes: The production logistics collaboration two-layer optimization model is constructed. The architecture of the production logistics collaboration two-layer optimization model includes a production scheduling optimization layer, a collaboration layer and a logistics distribution optimization layer. The production scheduling optimization layer corresponds to the production scheduling optimization model, the logistics distribution optimization layer corresponds to the logistics distribution optimization model, and the collaboration layer corresponds to the production logistics collaboration coupling mechanism. The production scheduling optimization model includes parameters related to the selection of multiple production stages and task execution. The total number of production tasks is... The number of production task types is Task set ,in Indicates the first There are [number] production tasks, and the number of production sections is [number]. The work sections are set as follows ,in Indicates the first Work section, task The corresponding processing section is ,in This is a state variable, representing the task. Do we need a work section? Processing, if Then the task No work section required Processing, or conversely, task Work section required The set of processing sections is as follows ,in In the workshop Number of work sections, total number of work sections The calculation formula is shown in equation (13): （13） Further construct the time status quantities related to production tasks, production tasks The arrival time is Production tasks The deadline is Production tasks At the work section The start time is This is the decision variable, the production task. At the work section The end time is Section The processing time is The relationship between the production task and the section task execution time is shown in equations (14) and (15): （14） （15） in, For production tasks Total processing time; Based on the association between work sections and production tasks, the association between work sections and material requirements is constructed. Material quantity required for one processing , For the number of material types, Indicator Section One processing requires the first Quantity of each type of material, number of each type of material in the internal warehouse ,in The first in the inner warehouse Quantity of materials, work section The upper limit of production capacity is That is, work section The maximum number of workpieces that can be processed at the same time; In the logistics and distribution optimization model, for In the warehouse at any time Quantity of different types of materials, from warehouse to work section The distance is The AGV cart's transport speed is constant. Then from the inner warehouse to the work section Delivery time The calculation formula is shown in equation (16): （16） In the production logistics collaborative coupling mechanism, production logistics collaborative state variables are constructed. express Production tasks at all times Is it in the work section? For the upper processing, based on the production scheduling optimization model and the logistics distribution optimization model, the constraints of the production logistics collaboration scenario are extracted, and the constraint formulas are shown in equations (17)-(21): （17） （18） （19） （20） （21） Among them, Equation (17) is the internal warehouse material inventory constraint, the material demand of the work section at the same time cannot exceed the internal warehouse material inventory; Equation (18) is the work section sequence constraint, the start time of the task in the previous work section cannot be later than the start time of the next work section; Equation (19) is the work section capacity constraint, the processing task of the work section at the same time cannot exceed the maximum capacity of the work section; Equation (20) is the process sequence constraint, the processing time of the same task in the previous period cannot be later than the processing time of the next work section, that is, it must be processed in the previous work section before it can enter the next work section for processing; Equation (21) is the work section quantity constraint, the number of work sections occupied at the same time cannot exceed the total number of such work sections.

6. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 5, characterized in that, The optimization objectives of the production logistics collaborative two-layer optimization model are to minimize the total waiting time, minimize the maximum production time, minimize the total logistics waiting time, and maximize the task benefits.

7. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 4, characterized in that, The maximum task completion time is the completion time of the last task completed within that time period. The internal warehouse turnover rate is the ratio of the change in internal warehouse materials to the internal warehouse capacity per unit time. The external warehouse turnover rate is the ratio of the change in external warehouse materials to the external warehouse capacity per unit time. The transportation cost is the cost of manpower and vehicles consumed during the transportation process.

8. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 2, characterized in that, The number of workshops is equal to the number of internal warehouses, that is... .

9. The multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 6, characterized in that, The total waiting time is the sum of the start time and arrival time of all production tasks. The maximum production time is the time when all production tasks are completed. The total logistics waiting time is the sum of the start time and arrival time of all logistics tasks. The task benefits are the sum of the revenue from each task.

10. A multi-warehouse collaborative dynamic optimization method for adapting to multi-variety, variable-batch mixed-line production as described in claim 5, characterized in that, Extract constraints from the production logistics collaboration scenario, and construct the scenario based on the following assumptions: Once a workpiece is finished in one section, it can immediately proceed to the next section for processing. There are enough AGV carts to transport materials to specific work sections, and the number of AGV carts is not a concern. The materials in the inner warehouse are replenished regularly and will not exceed the maximum capacity of the inner warehouse; The AGV transports materials at a constant speed.

Images