A multi-objective layout optimization method for material distribution based on hierarchical weight balancing

Abstract

The application discloses a material distribution multi-objective layout optimization method based on layered weight balance, relates to the technical field of material distribution layout optimization, and comprises the following steps: obtaining material distribution demand and space constraints, determining a layout scheme and a distribution path; determining a distribution network state containing a distribution path distribution load and congestion degree according to the layout scheme; determining the weight of each layer according to the deviation degree of each layer optimization target and in combination with the distribution network state; performing balance correction on the concentration degree of the weight of each layer to obtain balanced weight; calculating the comprehensive fitness of the layout scheme according to the balanced weight by layer, and updating the layout scheme in the space constraints according to the comprehensive fitness; and iteratively performing until a termination condition is met, so that a target layout scheme is obtained. Through dynamic association of the layout scheme and the distribution network, the application realizes dynamic balance of multi-objective weight, overcomes the problems of fixed weight deviation and static evaluation disconnection, and obtains a coordinated and optimized layout scheme under space constraints.

Description

Technical Field

[0001] This invention relates to the field of material distribution layout optimization technology, and more specifically, to a multi-objective layout optimization method for material distribution based on hierarchical weight balancing. Background Technology

[0002] With the development of intelligent manufacturing and flexible production, material distribution in manufacturing workshops involves various materials, distribution equipment, and distribution routes. The spatial layout of material distribution-related facilities directly affects distribution distance, delivery time, space utilization, and the operational status of distribution channels. Therefore, multi-objective optimization of material distribution layout is of great significance.

[0003] Currently, material distribution layout optimization typically employs system layout design, mathematical programming, and heuristic optimization methods such as genetic algorithms and ant colony algorithms to determine layout schemes through parameters such as logistics intensity and spatial distance. For multiple optimization objectives, some methods use preset weights to comprehensively evaluate objectives such as delivery distance, delivery time, and space utilization, or use multi-objective optimization methods to obtain candidate layout schemes.

[0004] However, existing technologies still have the following problems: Different optimization objectives have different dimensions, magnitudes of change, and actual priorities. Using fixed weights or parallel evaluation can easily cause some objectives to have an excessive impact on the comprehensive evaluation results, making it difficult to coordinate the optimization relationship between objectives at different levels. At the same time, existing layout evaluations are usually calculated based on static logistics distance or volume, which makes it difficult to reflect changes in the load and congestion status of delivery channels caused by layout changes, resulting in deviations between the optimized layout scheme and the actual delivery operation status. In addition, the objective weights usually remain unchanged during the optimization process, making it difficult to adjust the search direction according to the current optimization status.

[0005] Therefore, existing technologies require a multi-objective layout optimization method for material distribution that can dynamically adjust the multi-objective evaluation relationship by combining the target level of the material distribution layout and the distribution operation status, and optimize the layout scheme accordingly. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a multi-objective layout optimization method for material distribution based on hierarchical weight balancing. This method dynamically adjusts and balances hierarchical weights by driving the distribution network status, and iteratively updates the layout scheme accordingly. This solves the problems in the prior art where fixed weights lead to target bias, static distance evaluation cannot reflect dynamic congestion, and weights cannot be adjusted with the optimization status.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A multi-objective layout optimization method for material distribution based on hierarchical weight balancing includes the following steps: obtaining material distribution demand and spatial constraints, determining layout schemes and distribution routes; determining the distribution network status based on the layout scheme, wherein the distribution network status includes the distribution load and congestion level of the distribution routes; determining the weights of each level based on the deviation of the optimization objectives at each level and the distribution network status; balancing and correcting the weights of each level based on the concentration of the weights at each level to obtain balanced weights; calculating the comprehensive fitness of the layout scheme by level based on the balanced weights, and updating the layout scheme within the spatial constraints based on the comprehensive fitness; replacing the current layout scheme with the updated layout scheme, and returning to the step of determining the distribution network status until the termination condition is met to obtain the target layout scheme.

[0008] In a preferred embodiment, determining the layout scheme and delivery route includes: determining the deployment priority of each delivery facility based on the delivery volume and frequency in the material delivery demand; determining the location of each delivery facility in sequence according to the deployment priority, and performing constraint verification on the determined locations based on spatial constraints; and determining the corresponding delivery route based on the locations of each delivery facility that have passed the constraint verification and the delivery start point and delivery end point in the material delivery demand.

[0009] In a preferred embodiment, determining the delivery network status according to the layout scheme includes: mapping the layout scheme and delivery paths to a delivery topology network containing network nodes and network edges; counting the number of delivery tasks and the passage capacity of each network edge, and calculating the delivery load of the delivery path; detecting spatiotemporal overlap conflicts between tasks based on the expected passage time window of each delivery task in the delivery topology network, and calculating the congestion degree of the delivery path based on the queuing waiting time caused by the overlap conflict.

[0010] In a preferred embodiment, determining the weights of each level based on the deviation degree of the optimization objectives at each level, combined with the delivery network status, includes: calculating the deviation degree based on the difference between the current evaluation value and the benchmark value of each level's optimization objectives; determining the physical feedback factor of the corresponding level's optimization objectives based on the delivery load and congestion level; calculating the preliminary weights of each level's optimization objectives based on the deviation degree and the physical feedback factor, and normalizing them by level to obtain the weights of each level; wherein, the more severe the congestion represented by the deviation degree or the physical feedback factor, the greater the corresponding preliminary weight.

[0011] In a preferred embodiment, the step of balancing and correcting the weights at each level to obtain balanced weights includes: obtaining the current distribution of weights at each level and calculating the concentration of weights at each level; when the concentration exceeds a preset concentration threshold, balancing and correcting the weights at each level to reduce the relative differences between weights at each level; and normalizing the balanced and corrected weights at each level to obtain balanced weights.

[0012] In a preferred embodiment, the step of calculating the overall fitness of the layout scheme according to the equilibrium weights by level includes: obtaining the evaluation value of the layout scheme in each level of optimization objective; weighting each evaluation value according to the equilibrium weights by level to obtain the level fitness component of each level; and calculating the overall fitness of the layout scheme based on the level fitness component.

[0013] In a preferred embodiment, updating the layout scheme within spatial constraints based on comprehensive fitness includes: selecting parent layout schemes through a selection operation based on comprehensive fitness, and performing position adjustment or spatial reorganization operations on the parent layout schemes to generate child layout schemes; performing physical space verification on the child layout schemes, eliminating schemes that violate spatial constraints or cause physical overlap of facilities, and using the verified schemes as the updated layout schemes.

[0014] In a preferred embodiment, the location adjustment or spatial reorganization operation includes a targeted mutation operation performed based on the delivery network status. The targeted mutation operation includes: extracting network edges and network nodes whose congestion level exceeds a preset congestion warning value as candidate congestion areas based on the delivery network status; identifying target delivery facilities associated with the candidate congestion areas; and performing position translation or orientation adjustment on the target delivery facilities within spatial constraints, using the rule of increasing the physical distance between target delivery facilities or changing the material entry and exit orientation of the target delivery facilities, to obtain the offspring layout scheme after targeted mutation.

[0015] In a preferred embodiment, the physical space verification includes: checking whether the geometric contours of each delivery facility exceed the preset spatial boundaries; checking whether the spacing between adjacent delivery facilities meets the preset channel width requirements; and checking whether the geometric contours of each delivery facility intersect or overlap.

[0016] In a preferred embodiment, the termination condition includes reaching a preset maximum number of iterations, or the change in the overall fitness of consecutive preset generations being less than a preset change threshold.

[0017] This invention maps the layout scheme to a delivery topology network and obtains the delivery load and congestion level of the delivery path, thus forming a closed-loop correlation between the spatial characteristics of the layout scheme and the dynamic operation status of the delivery network. This overcomes the shortcomings of existing static distance evaluations that are disconnected from the actual dynamic delivery. By determining the weights of each level based on the deviation of the optimization objectives at each level and the state of the delivery network, and by balancing and correcting the weights based on the concentration of each level's weights, the weights of multiple objectives can be dynamically adjusted according to the optimization and delivery status. This avoids the problem of fixed weights causing some objectives to overly dominate the evaluation results, ensuring balanced coordination among multiple objectives. Furthermore, through a closed-loop iterative structure, the material delivery layout achieves multi-objective coordinated optimization under the condition of satisfying spatial geometric constraints, and the final layout scheme has robustness against congestion. Attached Figure Description

[0018] Figure 1 A schematic diagram of the multi-objective layout optimization method for material distribution based on hierarchical weight balancing provided in an embodiment of the present invention; Figure 2 A flowchart for weight distribution concentration detection and balance correction provided in an embodiment of the present invention; Figure 3 A flowchart of layout update and physical space verification based on comprehensive fitness evaluation provided for embodiments of the present invention; Figure 4 The comprehensive fitness iterative convergence curve provided in the embodiments of the present invention; Figure 5 A bar chart comparing the core evaluation index results of different optimization methods provided in the embodiments of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1 provides a multi-objective layout optimization method for material distribution based on hierarchical weight balancing, applicable to material distribution scenarios such as manufacturing workshops and warehouse areas with multiple distribution facilities, multiple distribution routes, and multiple layout optimization objectives. In this example, the distribution facilities to be configured, such as shelves, material storage areas, and material loading and unloading locations, are considered as layout objects, and the actual passage areas connecting the distribution facilities are considered as distribution channels. The material distribution requirements include the delivery start point, delivery end point, delivery volume, and delivery frequency; the spatial constraints include the deployable area of ​​the distribution facilities, the allowable spacing between distribution facilities, and the width requirements of the distribution channels.

[0021] The method is based on material delivery demand and spatial constraints. By optimizing the location of delivery facilities and delivery routes, the layout scheme can simultaneously reflect material delivery demand, spatial conditions, and the operational status of the delivery network. During the optimization process, a delivery network is established based on the delivery routes corresponding to the current layout scheme. The delivery load and congestion level of the delivery routes are determined according to the operation of delivery tasks within the network. The weights of each level are determined based on the deviation of the optimization objectives at each level and the status of the delivery network. The weights are then balanced based on the concentration of each level's weights to obtain balanced weights. Based on these balanced weights, the evaluation values ​​of the optimization objectives at each level are weighted to obtain the level fitness components. The overall fitness of the layout scheme is calculated based on these fitness components. The layout scheme is updated within the spatial constraints based on the overall fitness, replacing the current layout scheme. The delivery network status is then redefined, and the processes of weight determination, weight balancing, fitness calculation, and layout update are executed sequentially until a preset termination condition is met, resulting in the target layout scheme.

[0022] Specifically, such as Figure 1 As shown in the figure, the multi-objective layout optimization method for material distribution based on hierarchical weight balancing described in this embodiment includes the following steps: S1, Obtain material delivery requirements and spatial constraints, and determine the layout plan and delivery route; S101, In this step, we first obtain the material delivery requirements and space constraints.

[0023] The material delivery requirements include the delivery origin, delivery destination, delivery quantity per unit time, and delivery frequency for each material. The delivery origin is the material's outbound location or temporary storage area, and the delivery destination is the corresponding production workstation or demand point.

[0024] The spatial constraints include the workshop boundary range, the minimum allowable distance between adjacent delivery facilities, the minimum width requirement for delivery channels, and the location information of fixed obstacles. These spatial constraints are used to define the deployable area of ​​delivery facilities and the passable space of delivery channels.

[0025] S102, after obtaining the material distribution requirements and space constraints, determine the deployment priority of each distribution facility based on the material distribution requirements.

[0026] Specifically, for any delivery facility The overall demand intensity is determined based on the delivery volume and frequency of delivery tasks associated with the delivery facility. Since delivery volume and delivery frequency have different dimensions and numerical ranges, they are first normalized separately: (1) (2) in, In order to cooperate with distribution facilities The number of deliveries per unit time for associated delivery tasks. In order to cooperate with distribution facilities The delivery frequency of associated delivery tasks, and These are the maximum and minimum delivery volumes associated with all delivery facilities, respectively. and These are the maximum and minimum delivery frequencies associated with all delivery facilities, respectively.

[0027] The overall demand intensity is calculated according to the following formula: (3) in, and The preset weighting coefficients are used, and they satisfy the following conditions: This is used to balance the impact of delivery volume and delivery frequency on deployment priority.

[0028] The overall demand intensity is used to characterize the material delivery demand of the corresponding distribution facility. The higher the overall demand intensity, the higher the priority of the corresponding distribution facility. The locations of each distribution facility are determined in descending order of overall demand intensity.

[0029] As one implementation method, the nearest neighbor insertion method is used to determine the location of each distribution facility, specifically including: Step 1: Select the delivery facility with the highest overall demand intensity as the first deployment facility, and place the first deployment facility in the center of the distribution area of ​​the delivery destination; Step 2: For the remaining undeployed delivery facilities, select the delivery facilities to be deployed in descending order of overall demand intensity. Step 3: Generate multiple candidate locations within the spatially constrained deployment area. Calculate the weighted total distance between the delivery facility to be deployed and the already deployed delivery facilities for each candidate location. Then, determine the candidate location that satisfies the spatial constraints and has the smallest weighted total distance as the deployment location of the delivery facility to be deployed. Step four: Repeat steps two and three until the locations of all delivery facilities have been determined.

[0030] The weighted total distance is determined based on the degree of correlation between the current delivery facility and the existing delivery facilities in terms of delivery demand, and is used to ensure that delivery facilities with a high degree of correlation in delivery demand maintain a shorter spatial connection distance in the initial layout.

[0031] After determining the location of each delivery facility, the initial layout scheme is checked for spatial constraints. This involves checking whether each delivery facility is located within the workshop boundary, whether the distance between adjacent delivery facilities meets the preset minimum allowable distance, and whether the width of the delivery channel meets the preset minimum width requirement. For locations that fail the check, the locations are redefined within the available area of ​​the corresponding delivery facility until an initial layout scheme that satisfies the spatial constraints is formed.

[0032] S103 After determining the location of each distribution facility, based on the delivery start point and delivery end point in the material distribution requirements, a delivery route is planned for each delivery task within the delivery channel space corresponding to the initial layout scheme.

[0033] Specifically, the intersection of delivery channels and the corresponding access points of delivery facilities are taken as path nodes, the channel segments connecting adjacent path nodes are taken as path edges, and the actual length of the channel segments is taken as the weight of the path edges to construct a delivery channel network for path search.

[0034] For each delivery task, the shortest path algorithm is used to search for a feasible path that satisfies the spatial constraints between the corresponding delivery origin and delivery destination, and the searched path is determined as the delivery path corresponding to the delivery task.

[0035] When there are multiple feasible paths that satisfy spatial constraints and have the same path length, one of them is determined as the delivery path for the corresponding delivery task according to the preset path selection rules.

[0036] S2, determine the delivery network status according to the layout plan, the delivery network status includes the delivery load and congestion level of the delivery routes; S201. In this step, based on the current layout scheme determined in S1, a corresponding delivery topology network is established to quantify the operational status of each delivery route during the actual delivery process.

[0037] First, a basic delivery channel network is constructed based on the actual aisle space within the workshop. This basic delivery channel network uses the intersections of delivery channels (such as crossroads, T-junctions, and turning points) and the corresponding access points of delivery facility entrances and exits as network nodes, and the actual channel segments connecting adjacent network nodes as network edges. This basic delivery channel network represents all physical aisle space within the workshop available for delivery passage.

[0038] Then, each delivery path in the current layout scheme output by S1 is mapped to the basic delivery channel network. Each delivery path consists of several network edges connected in series, and each network edge can be shared by multiple delivery paths. After mapping, the delivery topology network corresponding to the current layout scheme is obtained.

[0039] The delivery topology network records the physical attributes of each network edge (including channel length, channel width, and throughput capacity) and the path information occupied by each delivery task, providing a data foundation for subsequent calculations of delivery load and congestion levels.

[0040] S202. After establishing the delivery topology network, count the number of delivery tasks on each network edge, and calculate the delivery load of each network edge by combining the passage capacity of each network edge.

[0041] Specifically, the statistical preset period The number of delivery tasks passing through each network edge The preset cycle can be a shift, a workday, or a complete delivery cycle set according to the actual production rhythm. For any network edge , This represents the total number of delivery tasks that pass through this network edge within the preset period.

[0042] Obtain the channel throughput capacity corresponding to each network edge The passage capacity is represented in the preset period. The maximum number of delivery tasks that the network edge can handle. As an optional implementation, the throughput capacity... Determined in the following manner: (4) in, The average operating speed of the delivery equipment in the aisle. The time length of the preset period. For network edge The corresponding channel length, To ensure a safe distance between delivery equipment, This is a floor function. The physical meaning of the above formula is: within a preset period, the upper limit of the number of delivery devices that can continuously pass through this channel segment while ensuring a safe distance.

[0043] Delivery load at each network edge Calculate using the following formula: (5) Among them, delivery load Representing network edges Saturation within a preset period, A larger value indicates that the delivery tasks on the edge of the network are more intensive and the channel resources are more strained.

[0044] S203, after determining the delivery load of each network edge, further detect the spatiotemporal overlap conflict on each network edge based on the expected travel time window of each delivery task in the delivery topology network, and determine the congestion degree of each network edge according to the spatiotemporal overlap conflict.

[0045] Specifically, for those passing through the same network edge For each delivery task, determine the estimated travel time for each delivery task on the edge of the network. ,in For the first Each delivery task enters the network edge The expected time For the first The delivery task is off-network. The expected time.

[0046] The estimated travel time window is determined by propagating the delivery tasks sequentially along the delivery path. For the first network edge on the delivery path, This refers to the estimated time when the delivery task departs from the delivery origin; for subsequent network edges, It equals the sum of the actual completion time of the delivery task at the previous network edge and the travel time from the previous network edge to the current network edge. That is: (6) in, The previous network edge of the current network edge. For the first The actual departure time of a delivery task at the previous network edge (including the queuing time that may occur at the previous network edge). This represents the travel time required to travel from the previous network edge to the current network edge. Thus, congestion at the previous network edge will correspondingly delay the estimated entry time of subsequent network edges, allowing the time window propagation to reflect the actual impact of congestion.

[0047] Detecting different delivery tasks on the same network edge The estimated travel times of two or more delivery tasks overlap. If the estimated travel times of two or more delivery tasks intersect on the timeline, it is determined that these delivery tasks have a spatiotemporal overlap conflict on that network edge.

[0048] For network edges that experience spatiotemporal overlap, the queuing time caused by the overlap is calculated. As one implementation, when the expected travel times of multiple delivery tasks overlap on the same network edge, they are queued according to the order in which they arrive at the edge. A later-arriving delivery task must wait for the previous task to leave before it can enter; the resulting waiting time is the queuing time.

[0049] Congestion levels at each network edge Calculate using the following formula: (7) in, For network edge The above refers to the cumulative queuing time caused by the expected overlap of traffic periods (i.e., the sum of the waiting times of all conflicting delivery tasks at this network edge). For network edge The estimated total travel time for all delivery tasks (i.e., the sum of the estimated travel times for each delivery task). The congestion level characterizes the network edge. The ratio of waiting time to travel time caused by spatiotemporal overlap and conflict within a preset period. The larger the value, the more severe the loss of traffic efficiency caused by task overlap at the network edge.

[0050] S204, after calculating the delivery load of each network edge. and congestion level Then, for any delivery route The delivery load and congestion level of the delivery route are determined based on the delivery load and congestion level of each network edge included in the delivery route.

[0051] Specifically, delivery routes Delivery load Calculate using the following formula: (8) That is, the delivery route The delivery load of a path is equal to the maximum delivery load of all its network edges. This is because the capacity of a delivery path is limited by the network edge with the highest load, which constitutes the bottleneck of the entire path, and the actual efficiency of the entire path is determined by this bottleneck segment.

[0052] Delivery route congestion level Calculate using the following formula: (9) That is, the delivery route The congestion level is equal to the sum of the congestion levels of all network edges it contains. The path-level congestion level is used to characterize the cumulative congestion impact on the delivery task along the entire delivery path. The more congested edges the delivery path passes through, and the more severe the congestion on each edge, the greater the cumulative loss of actual travel time for the entire path.

[0053] in, Indicate delivery route The set of network edges it contains.

[0054] S3, determine the weight of each level based on the degree of deviation from the optimization objective of each level and the status of the delivery network; S301, In this step, we first establish a multi-level optimization target system.

[0055] The multi-level optimization objective system includes multiple optimization levels, arranged from highest to lowest project priority. Each optimization level contains at least one optimization objective. As an optional implementation, the multi-level optimization objective system includes three optimization levels: The first optimization level is the delivery efficiency level, which includes delivery distance and delivery time. The optimization objective of this level directly determines the basic efficiency of material delivery. The second optimization level is the operation coordination level, which includes the load balancing and congestion level of each delivery route. The optimization objective of this level reflects the rationality of the load distribution and traffic efficiency of each route in the delivery network. The third optimization level is the comprehensive improvement level, which includes long-term operating energy consumption and overall logistics costs. The optimization objective of this level seeks further overall improvement based on meeting the objectives of the first two levels.

[0056] The priority order of each optimization level is as follows: delivery efficiency level is higher than operation coordination level, and operation coordination level is higher than comprehensive improvement level.

[0057] It should be noted that the optimization hierarchy established in this step is only for performance evaluation of layout schemes and does not include basic spatial constraints such as spatial boundaries, facility spacing, and passage width. These basic spatial constraints are considered hard constraints and are verified during the initial layout generation in S1, and further filtered through physical space verification during the subsequent layout update in S5. In other words, schemes that violate spatial constraints will not participate in the determination of hierarchy weights in this step or in the subsequent comprehensive fitness calculation.

[0058] S302, after establishing a multi-level optimization target system, obtain the current evaluation value of each optimization target in each optimization level of the current layout scheme, and calculate the degree of deviation of each optimization level.

[0059] For the The first in the optimization hierarchy There are several optimization objectives, and their current evaluation value is defined as follows: The preset benchmark evaluation value is .

[0060] The benchmark evaluation value is preset based on historical data, theoretical optimal value or engineering experience value, and is used to characterize the expected level of the optimization objective under ideal or acceptable conditions.

[0061] Deviation of each optimization objective Calculate using the following formula: (10) in, It is a preset positive decimal to avoid the denominator being zero.

[0062] For cost-based optimization objectives (such as delivery distance, delivery time, energy consumption, and cost), when When, it indicates that the current value is worse than the benchmark value, resulting in a positive deviation; when When the deviation is 0, it means the current value is not worse than the benchmark value, and there is no need to pay more attention to this target. For benefit-oriented optimization targets (such as space utilization), first convert the target value into the corresponding loss value (such as taking the reciprocal or a negative value) to unify it into a cost-oriented direction, and then calculate the deviation according to the above formula.

[0063] No. Overall deviation of the optimization level The following is obtained by weighted summation of the deviations of each optimization objective within this level: (11) in, For the first The total number of optimization objectives within the optimization level. For the first In the optimization level, the first The preset weight coefficients of each optimization objective satisfy... .

[0064] S303, while calculating the deviation of each optimization level, also considers the delivery load of each delivery route output by S2. and congestion level Extract the physical feedback factors corresponding to each optimization level.

[0065] The physical feedback factor is used to characterize the degree of influence of the actual operating status of the delivery network on the optimization objectives of each optimization level.

[0066] For the first optimization level (delivery efficiency level), the corresponding physical feedback factor Determine using the following formula: (12) in, This is the set of all delivery routes in the current layout plan. This represents the total number of delivery routes. This reflects the average load and congestion coupling state of the delivery network under the current layout scheme. When both load and congestion are at a high level, the coupling effect between the two has an amplifying effect on delivery efficiency; therefore, this coupling relationship is represented by a product form.

[0067] For the second optimization level (operational coordination layer), the corresponding physical feedback factor Determine using the following formula: (13) in, The variance of the delivery load across all delivery routes characterizes the degree of balance in load distribution among the routes. This is the average congestion level across all delivery routes, used to characterize the overall traffic efficiency of the delivery network. This is a preset fusion coefficient used to balance the relative importance of load balancing and congestion levels. The larger the value, the worse the operational coordination of the delivery network—the more uneven the load distribution or the more severe the overall congestion, the lower the operational coordination.

[0068] For the third optimization level (comprehensive improvement level), the corresponding physical feedback factor Determine using the following formula: (14) in, and The coefficient is a preset fusion factor. It should be noted that delivery load and congestion level are not direct measures of energy consumption or cost, but rather represent the operational status of delivery equipment, including the frequency of operation, waiting time, and channel utilization intensity. These operational statuses affect the operating time, idle waiting time, and route execution efficiency of delivery equipment, and therefore can serve as a physical feedback representation of long-term operational energy consumption and overall logistics costs. The larger the value, the greater the deviation of energy consumption or cost from expectations under the current delivery network status.

[0069] S304, obtaining the deviation level of each optimization level. and the corresponding physical feedback factor Later, due to and The numerical scale and dimensions may be different, so we first normalize them separately.

[0070] Divide the deviation at each level by the maximum deviation among all levels to obtain the normalized deviation. Divide the physical feedback factor at each level by the maximum value among all physical feedback factors to obtain the normalized physical feedback factor. The normalized deviation and the physical feedback factor both fall between 0 and 1.

[0071] S305, determine the dynamic state weights for each optimization level. Specifically, the... Optimize the dynamic state weights of the hierarchy Calculate using the following formula: (15) in, and The preset fusion coefficient, with the sum of the two being 1, is used to balance the degree of deviation and the influence of the physical feedback factor on the dynamic state weights. It is a preset positive decimal to ensure that each level has a non-zero base weight, so as to avoid a certain optimization level from completely losing its evaluation function in subsequent iterations because the current state has reached the baseline level.

[0072] S306 After determining the dynamic state weights of each optimization level, the preliminary weights of each optimization level are further determined by combining the preset fixed level priority coefficients of each optimization level.

[0073] The fixed-level priority coefficient is used to characterize the inherent priority order of each optimization level in the project, and does not change with the optimization state. Let the first... The fixed priority coefficient for the optimization level is And satisfy .

[0074] No. Preliminary weights of the optimization level Calculate using the following formula: (16) To ensure the priority order of each optimization level (i.e. (Always true), perform priority preservation processing on the initial weights: (17) in, This is a preset priority enhancement coefficient used to ensure that the fixed-level priority coefficient plays a dominant role in the final weight ranking. This is achieved by introducing... Item, even if it is a dynamic state weight at a certain low priority level The weight is much greater than that of higher priority levels, and the final weight of higher priority levels can still maintain an advantage.

[0075] Divide the weights of each optimized level by the sum of all level weights after priority preservation processing, so that the sum of the normalized level weights is 1 and each level weight is greater than 0, to obtain the level weights of each optimization level. The normalized weights at each level satisfy... The hierarchical priority order.

[0076] S4, adjust the weights of each level according to the concentration of each level's weights to obtain balanced weights; In this step, the concentration of the weights of each optimization level determined in S3 is detected, and the over-concentrated weight distribution is balanced and corrected according to the detection results, so that each optimization level obtains a more coordinated weight allocation while maintaining the original priority relationship.

[0077] Figure 2 The flowchart for weight distribution concentration detection and balance correction provided in this embodiment of the invention is a detailed process for step S4. Figure 2 As shown, the process includes the following steps: S401, obtain the weights of each optimization level and calculate the concentration of the current weight distribution.

[0078] This embodiment uses information entropy to characterize the dispersion of weights at each optimization level. The information entropy... Calculate using the following formula: (18) in, for The natural logarithm of information entropy. The larger the information entropy, the more dispersed (i.e., more balanced) the weight distribution across levels; the smaller the information entropy, the more concentrated (i.e., more concentrated in a certain level or a few levels) the weight distribution.

[0079] For the three optimization levels, the maximum value of information entropy is This corresponds to the state where the weights of the three levels are completely equal, i.e. .

[0080] To align the measurement of concentration with intuitive understanding (i.e., the higher the concentration, the larger the value), information entropy is converted into normalized concentration. : (19) in, The value ranges from 0 to 1. When the weights of the three levels are perfectly balanced, ,at this time A value of 0 indicates that the weight distribution is not concentrated at all; when the weights are completely concentrated in a certain level, ,at this time A value of 1 indicates that the weight distribution is completely concentrated. Therefore, The larger the value, the more concentrated the current weight distribution, and the more urgent the need for balance correction.

[0081] As another optional implementation, the degree of weight concentration can also be measured by the variance of the weights at each level; the larger the variance, the more concentrated the weight distribution.

[0082] S402, the normalized concentration Compared with the preset concentration threshold Compare. When normalized concentration Greater than the preset concentration threshold If the current weight distribution is found to be in a state of concentrated imbalance, a balance correction is triggered; otherwise, if the current weight distribution is not found to be in a state of concentrated imbalance, a balance correction is not performed, and the weights of each optimization level output by S3 are directly adjusted. The weights are output to S5 as equalization weights.

[0083] The concentration threshold The values ​​are pre-set based on the allowable weight differences between optimization objectives and the optimization requirements of the actual delivery scenario, with a range of values ​​ranging from [value range missing]. As one implementation method, the... It can be set to a value between 0.3 and 0.5. When the weight exceeds this range, it indicates that the weight of one or more levels is significantly higher than that of other levels, and it is necessary to adjust the balance to avoid the optimization process from being overly biased towards that level.

[0084] S403, when it is determined that the current weight distribution is in a state of concentrated imbalance, an equilibrium correction is performed on the weights of each optimization level to shrink back to a uniform weight distribution. Specifically, the first... Balanced adjustment weights at each optimization level Calculate using the following formula: (20) in, For the first time before the correction Hierarchical weights This is a preset equilibrium strength coefficient, ranging from 0 to 1, used to control the degree to which the current weights shrink back towards a uniform weight distribution.

[0085] when When the value is small, the equilibrium correction magnitude is small, and more of the current weight distribution characteristics can be preserved; when When the weights of each optimization level increase, the weights of each optimization level are directed towards... The uniform distribution is further reduced, and the concentration of weights decreases. As one implementation method, the equilibrium intensity coefficient... It can be set to a value between 0.2 and 0.4 to achieve gradual balancing rather than one-time balancing, avoiding drastic changes in weights that could cause sudden changes in the direction of subsequent optimization searches.

[0086] S404 outputs the weights from S3 when the equalization correction is not triggered, or the weights after the equalization correction is triggered. Normalization is performed to make the sum of the weights at each level equal to 1, thus obtaining a balanced weight. : (twenty one) because The corrected weights still satisfy In other words, the balance correction only reduces the weight differences between each optimization level, without changing the established priority relationship between the delivery efficiency layer, the operation coordination layer, and the comprehensive improvement layer.

[0087] S5, calculate the overall fitness of the layout scheme according to the balanced weights at each level, and update the layout scheme within the spatial constraints based on the overall fitness. Figure 3 The flowchart for layout update and physical space verification based on comprehensive fitness evaluation provided in this embodiment of the invention is specifically a detailed process for step S5. For example... Figure 3 As shown, the layout update and physical space verification process includes the following steps: S501, obtain the evaluation value of each optimization objective of the current layout scheme at each optimization level.

[0088] For the The first in the optimization hierarchy An optimization objective, and its evaluation value It is calculated based on the type of the optimization objective and the actual parameters of the current layout scheme.

[0089] For the delivery efficiency layer (level 1), the evaluation value of the delivery distance target is the sum of the delivery distances of each delivery route under the current layout plan; the evaluation value of the delivery time target is the sum of the estimated travel times of each delivery route.

[0090] For the operation coordination layer (second level), the evaluation value of the delivery load balancing target is the variance of the load of each delivery route; the evaluation value of the congestion target is the average value of the congestion level of each delivery route.

[0091] For the comprehensive improvement layer (third level), the evaluation value of the long-term operating energy consumption target is the total energy consumption under the current layout scheme; the evaluation value of the comprehensive logistics cost target is the total logistics cost under the current layout scheme.

[0092] Before calculating the overall fitness, all optimization objectives are uniformly converted into loss-type evaluation values, meaning that the higher the evaluation value, the worse the performance of the optimization objective. For cost-type optimization objectives (such as delivery distance, delivery time, energy consumption, and cost), their evaluation values ​​are used directly; for benefit-type optimization objectives, they are converted into loss-type evaluation values.

[0093] S502 After obtaining the loss-type evaluation values ​​of each optimization objective, since the dimensions and numerical ranges of different optimization objectives may differ, the evaluation values ​​of each optimization objective are standardized to make the different objectives comparable in the subsequent weighted summation.

[0094] For the The first in the optimization hierarchy An optimization objective, and its standardized evaluation value Calculate using the following formula: (twenty two) in, The loss-type evaluation value before standardization. and These are the minimum and maximum values ​​of the optimization objective within the current set of candidate layout schemes, respectively. This is a pre-defined positive decimal to avoid a denominator of zero. The standardized evaluation value. The value range is [0,1], and the smaller the value, the better the performance of the optimization objective.

[0095] S503, after obtaining the standardized evaluation values ​​of each optimization objective, based on the equilibrium weights output in S4. The fitness components of each level are calculated according to the optimization level.

[0096] For the Optimization hierarchy, its hierarchy fitness component The results are obtained by weighted summation of the standardized evaluation values ​​of each optimization objective within this level: (twenty three) The hierarchical fitness component Characterizing the current layout scheme in the th... Optimize overall performance at different levels. The smaller the value, the better the overall performance of that level.

[0097] S504, weighted summation of the fitness components at each level to obtain the overall fitness of the current layout scheme. : (twenty four) Overall fitness The smaller the value, the better the overall performance of the layout scheme.

[0098] S505, after calculating the overall fitness of each layout scheme, update the layout scheme within the spatial constraints based on the overall fitness.

[0099] The selection process is based on the overall fitness of each layout scheme, retaining the layout scheme with the best overall fitness as the parent layout scheme. As one implementation method, a tournament selection strategy is used: several candidate schemes are randomly selected from the current candidate layout scheme set, and the scheme with the best overall fitness is selected to enter the parent set. This process is repeated until the size of the parent set reaches a preset number.

[0100] Furthermore, position adjustment or spatial reorganization operations are performed on the parent layout scheme to adjust the spatial coordinates of each distribution facility and generate the child layout scheme.

[0101] In one implementation, the position adjustment operation includes exchanging partial spatial layout information between two parent layout schemes to generate a new child scheme. Specifically, position adjustment is performed with a preset probability, randomly selecting spatial regions in the parent schemes for exchange, and performing conflict detection on the exchanged layout schemes.

[0102] In one implementation, the spatial reorganization operation includes making local spatial adjustments to a single parent layout scheme to generate a new child scheme. Specifically, spatial reorganization is performed with a preset probability, including shifting the location or adjusting the orientation of delivery facilities.

[0103] Furthermore, the location adjustment or spatial reorganization operation includes a targeted mutation operation performed based on the delivery network status. This targeted mutation operation performs targeted adjustments to localized areas of high congestion within the delivery network. Based on the delivery network status determined by S2, network edges and network nodes with congestion levels exceeding a preset warning value are extracted as candidate congestion areas; target delivery facilities associated with the candidate congestion areas are identified; and within the spatial constraints, the target delivery facilities are translated or their orientations are adjusted by increasing the physical distance between them or changing the material entry and exit orientation of them, to obtain a targeted mutation offspring layout scheme.

[0104] Unlike random adjustments, the targeted mutation operation identifies congestion hotspots and their associated delivery facilities, and then spatially adjusts the facilities causing congestion. Moving facilities alters the spatial connectivity between delivery channels and the length of delivery paths, thus changing the distribution of delivery tasks along the network edges. Adjusting facility orientation changes the connectivity between delivery facility entrances / exits and delivery channels, thereby altering the spatiotemporal distribution of delivery tasks entering the corresponding network edges. Therefore, the targeted mutation operation is driven by the delivery network state fed back from S2 and has a clear physical optimization direction.

[0105] Furthermore, after generating the child layout schemes, a physical space verification is performed on the child layout schemes to eliminate schemes that violate the spatial constraints or cause physical overlap of facilities. The physical space verification includes the following four checks: Check whether the geometric outline of each delivery facility exceeds the preset spatial boundary; check whether the geometric outline of each delivery facility intersects or overlaps with fixed obstacles; check whether the distance between adjacent delivery facilities meets the preset channel width requirements; check whether the geometric outlines of each delivery facility intersect or overlap.

[0106] A child layout scheme is retained as a valid candidate only if it passes all four of the above checks. Child layout schemes that fail the checks are directly eliminated and do not participate in subsequent iterations.

[0107] From the child layout schemes that pass the physical space verification, a predetermined number of layout schemes are selected as the next generation layout schemes based on their overall fitness. That is, all child layout schemes that pass the physical space verification are sorted in ascending order of overall fitness, and the top-performing scheme with the best overall fitness is selected. These schemes constitute the next-generation candidate layout scheme set, among which The preset population size is used. Child schemes that do not enter the next generation set are not retained in the iteration to control the population size and maintain optimization pressure. The selected next generation layout scheme is used as the updated layout scheme and output to step S2.

[0108] S6, replace the current layout scheme with the updated layout scheme, return to the step of determining the delivery network status, until the termination condition is met, and obtain the target layout scheme.

[0109] This step controls the iterative optimization process from S1 to S5, determines whether the termination condition is met, and outputs the target layout scheme.

[0110] In this process, S1 is only used to generate the initial layout scheme; subsequent iterations use the updated layout scheme output by S5 as the current layout scheme, return to step S2 to redetermine the delivery network status, and do not repeat S1. The layout schemes output by S5 have all passed the physical space verification, so S6 does not repeat the space constraint verification.

[0111] S601: After outputting the updated layout scheme in S5, obtain the current iteration number. And obtain the overall fitness corresponding to the current iteration.

[0112] During the iteration process, the effective layout schemes and their overall fitness obtained in each iteration are recorded, and the overall fitness of the current iteration is compared with the historical best overall fitness.

[0113] For the The overall fitness of the layout scheme obtained from rounds of iteration is denoted as . The overall fitness is calculated by equation (24) in S5.

[0114] The overall fitness of the current round Overall fitness with historical best The comparison is performed, and if the overall fitness of the current round is less than the historical best overall fitness, the current layout scheme is determined as the new historical best layout scheme, and its overall fitness is updated as follows: (25) Otherwise, the historical best layout and its overall fitness remain unchanged. This method preserves the historical best layout independently of the layout used in the current iteration, preventing inferior layouts from overwriting already obtained optimal layouts in subsequent iterations.

[0115] S602, Pre-set the maximum number of iterations After each iteration from S2 to S5 is completed, the current iteration number is recorded. Increment by 1, and set the current iteration number. With maximum number of iterations Comparison. When When the maximum number of iterations has been reached, the iteration is terminated, and the target layout scheme is output.

[0116] The maximum number of iterations is preset based on factors such as the number of delivery facilities, the number of candidate layout schemes, the number of optimization targets, and computing resources. For example, the maximum number of iterations can be set to 500-2000 iterations. At that time, a convergence determination is further performed.

[0117] S603 records the optimal overall fitness for each round during the iteration process and sets the observation window length. And the threshold for the amount of change. For the current number... Round iteration, when At that time, calculate the relative change in overall fitness between the current round and the starting round of the observation window. : (26) in, The preset observation window length represents the number of iterations used to determine whether the optimization state has reached a stable state. This is a preset threshold for the amount of change.

[0118] when Less than If the change in the historical best overall fitness within the observation window is less than a preset threshold, the current layout optimization process is considered to have stabilized and meets the convergence termination condition. Otherwise, the current optimization process is considered not to have converged, and iteration continues.

[0119] As one implementation method, the observation window length The number of generations can be set to 20 to 50, and the change threshold δ can be set to 0.001 to 0.01. These parameters can be adjusted according to the search size and convergence characteristics of the actual layout optimization problem.

[0120] S604: If either of the above two conditions is met, the iteration terminates. If neither condition is met, the updated layout scheme returns to step S2, and the next iteration continues.

[0121] After the iteration ends, the historically optimal layout scheme recorded during the iteration process is determined as the target layout scheme. The target layout scheme includes the final deployment location of each delivery facility, the delivery path of each delivery task, and the corresponding comprehensive fitness. This scheme has undergone delivery network status evaluation and spatial constraint verification in multiple iterations, and can achieve coordinated optimization of multi-level objectives such as delivery efficiency, operational coordination, and comprehensive improvement while meeting spatial constraints.

[0122] Example 2: To further verify the applicability of the present invention in different production scenarios, this example uses the optimization of material distribution layout in an automotive parts manufacturing workshop as an example for simulation verification. The selection of parameters in this example is only for illustrating the implementation process of the present invention in a specific industrial scenario and does not constitute a limitation on the scope of protection of the present invention.

[0123] In this embodiment, the workshop has a floor plan of 90m × 60m and is equipped with 12 delivery facilities. The workshop has a herringbone-shaped delivery channel, with the main channel being 3.5m wide and the branch channels being 2.5m wide, and 3 fixed obstacle areas are set up.

[0124] A total of 8 types of materials are involved, forming 18 delivery tasks, with a delivery volume of 3-30 pieces per shift and a delivery frequency of 1-6 times per shift. 5 AGVs are used for delivery, with an average operating speed of 0.6m / s, a safety distance of 1.8m, and a preset delivery cycle of 480min per shift.

[0125] Spatial constraints: Workshop boundary 90m×60m, minimum facility spacing 1.5m, minimum passage width 2.5m.

[0126] Optimization goal: Adopt the same three-tier structure as Example 1 (delivery efficiency layer, operation coordination layer, and comprehensive improvement layer).

[0127] Algorithm parameters: Population size 40, maximum number of iterations 800, crossover probability 0.80, mutation probability 0.10. Fixed hierarchy priority coefficient. Degree of deviation and fusion coefficient of physical feedback factor Normalized concentration threshold Equilibrium strength coefficient Convergence criteria: observation window of 25 generations, change threshold of 0.005.

[0128] Figure 4 The image shows the iterative convergence curve of the overall fitness provided in an embodiment of the present invention. Figure 4As shown, the optimization was performed according to the method described in Example 1. After 620 iterations, the convergence condition was met, and the target layout scheme was output. The comparison of the indicators before and after optimization is shown in Table 1.

[0129] Table 1 Comparison of key indicators before and after optimization

[0130] Three control experiments were set up under the same conditions, and the results of each control group were compared with those of the present invention, as shown in Table 2.

[0131] Control group settings: ① Fixed weight method (no dynamic weighting and balance correction); ② No balance correction method (no S4); ③ No target mutation method (no target mutation).

[0132] Table 2 Comparison of optimization results from different methods

[0133] Figure 5 This is a bar chart comparing the core evaluation index results of different optimization methods provided in the embodiments of the present invention. The normalized comparison results of each evaluation index are displayed as a percentage, with the control group (fixed weight method) as the 100% baseline.

[0134] This embodiment was validated under different workshop sizes, layouts, and delivery task densities compared to Embodiment 1. After optimization, the total delivery distance decreased by 22.92%, the average congestion level decreased by 44.44%, and the overall logistics cost decreased by 14.92%, with significant improvements in all evaluation indicators. Comparative experiments further verified that dynamic weight determination, weight balance correction, and targeted mutation mechanisms still have independent and cumulative contributions in different scenarios.

[0135] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0136] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0137] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0138] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0139] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0140] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-objective layout optimization method for material distribution based on hierarchical weight balancing, characterized in that, Includes the following steps: Obtain material delivery requirements and spatial constraints, and determine the layout plan and delivery routes; The delivery network status is determined based on the layout plan, and the delivery network status includes the delivery load and congestion level of the delivery routes; The weights of each level are determined based on the degree of deviation from the optimization objectives at each level, combined with the status of the delivery network. The weights at each level are adjusted to achieve a balanced weight based on the degree of concentration of weights at each level. The overall fitness of the layout scheme is calculated hierarchically based on the balanced weights, and the layout scheme is updated within the spatial constraints based on the overall fitness. Replace the current layout with the updated layout, return to the step of determining the delivery network status, and continue until the termination condition is met to obtain the target layout.

2. The method according to claim 1, characterized in that, The determination of the layout plan and delivery route includes: Based on the delivery volume and frequency in the material delivery requirements, determine the deployment priority of each delivery facility; The locations of each delivery facility are determined sequentially according to deployment priority, and the determined locations are verified based on spatial constraints. The corresponding delivery route is determined based on the location of each delivery facility that has passed the constraint verification and the delivery start and end points in the material delivery requirements.

3. The method according to claim 1, characterized in that, Determining the delivery network status based on the layout plan includes: The layout scheme and delivery routes are mapped into a delivery topology network that includes network nodes and network edges; Count the number of delivery tasks and traffic capacity at each network edge, and calculate the delivery load of the delivery path; Based on the estimated travel time window of each delivery task in the delivery topology network, spatiotemporal overlap and conflict between tasks are detected, and the congestion level of the delivery path is calculated based on the queuing waiting time caused by the overlap and conflict.

4. The method according to claim 3, characterized in that, The determination of weights for each level based on the degree of deviation from the optimization objectives at each level, combined with the status of the delivery network, includes: The degree of deviation is calculated based on the difference between the current evaluation value and the benchmark value of each level of optimization objective; Based on delivery load and congestion levels, determine the physical feedback factors for the corresponding optimization objectives at each level; The initial weights of the optimization objectives at each level are calculated based on the degree of deviation and the physical feedback factor, and then normalized according to the level to obtain the weights of each level; wherein, the more severe the congestion represented by the degree of deviation or the physical feedback factor, the greater the corresponding initial weight.

5. The method according to claim 4, characterized in that, The process of balancing and adjusting the weights at each level to obtain balanced weights includes: Obtain the distribution of weights at each level and calculate the degree of concentration of weights at each level; When the concentration level exceeds a preset concentration threshold, the weights of each level are balanced and corrected to reduce the relative differences between the weights of each level. The weights of each level after the balance correction are normalized to obtain the balance weights.

6. The method according to claim 5, characterized in that, The comprehensive adaptability of the layout scheme calculated hierarchically based on the balanced weights includes: Obtain the evaluation value of the layout scheme in each level of optimization objective; Based on the balanced weights, the evaluation values ​​are weighted according to the levels to obtain the level fitness components of each level; The overall fitness of the layout scheme is calculated based on the hierarchical fitness component.

7. The method according to claim 6, characterized in that, The method of updating the layout scheme within spatial constraints based on comprehensive fitness includes: Based on the overall fitness, the parent layout scheme is selected through selection operations, and the parent layout scheme is subjected to position adjustment or spatial reorganization operations to generate the child layout scheme. Physical space verification is performed on the child layout schemes, and schemes that violate spatial constraints or cause physical overlap of facilities are eliminated. The schemes that pass the verification are used as the updated layout schemes.

8. The method according to claim 7, characterized in that, The location adjustment or spatial reorganization operation includes targeted mutation operations performed based on the delivery network status, the targeted mutation operations including: Based on the delivery network status, network edges and network nodes whose congestion level exceeds the preset congestion warning value are extracted as candidate congestion areas; Identify target delivery facilities associated with candidate congestion areas; By increasing the physical distance between target delivery facilities or changing the material entry and exit orientation of target delivery facilities as rules, the target delivery facilities are translated or their orientation is adjusted within spatial constraints to obtain the offspring layout scheme after targeted mutation.

9. The method according to claim 8, characterized in that, The physical space verification includes: Check whether the geometric outline of each delivery facility exceeds the preset spatial boundaries; Check whether the spacing between adjacent delivery facilities meets the preset aisle width requirements; Check whether the geometric outlines of each distribution facility intersect or overlap.

10. The method according to claim 9, characterized in that, The termination conditions include reaching a preset maximum number of iterations, or the change in the overall fitness of consecutive preset generations being less than a preset change threshold.

Images