Intelligent logistics order dispatching method and system

Abstract

The application relates to the technical field of intelligent logistics, in particular to an order dispatching method and system for intelligent logistics. The method comprises the following steps: dividing a logistics area into multiple grids; in response to a new order event, generating a preset initial value of a profit pheromone; acquiring a transport capacity node density in each grid, and generating a congestion pheromone when the transport capacity node density exceeds a preset threshold; calculating a composite pheromone in a preset grid range for each transport capacity node; based on the composite pheromone, calculating a state transition probability of the transport capacity node moving to an adjacent grid or accepting an order by using a probability selection algorithm; and generating a dispatching instruction according to the state transition probability of each transport capacity node. By means of the double pheromone mechanism, the problems of global supply-demand imbalance and local transport capacity congestion in the traditional dispatching method are effectively solved, and the scheduling efficiency and response speed of the logistics system are significantly improved.

Description

Technical Field

[0001] This application relates to the field of smart logistics technology. More specifically, this application relates to a smart logistics order dispatching method and system. Background Technology

[0002] Smart logistics is a crucial direction for the development of modern logistics, its core being the optimization and efficient scheduling of logistics resources through information technology. In scenarios such as instant delivery and same-city logistics, the order dispatch system acts as a bridge connecting customer demand and transportation capacity supply, and its performance directly impacts service quality and operational efficiency. The order dispatch problem is essentially a dynamic, multi-objective optimization problem, requiring the maximization of transportation capacity utilization and the reduction of delivery costs while meeting order timeliness requirements.

[0003] Traditional order dispatching methods primarily rely on rule-driven or simple distance-matching strategies. These methods have significant drawbacks: First, they cannot effectively address the spatiotemporal imbalances in order distribution, leading to over-concentration of capacity in densely populated areas and under-concentration in sparsely populated areas. Second, traditional methods essentially treat capacity nodes as isolated individuals for scheduling, lacking a collective coordination mechanism. This contrasts sharply with the efficient division of labor and cooperation achieved by organisms like ant colonies releasing chemical pheromones in nature. Without such pheromone-like global guidance signals, drivers often rely solely on personal experience or localized vision to blindly navigate, resulting in highly random and unpredictable capacity distribution, making it difficult to form an efficient self-organizing network.

[0004] While some existing dispatching methods are based on machine learning or optimization algorithms, most focus on static matching and optimization of orders and drivers, and route planning. They lack a systematic solution for proactive spatial scheduling of transportation resources through information guidance mechanisms. Inspired by the pheromone feedback mechanism in ant colony optimization, this application aims to introduce the principle of using pheromone concentration to guide group behavior in the biological world into the field of logistics scheduling. However, traditional ant colony optimization mainly uses positive feedback mechanisms to promote path convergence (i.e., all ants take the same optimal path), which, if directly applied to logistics dispatching, would lead to capacity congestion. Therefore, existing technologies lack an intelligent dispatching method that can both utilize pheromones to simulate the attraction of food sources to transportation capacity and utilize congestion pheromones to simulate the biological aversion to overcrowded environments, thereby achieving real-time perception of regional supply and demand and dynamic adjustment of transportation capacity distribution. Summary of the Invention

[0005] The purpose of this application is to propose an intelligent logistics order dispatch method and system to solve the problems of low efficiency and uneven distribution of transportation capacity in existing logistics dispatch technologies.

[0006] In a first aspect, this application provides an order dispatching method for smart logistics, comprising: dividing a logistics area into multiple grids, each grid including two parameters: revenue pheromone and congestion pheromone; generating a revenue pheromone with a preset initial value in the grid where the order is located in response to a new order event; obtaining the density of transport nodes within each grid, and generating a congestion pheromone when the density of transport nodes exceeds a preset threshold; the order event includes at least the grid where the order is located and the order generation time; calculating a synthetic pheromone within a preset grid range for each transport node, the synthetic pheromone being obtained by weighted calculation of the revenue pheromone and the congestion pheromone; calculating the state transition probability of a transport node moving to an adjacent grid using a probabilistic selection algorithm based on the synthetic pheromone; and generating a dispatching instruction according to the state transition probability of each transport node. The revenue pheromone decays nonlinearly to surrounding grids based on the spatial diffusion coefficient within the logistics area at the current time; the spatial diffusion coefficient is determined based on the global supply-demand ratio at the current time, the global supply-demand ratio being the ratio between the total number of unresponsive orders and the total number of transport nodes to be dispatched within the current logistics area.

[0007] The core innovation of this application lies in establishing a capacity guidance mechanism based on dual-dimensional pheromones. By simultaneously maintaining two parameters—revenue pheromones and congestion pheromones—within the logistics area grid, a quantitative expression of order attractiveness and regional congestion status is achieved. This dual-dimensional pheromone guidance mechanism solves both the problem of capacity aggregation towards order areas and the problem of local congestion caused by uneven capacity distribution, achieving dual optimization of order response speed and capacity utilization efficiency. Compared with traditional single-matching methods, this application can maintain stable order dispatch performance under different order densities and traffic conditions, significantly reducing order waiting time and empty capacity rate.

[0008] Optionally, dividing the logistics area into multiple grids includes: constructing a two-dimensional grid map of the logistics area and establishing an adjacency matrix describing the spatial adjacency relationship of each grid; initializing two dimensions of data for each grid, which are used to store revenue pheromones and congestion pheromones respectively, and setting the initial values ​​of revenue pheromones and congestion pheromones of all grids to zero; and setting the spatial attributes of each grid, including the grid center coordinates and the grid coverage area.

[0009] By constructing a two-dimensional grid map of the logistics area and establishing an adjacency matrix, a structured computational foundation is provided for the spatial diffusion of pheromones and the state transitions of transport nodes. This standardized grid partitioning method ensures computational efficiency and facilitates the system's expansion to logistics areas of different sizes, providing a feasible technical path for large-scale transport capacity scheduling.

[0010] Optionally, the nonlinear decay of the revenue pheromone towards the surrounding grids includes: determining a set of affected grids within a preset range, centered on the grid where the order is located; calculating the spatial distance between the center of each grid in the affected grid set and the center of the grid where the order is located; and calculating the revenue pheromone at each affected grid, wherein the revenue pheromone decreases exponentially with the spatial distance, and the rate of decrease is controlled by the spatial diffusion coefficient as a parameter.

[0011] The spatial diffusion coefficient, used as a parameter to control the decay rate, enables the system to dynamically adjust the impact range of orders based on the global supply and demand status. This nonlinear decay mechanism aligns with the law that order attractiveness decreases with distance in real-world logistics scenarios. It avoids the ineffective mobilization of long-distance transport capacity while ensuring priority response from short-distance transport capacity, thus improving the spatial rationality of order dispatch.

[0012] Optionally, the calculation process of the spatial diffusion coefficient includes: presetting a first supply-demand ratio threshold and a second supply-demand ratio threshold, wherein the first supply-demand ratio threshold is greater than the second supply-demand ratio threshold; when the global supply-demand ratio is greater than the first supply-demand ratio threshold, the spatial diffusion coefficient is set to a first value, the first value corresponding to a slower decay rate; when the global supply-demand ratio is less than the second supply-demand ratio threshold, the spatial diffusion coefficient is set to a second value, the second value corresponding to a faster decay rate; wherein the first value is less than the second value.

[0013] Optionally, the positive correlation between the revenue pheromone and the order waiting time includes: monitoring the duration of the order waiting to be scheduled; calculating the time increment of the revenue pheromone, wherein the time increment is proportional to the duration, and superimposing the time increment onto the revenue pheromone value of the grid where the order is located; if the superimposed revenue pheromone exceeds a preset upper limit value, then setting the revenue pheromone to the preset upper limit value.

[0014] Optionally, generating congestion pheromones when the capacity node density exceeds a preset threshold includes: counting the number of capacity nodes in each grid at the current moment; calculating the capacity node density, which is the number of capacity nodes divided by the coverage area of ​​the grid; comparing the capacity node density with a preset threshold; if the capacity node density is greater than the preset threshold, the generated congestion pheromone value is positively correlated with the difference between the capacity node density and the preset threshold, otherwise the congestion pheromone value is zero.

[0015] By statistically analyzing the number of transport nodes within each grid and calculating their density, a congestion pheromone proportional to the excess is generated when the density exceeds a preset threshold. This enables the quantitative identification of areas with excessive transport capacity aggregation. This density-based congestion detection and avoidance mechanism effectively solves the problem of resource waste caused by the blind aggregation of transport capacity, promotes the balanced distribution of transport capacity within the logistics area, and improves the overall availability and responsiveness of transport capacity.

[0016] Optionally, the calculation process of the synthetic pheromone includes: determining the neighborhood grid set of the current location of each capacity node to be scheduled, wherein the neighborhood grid set is composed of the grid where the capacity node is currently located and multiple neighboring grids; obtaining the revenue pheromone and congestion pheromone of each grid in the neighborhood grid set; calculating the synthetic evaluation value of each grid, wherein the synthetic evaluation value is equal to the revenue pheromone multiplied by a first weighting coefficient minus the congestion pheromone multiplied by a second weighting coefficient; the synthetic evaluation value of each grid in the neighborhood grid set constitutes the synthetic pheromone.

[0017] By determining the neighborhood grid set of each capacity node and obtaining the revenue pheromone and congestion pheromone of each grid, a weighted linear combination method is used to calculate the composite evaluation value. This ensures that the decision-making of capacity nodes considers both the attractiveness of order revenue and the repulsive force of regional congestion. This multi-factor comprehensive evaluation method for calculating composite pheromones provides capacity nodes with a more comprehensive and accurate decision-making basis, improving the rationality of state transition probability calculation and the optimization of order dispatch results.

[0018] Optionally, the calculation process of the state transition probability includes: setting the negative elements in the synthetic pheromone to zero; calculating the sum of all elements in the processed synthetic pheromone; if the sum is greater than zero, calculating the ratio of the synthetic evaluation value of each grid to the sum, and using the ratio as the probability of the transport node transferring to the corresponding grid; if the sum is equal to zero, setting the probability of the transport node transferring to all adjacent grids to an equal value.

[0019] Optionally, the method further includes: real-time monitoring of order status; when an order is detected to be accepted by a capacity node, identifying the set of affected grids covered by the revenue pheromone generated by the order, and clearing the revenue pheromone values ​​generated by the order in the set to zero; and performing an evaporation operation on the congestion pheromone values ​​in all grids according to a preset time step, wherein the evaporation operation is to reduce the current congestion pheromone values ​​according to a preset attenuation ratio.

[0020] In the second aspect, a smart logistics order dispatch system includes: processor; The memory stores computer instructions for order dispatching in a smart logistics system, which, when executed by the processor, cause the system to perform the aforementioned smart logistics order dispatching method.

[0021] The beneficial effects of this application are as follows: By establishing a complete dispatch system from grid partitioning, pheromone generation, synthetic pheromone calculation to state transition probability determination, this application achieves an intelligent and adaptive solution to the logistics order dispatch problem. This application provides the logistics industry with an efficient, intelligent, and robust order dispatch solution, effectively addressing the insufficient ability of traditional dispatch methods to cope with spatiotemporal supply and demand imbalances, and significantly improving order response speed and capacity utilization. Attached Figure Description

[0022] Figure 1 This is a flowchart of an order dispatching method for smart logistics according to an embodiment of this application.

[0023] Figure 2 This is a global supply and demand ratio diagram of an intelligent logistics order dispatch method according to an embodiment of this application.

[0024] Figure 3 This is a performance comparison chart of an intelligent logistics order dispatch method according to an embodiment of this application.

[0025] Figure 4 This is a structural block diagram of an intelligent logistics order dispatch system according to an embodiment of this application. Detailed Implementation

[0026] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Figure 1 The diagram shown is a flowchart of an order dispatching method for smart logistics according to an embodiment of this application.

[0027] S1: Divide the logistics area into multiple grids and initialize the revenue pheromone and congestion pheromone parameters.

[0028] The grid division of the logistics area forms the foundational architecture of the entire dispatch system. In implementation, the first step is to construct a two-dimensional raster map of the logistics area, discretizing the continuous geographical space into a regular grid structure. The grid division method includes the following steps: Obtain the geographical boundary coordinates of the logistics area, including four boundary parameters: minimum longitude, maximum longitude, minimum latitude, and maximum latitude. Based on preset grid side length parameters, calculate the number of grids required in the longitude and latitude directions. In this embodiment, the grid side length is set to 500 meters, a scale that ensures sufficient spatial resolution without excessive computational burden due to an excessive number of grids. For a typical urban logistics area covering 10 km × 10 km, it can be divided into 20 × 20 grid units, totaling 400 grid cells. Preferably, in another embodiment, an adaptive grid partitioning strategy can be adopted, using a finer grid size (e.g., 200 meters) in the densely populated urban core area and a coarser grid size (e.g., 1000 meters) in the sparsely populated urban periphery area, thereby reducing overall computational complexity while maintaining control accuracy in the core area.

[0029] Construct an adjacency matrix describing the spatial adjacency relationships of each grid cell. The adjacency matrix is ​​an N×N two-dimensional square matrix, where N is the total number of grid cells, and matrix elements A ij This is used to describe the spatial adjacency relationship between grid i and grid j. If two grids are physically adjacent (including the four orthogonal directions and the four diagonal directions), the corresponding matrix element is assigned a value of 1; otherwise, it is assigned a value of 0. The construction of the adjacency matrix provides the topological foundation for the subsequent spatial diffusion and propagation of pheromones and the state transition of transport nodes between grids. Preferably, in another embodiment, for large-scale logistics areas, sparse matrix storage technology can be used to record only non-zero elements and their position indices, significantly reducing storage space requirements.

[0030] Two dimensions of pheromone data are initialized for each grid, used to store revenue pheromone and congestion pheromone respectively. Revenue pheromone reflects the attractiveness of orders to capacity nodes within the grid; a higher value indicates more substantial order revenue in that area. Congestion pheromone reflects the degree of aggregation of capacity resources within the grid; a higher value indicates greater capacity density in that area. At system startup, the initial values ​​of revenue and congestion pheromone for all grids are uniformly set to zero, indicating no order demand and no capacity aggregation in the initial state. The value range of revenue pheromone is set from 0 to a preset upper limit; in this embodiment, the upper limit is set to 100 to avoid numerical runaway due to long-term accumulation. The value range of congestion pheromone is from 0 to positive infinity, and in actual operation, it typically does not exceed 50.

[0031] Spatial attribute parameters are defined for each grid, including the grid center coordinates and grid coverage area. The grid center coordinates are obtained by the arithmetic mean of the boundary coordinates and are used for subsequent spatial distance calculations. The grid coverage area is calculated using a standard geographic surveying formula. In this embodiment, a grid with a side length of 500 meters corresponds to a coverage area of ​​approximately 0.25 square kilometers. This spatial attribute data is stored in a structured format, providing basic data support for calculations such as density statistics and distance attenuation.

[0032] S2: In response to new order events, generate and update revenue pheromones and congestion pheromones.

[0033] When the logistics system receives a new order request, it immediately triggers the pheromone generation and update process. First, the geographical coordinates of the order are identified, and the grid number to which the order belongs is determined based on the coordinates. A revenue pheromone with a preset initial value is generated at the grid location of the order. The initial value of the revenue pheromone is not a fixed constant but is determined based on multiple attributes of the order. These attributes include business elements such as the order's priority level, delivery fee amount, and timeliness requirement type. In this embodiment, the initial value of the revenue pheromone for ordinary orders is set to 10, for expedited orders (due to stricter timeliness requirements) it is set to 20, and for special orders such as high-value categories like fresh produce and pharmaceuticals it is set to 30. The generation of the revenue pheromone adopts an additive mode; that is, if revenue pheromones from other orders already exist within the grid, the pheromone from the new order will be added to the existing value, reflecting the clustering effect of order demand in that area. When the added value exceeds a preset upper limit, a truncation operation is performed to limit it within the upper limit.

[0034] The revenue pheromone is generated not only in the grid where the order is located, but also diffuses nonlinearly to surrounding grids according to the spatial diffusion coefficient. This mechanism simulates the attraction and radiation effect of orders on surrounding transportation capacity. The specific diffusion process includes: taking the grid where the order is located as the center, determining the set of affected grids within a preset number of hops. In this embodiment, the diffusion range is set to 3, meaning that all neighboring grids within a Manhattan distance of no more than 3 grids are affected, corresponding to an influence radius of approximately 1.5 kilometers in actual distance. In another embodiment, Euclidean distance can be used to define the diffusion range, including grids with a straight-line distance less than a preset threshold in the affected set.

[0035] Calculate the spatial distance between the center of each grid in the affected grid set and the center of the order grid. The distance calculation uses standard formulas for spherical geometry to ensure accurate geodesic distances across different latitudes. Calculate the pheromone gain increment at each affected grid. Pheromones decrease exponentially with distance, and the decay rate is controlled by the spatial diffusion coefficient, calculated using the following formula: ; in, Indicates distance as The incremental pheromone gain obtained at the affected grid j This represents the initial value of the revenue pheromone in the grid where the order is located. Represents the spatial diffusion coefficient. This represents the spatial distance between the centers of the two grids. The exponential decay function ensures that the pheromone intensity decreases rapidly with distance, consistent with the spatial diminishing returns of order attractiveness in real-world scenarios. The value of the spatial diffusion coefficient α directly determines the decay rate: a larger α value results in faster decay and a smaller influence range, while a smaller α value results in slower decay and a larger influence range.

[0036] The spatial diffusion coefficient is not a fixed parameter, but rather dynamically adjusted based on the global supply-demand ratio of the logistics region at the current moment. The global supply-demand ratio is defined as the ratio of the total number of unresponsive orders to the total number of waiting-to-be-scheduled transport capacity nodes. This indicator comprehensively reflects the supply and demand tension of the logistics system. When the supply-demand ratio is high, it indicates that orders are in short supply relative to transport capacity. In this case, the impact range of individual orders should be reduced to avoid attracting long-distance transport capacity and causing delays in the response of short-distance orders. Conversely, when the supply-demand ratio is low, it indicates that transport capacity is relatively abundant. In this case, the impact range of orders should be expanded to increase the responsiveness of long-distance transport capacity. Figure 2 The diagram shown is a global supply-demand ratio chart of an intelligent logistics order dispatch method according to an embodiment of this application.

[0037] The spatial diffusion coefficient is determined using a piecewise mapping mechanism: a first supply-demand ratio threshold and a second supply-demand ratio threshold are preset, which are set to 2.0 and 0.5 respectively in this embodiment. When the global supply-demand ratio is greater than 2.0, the system is determined to be in a high-demand state, and the spatial diffusion coefficient is set to a larger first value, such as 2.0, corresponding to a faster decay rate and a smaller impact range. When the global supply-demand ratio is less than 0.5, the system is determined to be in a low-demand state, and the spatial diffusion coefficient is set to a smaller second value, such as 0.5, corresponding to a slower decay rate and a larger impact range. In another embodiment, a continuous function can be used to achieve a smooth transition. When the supply-demand ratio is between the two thresholds, the diffusion coefficient of the intermediate state is calculated through linear interpolation or a nonlinear mapping function to avoid system oscillations caused by sudden parameter changes.

[0038] The pheromone also has the characteristic of accumulating and increasing over time, ensuring that orders that have not responded for a long time can receive higher scheduling priority. The system continuously monitors the duration of each order awaiting scheduling, i.e., the time interval from the order's creation time to the current time. The time increment of the pheromone is calculated, and this increment is proportional to the duration. This embodiment uses a linear growth model, where the time increment equals the time growth coefficient multiplied by the duration. The time growth coefficient is set to 0.1, meaning that for every 10 seconds an order waits, its pheromone increases by 1 unit. The system periodically performs the increment accumulation operation according to a preset time step, which is set to 10 seconds in this embodiment. The calculated time increment is added to the pheromone value of the grid where the order is located. If the accumulated value exceeds a preset upper limit, it is truncated to the upper limit to prevent a single long-waiting order from dominating the entire scheduling decision.

[0039] While updating pheromones, the density of transport nodes within each grid is acquired. When the density exceeds a preset threshold, a congestion pheromone is generated. The generation process of the congestion pheromone includes: counting the number of transport nodes in each grid at the current moment; transport nodes reporting their location information in real time via vehicle positioning devices or mobile terminals; the system determining the grid to which they belong based on their location coordinates and performing a count; and calculating the transport node density by dividing the number of transport nodes within the grid by the grid coverage area to obtain the transport capacity distribution intensity per unit area, expressed as vehicles per square kilometer.

[0040] The calculated capacity node density is compared with a preset threshold. In this embodiment, the threshold is set to 10 vehicles per square kilometer, corresponding to approximately 2.5 capacity nodes within a grid with a side length of 500 meters. If the capacity node density exceeds this threshold, it is determined that the grid exhibits capacity aggregation, requiring the generation of congestion pheromones to exert a repulsive effect. The generated congestion pheromone value is proportional to the difference between the density exceeding the threshold, calculated using a linear relationship: congestion pheromone equals congestion coefficient multiplied by the excess density. The congestion coefficient reflects the degree of congestion corresponding to a unit density exceeding the standard; in this embodiment, it is set to 2.0. For example, if a grid's capacity density is 15 vehicles per square kilometer, exceeding the threshold by 5 vehicles per square kilometer, then the congestion pheromone is 10 units. If the capacity density does not exceed the threshold, the congestion pheromone remains zero, and no repulsive effect is generated.

[0041] S3: Calculate the synthetic pheromone and determine the state transition probability.

[0042] For each transport node to be dispatched within the logistics area, a synthetic pheromone is calculated. The synthetic pheromone comprehensively considers information from both revenue attraction and congestion repulsion dimensions, providing a quantitative basis for the movement decisions of transport nodes. The calculation process first determines the neighborhood grid set of the current location of the transport node. This set consists of the grid where the transport node is currently located and several adjacent grids. This embodiment uses a 3×3 grid window, including the central grid and its eight adjacent grids, totaling nine candidate locations, corresponding to nine actions that the transport node can perform: maintaining its current location or moving in one of eight directions. In another embodiment, the neighborhood range can be dynamically adjusted according to the movement speed of the transport node. For high-speed transport nodes, the neighborhood is expanded to a 5×5 window; for low-speed or stationary transport nodes, it is reduced to the current grid and four orthogonal adjacent grids.

[0043] Obtain the revenue pheromone and congestion pheromone for each grid in the neighborhood grid set. Calculate the composite evaluation value for each neighborhood grid, which reflects the grid's overall attractiveness to capacity nodes. The calculation formula is as follows: ; Where V j Let w1 and w2 represent the composite evaluation value of grid j, respectively, and τ represent the first and second weighting coefficients. j and η j Let w1 and w2 represent the revenue pheromone and congestion pheromone of grid j, respectively. The first weighting coefficient controls the influence intensity of the revenue pheromone, and the second weighting coefficient controls the suppression intensity of the congestion pheromone. The ratio between the two reflects the system's trade-off strategy between order response and capacity balance. In this embodiment, w1 is set to 1.0 and w2 is set to 0.5, indicating that the system prioritizes order response speed while moderately suppressing capacity aggregation. In another embodiment, the weighting parameters can be dynamically adjusted according to the business period, increasing w1 during peak order periods to speed up response and increasing w2 during off-peak order periods to optimize distribution.

[0044] The combined evaluation values ​​of all grids in the neighborhood grid set constitute the combined pheromone vector of the capacity node. Each element of the vector corresponds to the comprehensive benefit of an optional action; a positive value indicates that the direction is attractive, while a negative value indicates that there is a repulsive effect, and the magnitude of the value reflects the strength of the attraction or repulsion.

[0045] Based on the synthetic pheromone vector, a probabilistic selection algorithm is used to calculate the state transition probability of a transport node moving to its neighboring grids. The calculation process first sets the negative elements in the synthetic pheromone vector to zero, because negative values ​​correspond to grids with excessive congestion and negative net attraction, and transport nodes should not move in these directions. After zeroing, the sum of all elements in the processed vector is calculated. If the sum is greater than zero, it indicates the existence of candidate grids with positive attraction within the neighborhood. In this case, the proportion of the synthetic evaluation value of each grid to the sum is calculated, and this proportion is used as the transition probability. This normalization operation ensures that the sum of all probabilities equals 1, satisfying the mathematical requirements of probability distribution.

[0046] If the combined evaluation values ​​of all grids are non-positive, resulting in a sum of zero, it indicates that there are neither order attraction nor clear guidance signals within the neighborhood. In this case, the system adopts an equal probability strategy, setting the probability of transfer to all neighboring grids to an equal value, for example, allocating a one-eighth probability to each of the eight neighboring grids. This mechanism ensures that capacity nodes can still conduct random walk exploration even in the absence of clear signals, maintaining the system's dynamism and responsiveness to sudden orders.

[0047] S4: Generate dispatch instructions and execute capacity scheduling.

[0048] Based on the state transition probability vector calculated for each capacity node, corresponding dispatch instructions are generated and issued for execution. Instruction generation employs a roulette wheel selection algorithm, which randomly samples based on probability distributions to ensure that high-probability actions have a greater chance of being selected, while retaining the possibility of low-probability actions being selected, thus balancing deterministic utilization with random exploration. Specifically, uniformly distributed random numbers between 0 and 1 are generated, the cumulative distribution function of the probability vector is calculated, the probability interval in which the random number falls is determined, and the corresponding grid is selected as the target location for the capacity node.

[0049] The generated dispatch instruction contains complete scheduling information: capacity node identifier, current location, target grid, recommended route, and estimated arrival time. If there are pending orders within the target grid, the instruction directly includes the order details, instructing the capacity node to pick up the order; if there are no orders in the target grid, the instruction guides the capacity node to move in that direction, entering a potential area where orders are highly likely to occur. The recommended route is generated based on the current road network conditions and traffic information, providing the optimal driving route for the capacity node. The estimated arrival time is estimated based on distance and average speed and is used for subsequent order matching and timeliness monitoring.

[0050] Order dispatch instructions are sent to the mobile terminal devices of the transport nodes in real time via a push notification mechanism. Upon receiving the instruction, the transport node can choose to accept or reject it, and the system records the response for subsequent strategy optimization. When a transport node accepts the instruction and successfully accepts the order, the system immediately updates the order status and performs the corresponding pheromone cleanup operation.

[0051] The system monitors order status changes in real time. When an order is detected as being accepted by a capacity node, it immediately identifies the set of affected grids covered by the revenue pheromones generated by that order and resets all revenue pheromone values ​​generated by that order within that set to zero. This reset operation is achieved by iterating through the list of affected grids and subtracting the order's contribution from the pheromone storage of each grid. If a grid is affected by multiple orders simultaneously, only the pheromone portion of the accepted orders is cleared, while the pheromone from other pending orders is retained. This mechanism avoids resource waste caused by completed orders continuing to attract capacity.

[0052] Following a preset time step, the system performs an evaporation operation on congestion pheromones in all grids. This evaporation simulates the natural dissipation of traffic congestion in the real world; as transport nodes depart and time passes, the congestion level gradually decreases. Evaporation is achieved by multiplying the current congestion pheromone value by a preset attenuation ratio. In this embodiment, the attenuation ratio is set to 0.9, meaning that 90% of the value is retained after each time step. The time step is set to 30 seconds, and the system performs a global evaporation operation every 30 seconds, traversing all grids and updating their congestion pheromones. This dynamic attenuation mechanism ensures that congestion pheromones can reflect the current transport capacity distribution in a timely manner, avoiding the undue influence of historical congestion information on current decisions.

[0053] like Figure 3 The figure shown is a performance comparison chart of an intelligent logistics order dispatch method according to an embodiment of this application. Figure 3 This paper presents a comparison of the proposed method with traditional greedy algorithms and random walk strategies in terms of the key metric of average order waiting time, under the same logistics area and order flow input conditions. Experimental data shows that the random walk strategy results in the highest average order waiting time due to the lack of clear target guidance for capacity nodes. While the greedy algorithm (which only considers the closest node) improves performance, it easily leads to local capacity clustering and competition, resulting in an average waiting time of approximately 15.4 seconds. In contrast, the proposed method attracts capacity through the nonlinear diffusion of revenue pheromones and avoids excessive capacity aggregation using congestion pheromones, achieving a dynamic balance between global supply and demand. The average order waiting time of the proposed method is significantly reduced to approximately 8.5 seconds. This fully demonstrates the significant advantages of this application in improving the scheduling efficiency and response speed of logistics systems.

[0054] According to a second aspect of this application, this application also provides an intelligent logistics order dispatch system. Figure 4 This is a structural block diagram of an intelligent logistics order dispatch system according to an embodiment of this application. Figure 4As shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an order dispatching method for smart logistics according to the first aspect of this application. The system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface. Their configurations and functions are known in the art and will not be described further here.

[0055] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in this application, based on the technical solution and concept of this application, should be within the scope of protection of this application.

Claims

1. A smart logistics order dispatching method, characterized in that, The method includes: The logistics area is divided into multiple grids, and each grid includes two parameters: revenue pheromone and congestion pheromone. In response to a new order event, a revenue pheromone with a preset initial value is generated in the grid where the order is located; the density of transport nodes in each grid is obtained, and a congestion pheromone is generated when the density of transport nodes exceeds a preset threshold; the order event includes at least the grid where the order is located and the order generation time. For each capacity node, a synthetic pheromone within a preset grid range is calculated. The synthetic pheromone is obtained by weighting the revenue pheromone and the congestion pheromone. Based on the synthetic pheromone, a probabilistic selection algorithm is used to calculate the state transition probability of the capacity node moving to an adjacent grid. A dispatch instruction is generated according to the state transition probability of each capacity node. The revenue pheromone decays nonlinearly to the surrounding grid based on the spatial diffusion coefficient within the logistics area at the current moment; the spatial diffusion coefficient is determined based on the global supply-demand ratio at the current moment, and the global supply-demand ratio is the ratio between the total number of unresponsive orders and the total number of transport capacity nodes to be scheduled within the current logistics area.

2. The intelligent logistics order dispatch method according to claim 1, characterized in that, The method of dividing the logistics area into multiple grids includes: Construct a two-dimensional grid map of the logistics area and establish an adjacency matrix describing the spatial adjacency relationships of each grid. Initialize two dimensions of data for each grid, to store the gain pheromone and congestion pheromone respectively, and set the initial values ​​of the gain pheromone and congestion pheromone for all grids to zero; Define the spatial attributes for each grid, including the grid center coordinates and the grid coverage area.

3. The intelligent logistics order dispatch method according to claim 1, characterized in that, The nonlinear decay of the gain pheromone towards the surrounding grid includes: Centered on the grid where the order is located, determine the set of affected grids within a preset range; Calculate the spatial distance between the center of each grid in the affected grid set and the center of the grid where the order is located; Calculate the gain pheromone at each affected grid, wherein the gain pheromone decreases exponentially with the spatial distance, and the rate of decrease is controlled by the spatial diffusion coefficient as a parameter.

4. The intelligent logistics order dispatch method according to claim 1, characterized in that, The calculation process of the spatial diffusion coefficient includes: A first supply-demand ratio threshold and a second supply-demand ratio threshold are preset, wherein the first supply-demand ratio threshold is greater than the second supply-demand ratio threshold; When the global supply-demand ratio is greater than the first supply-demand ratio threshold, the spatial diffusion coefficient is set to a first value, which corresponds to a slower decay rate. When the global supply-demand ratio is less than the second supply-demand ratio threshold, the spatial diffusion coefficient is set to a second value, which corresponds to a faster decay rate; wherein, the first value is greater than the second value.

5. The intelligent logistics order dispatch method according to claim 1, characterized in that, The method also includes the fact that the revenue pheromone is positively correlated with order waiting time: Monitor the duration of orders awaiting scheduling based on their creation time; The time increment of the yield pheromone is calculated, and the time increment is positively correlated with the duration. The time increment is superimposed on the pheromone value of the grid where the order is located; If the combined benefit pheromone exceeds the preset upper limit, then the benefit pheromone will be set to the preset upper limit.

6. The intelligent logistics order dispatch method according to claim 1, characterized in that, The step of generating congestion pheromones when the density of transport nodes exceeds a preset threshold includes: Count the number of transport nodes in each grid at the current moment; Calculate the capacity node density, which is the number of capacity nodes divided by the coverage area of ​​the grid; The density of the transportation nodes is numerically compared with a preset threshold. If the density of transport nodes is greater than the preset threshold, the generated congestion pheromone value is positively correlated with the difference between the density of transport nodes and the preset threshold; otherwise, the congestion pheromone value is zero.

7. The intelligent logistics order dispatch method according to claim 1, characterized in that, The calculation process for the synthetic pheromone includes: Determine the neighborhood grid set of the current location of each capacity node to be scheduled. The neighborhood grid set is composed of the grid where the capacity node is currently located and multiple neighboring grids. Obtain the benefit pheromone and congestion pheromone for each grid in the neighborhood grid set; Calculate the composite evaluation value for each grid, which is equal to the benefit pheromone multiplied by the first weighting coefficient minus the congestion pheromone multiplied by the second weighting coefficient; The composite evaluation value of each grid in the neighborhood grid set constitutes the composite pheromone.

8. The intelligent logistics order dispatch method according to claim 1, characterized in that, The calculation process for the state transition probability includes: Set the negative elements in the synthesized pheromone to zero; Calculate the sum of all elements in the processed synthetic pheromone; If the sum is greater than zero, the ratio of the composite evaluation value of each grid to the sum is calculated, and the ratio is used as the probability of the capacity node being transferred to the corresponding grid. If the sum is equal to zero, then the probability of a capacity node being transferred to all adjacent grids is set to an equal value.

9. The intelligent logistics order dispatch method according to claim 1, characterized in that, The method further includes: Real-time monitoring of order status; Once an order is detected to be accepted by a capacity node, the set of affected grids covered by the revenue pheromone generated by the order is identified, and the value of the revenue pheromone generated by the order in that set is cleared to zero. According to the preset time step, the evaporation operation is performed on the congestion pheromone values ​​in all grids. The evaporation operation involves reducing the current congestion pheromone value by a preset attenuation ratio.

10. An intelligent logistics order dispatch system, characterized in that, include: processor; A memory, wherein a computer program is stored; Wherein, the processor is configured to execute the computer program to implement an order dispatching method for smart logistics as described in any one of claims 1 to 9.

Images