Collaborative vehicle scheduling optimization method based on shared warehouse

Abstract

The invention relates to a shared warehouse-based collaborative vehicle scheduling optimization method, which comprises the following steps of: based on shared warehouse information, grouping warehouses by adopting a clustering algorithm to obtain a plurality of warehouse clusters; wherein the shared warehouse information comprises geographic position distribution and cargo storage amount information of the shared warehouse; in each warehouse cluster, according to real-time road traffic condition information, an ant colony algorithm is adopted to dynamically plan a driving path of a vehicle; in the shared warehouse, according to the layout characteristics of the warehouse and the cargo distribution condition, a genetic algorithm is adopted to optimize the loading and unloading route of the vehicle, and the optimal loading and unloading route is obtained; in the scheduling task execution process, the scheduling strategy is updated in real time through an online learning algorithm according to the cargo distribution change condition. According to the method, the transportation time can be effectively shortened, the scheduling accuracy and real-time performance are improved, the energy consumption is reduced, and the overall logistics efficiency and the scheduling stability are improved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a collaborative vehicle scheduling optimization method based on a shared warehouse. Background Art

[0002] In the collaborative vehicle scheduling optimization method based on shared warehouses, vehicle vector optimization is a key technical issue. Due to the dispersed geographical locations of shared warehouses and the uneven distribution of goods, vehicles often encounter problems such as unreasonable path planning and insufficient dynamic obstacle avoidance when performing scheduling tasks, which leads to increased vehicle travel distance, longer transportation time, and reduced transportation efficiency.

[0003] Specifically, when vehicles transport goods back and forth between shared warehouses, how to dynamically optimize the driving path based on factors such as real-time traffic conditions and changes in cargo distribution is a technical problem that needs to be solved urgently. Traditional static path planning methods can no longer meet the needs of real-time scheduling, and it is necessary to introduce intelligent algorithms and big data analysis technologies to achieve dynamic optimization of vehicle driving paths. At the same time, during the driving process, it is also necessary to perceive changes in road conditions in real time, avoid congested sections and accident-prone areas in a timely manner, and ensure that the vehicle can reach its destination safely and efficiently.

[0004] In addition, when vehicles are loading and unloading in shared warehouses, they also need to plan their routes reasonably to reduce unnecessary repeated movements and avoid mutual interference, so as to improve work efficiency. This requires optimizing the driving vectors of vehicles in the warehouse based on the layout characteristics of shared warehouses, so as to achieve precise scheduling and intelligent collaboration. Summary of the invention

[0005] The purpose of the present invention is to propose a collaborative vehicle scheduling optimization method based on a shared warehouse to solve the problems existing in the above-mentioned prior art.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A collaborative vehicle scheduling optimization method based on a shared warehouse, comprising:

[0008] Based on the shared warehouse information, a clustering algorithm is used to group the warehouses to obtain a number of warehouse clusters; wherein the shared warehouse information includes: geographical location distribution and cargo storage volume information of the shared warehouses;

[0009] In each warehouse cluster, the ant colony algorithm is used to dynamically plan the vehicle's driving path based on real-time road traffic information;

[0010] Inside the shared warehouse, according to the layout characteristics of the warehouse and the distribution of goods, a genetic algorithm is used to optimize the loading and unloading routes of vehicles to obtain the optimal loading and unloading routes;

[0011] During the execution of the dispatching task, the dispatching strategy is updated in real time using the online learning algorithm according to the changes in the distribution of goods.

[0012] Optionally, obtain several warehouse clusters including:

[0013] Collect the longitude and latitude coordinate information of shared warehouses and count the cargo storage data of each warehouse;

[0014] The K-means clustering algorithm is used to set the number of clusters. The Euclidean distance between the longitude and latitude coordinates of the warehouses and the difference in cargo storage volume are calculated, and the cluster centers are iteratively optimized. Finally, warehouses that meet the preset distance and the preset cargo storage volume are divided into the same cluster.

[0015] Optionally, dynamically planning the vehicle's driving path using an ant colony algorithm includes:

[0016] Acquire real-time traffic status information of the road network within the warehouse cluster, wherein the real-time traffic status information includes the travel time and traffic flow data of each node and edge;

[0017] Initialize the pheromone concentration of each node and edge in the road network within the warehouse cluster according to the real-time traffic status information, wherein the pheromone concentration is inversely proportional to the travel time and directly proportional to the traffic flow;

[0018] The ant colony algorithm is used to dynamically plan the vehicle's driving path, simulating ants walking in the road network, and determining the next direction of each ant's movement based on the pheromone concentration of the nodes and edges;

[0019] In each iteration of the ant colony algorithm, the pheromone concentration of each node and edge on the path is updated according to the path length and travel time of the ants, wherein the shorter the travel time and the smaller the path length, the greater the increase in pheromone concentration;

[0020] In the iterative process of the ant colony algorithm, a pheromone evaporation mechanism is introduced. If a node or edge is not passed by ants, its pheromone concentration will gradually decrease over time.

[0021] After several iterations, the path with pheromone concentration higher than the preset threshold is obtained as the recommended driving path for the vehicle under the current traffic conditions.

[0022] Optionally, a genetic algorithm is used to optimize the loading and unloading routes of vehicles, and obtaining the optimal loading and unloading routes includes:

[0023] Based on the loading and unloading locations and routes, construct a loading and unloading route map for the shared warehouse;

[0024] Designing a fitness function of a genetic algorithm for the loading and unloading route map; wherein the fitness function combines the distance, time, and congestion level of the loading and unloading route;

[0025] According to the fitness function, an initial population is randomly generated, wherein each individual in the initial population corresponds to a complete loading and unloading route;

[0026] Using a roulette wheel selection algorithm, the initial population is evaluated according to the fitness function, and selection, crossover and mutation operations are performed to obtain a mutated loading and unloading route;

[0027] According to the preset termination conditions, the selection, crossover and mutation operations are repeated until the optimal loading and unloading route is obtained.

[0028] Optionally, constructing a loading and unloading route map of the shared warehouse includes:

[0029] Constructing a digital warehouse model of the shared warehouse, and mapping loading and unloading locations and routes to the digital warehouse model;

[0030] Based on the digital model of the warehouse, a graph theory algorithm is used to define the loading and unloading locations as nodes of a graph, and the routes as edges of a graph, so as to construct a loading and unloading route map of the shared warehouse.

[0031] Optionally, perform selection, crossover, and mutation operations, including:

[0032] Randomly select two individual fragments to exchange and combine to generate new loading and unloading routes;

[0033] A mutation operation is performed on the new loading and unloading route, nodes in the new loading and unloading route are randomly replaced or deleted, a new path is introduced, and a mutated loading and unloading route is obtained.

[0034] Optionally, using an online learning algorithm to update the scheduling strategy in real time includes:

[0035] Obtain real-time feedback information based on changes in cargo distribution as input data for online learning algorithms;

[0036] An online learning algorithm is used to process the input feedback information to obtain the characteristic representation of the change in cargo distribution;

[0037] According to the characteristic representation of cargo distribution changes, determine whether the current scheduling strategy needs to be optimized and adjusted; if optimization and adjustment are needed, generate a new scheduling strategy through online learning algorithm to replace the original scheduling strategy;

[0038] The new scheduling strategy is applied to the execution of scheduling tasks to schedule goods in real time.

[0039] Optionally, updating the scheduling strategy in real time using an online learning algorithm also includes:

[0040] During the execution of the scheduling task, feedback information on changes in cargo distribution is continuously obtained as input data for the next online learning.

[0041] The beneficial effects of the present invention are:

[0042] The present invention discloses a collaborative vehicle scheduling optimization method based on shared warehouses. The method first clusters and groups shared warehouses to form warehouse clusters with similar geographical locations and cargo storage volumes to reduce long-distance transportation. In each cluster, an ant colony algorithm is used to dynamically plan vehicle routes based on real-time traffic conditions. A genetic algorithm is used to optimize loading and unloading routes inside the warehouse. At the same time, the present invention introduces an online learning algorithm to update scheduling strategies in real time to adapt to changes in cargo distribution. The present invention can effectively shorten transportation time, improve scheduling accuracy and real-time performance, reduce energy consumption, and improve overall logistics efficiency and scheduling stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A schematic flow chart of a collaborative vehicle scheduling optimization method based on a shared warehouse according to an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of a method for dynamically planning a vehicle's driving path using an ant colony algorithm according to an embodiment of the present invention;

[0046] Figure 3 A schematic flow chart of a method for optimizing vehicle loading and unloading routes by using a genetic algorithm to obtain an optimal loading and unloading route according to an embodiment of the present invention;

[0047] Figure 4 The present invention is a flowchart of a method for updating a scheduling strategy in real time by using an online learning algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] like Figure 1 As shown, this embodiment proposes a collaborative vehicle scheduling optimization method based on a shared warehouse, including:

[0051] Based on the shared warehouse information, a clustering algorithm is used to group the warehouses to obtain a number of warehouse clusters; wherein the shared warehouse information includes: geographical location distribution and cargo storage volume information of the shared warehouses;

[0052] In each warehouse cluster, the ant colony algorithm is used to dynamically plan the vehicle's driving path based on real-time road traffic information;

[0053] Inside the shared warehouse, according to the layout characteristics of the warehouse and the distribution of goods, a genetic algorithm is used to optimize the loading and unloading routes of vehicles to obtain the optimal loading and unloading routes;

[0054] During the execution of the dispatching task, the dispatching strategy is updated in real time using the online learning algorithm according to the changes in the distribution of goods.

[0055] Specifically, this embodiment first clusters and groups shared warehouses to form warehouse clusters with similar geographical locations and cargo storage volumes to reduce long-distance transportation; within each cluster, the ant colony algorithm is used to dynamically plan vehicle routes based on real-time traffic conditions; a genetic algorithm is used to optimize loading and unloading routes within the warehouse; at the same time, the present invention introduces an online learning algorithm to update the scheduling strategy in real time to adapt to changes in cargo distribution. This embodiment can effectively shorten transportation time, improve scheduling accuracy and real-time performance, reduce energy consumption, and improve overall logistics efficiency and scheduling stability.

[0056] Furthermore, several warehouse clusters are obtained including:

[0057] Collect the longitude and latitude coordinate information of shared warehouses and count the cargo storage data of each warehouse;

[0058] The K-means clustering algorithm is used to set the number of clusters. The Euclidean distance between the longitude and latitude coordinates of the warehouses and the difference in cargo storage volume are calculated, and the cluster centers are iteratively optimized. Finally, warehouses that meet the preset distance and the preset cargo storage volume are divided into the same cluster.

[0059] Specifically, in this embodiment, the latitude and longitude coordinate information of several target shared warehouses is first obtained through GPS positioning technology, and the cargo storage data of each warehouse is counted in real time, and these data are used as input features of the clustering algorithm. Then the K-means clustering algorithm is used, and the number of clusters k=20 is set. By calculating the Euclidean distance between the latitude and longitude coordinates of the warehouses and the difference in cargo storage, the cluster center is iteratively optimized, and finally the warehouses with similar distances and similar cargo storage are divided into the same cluster. After the clustering is completed, the vehicle transportation route is optimized according to the clustering results, and vehicle scheduling and cargo distribution are preferentially performed within the same cluster. For example, if the average distance between warehouses within a cluster is 50 kilometers and the cargo storage capacity is about 1,000 tons, the cargo distribution within the cluster can be efficiently transported by vehicles within the cluster. When the cargo storage capacity of the warehouse within the cluster is insufficient, the distance and cargo storage capacity difference with the adjacent cluster are automatically calculated, and the cluster with the smallest span is selected for cross-cluster vehicle scheduling to replenish the supply of goods. Through the above clustering optimization method, the shared warehouses within the target range can be divided into several regional clusters, which can significantly improve the efficiency of vehicle scheduling and cargo distribution within the region. At the same time, through cross-cluster vehicle scheduling, efficient collaboration between clusters can also be achieved to ensure the stability of cargo supply.

[0060] Furthermore, the ant colony algorithm is used to dynamically plan the vehicle's driving path, such as Figure 2 As shown, including:

[0061] Acquire real-time traffic status information of the road network within the warehouse cluster, wherein the real-time traffic status information includes the travel time and traffic flow data of each node and edge;

[0062] Initialize the pheromone concentration of each node and edge in the road network within the warehouse cluster according to the real-time traffic status information, wherein the pheromone concentration is inversely proportional to the travel time and directly proportional to the traffic flow;

[0063] The ant colony algorithm is used to dynamically plan the vehicle's driving path, simulating ants walking in the road network, and determining the next direction of each ant's movement based on the pheromone concentration of the nodes and edges;

[0064] In each iteration of the ant colony algorithm, the pheromone concentration of each node and edge on the path is updated according to the path length and travel time of the ants, wherein the shorter the travel time and the smaller the path length, the greater the increase in pheromone concentration;

[0065] In the iterative process of the ant colony algorithm, a pheromone evaporation mechanism is introduced. If a node or edge is not passed by ants, its pheromone concentration will gradually decrease over time.

[0066] After several iterations, the path with pheromone concentration higher than the preset threshold is obtained as the recommended driving path for the vehicle under the current traffic conditions.

[0067] Specifically, in this embodiment, in order to obtain the real-time traffic status information of the road network in the warehouse cluster, the travel time and traffic flow data of each node and edge in the road network can be obtained every 5 minutes. According to the real-time traffic status information obtained, the pheromone concentration of each node and edge in the road network is initialized. Among them, the travel time is inversely proportional to the pheromone concentration. The longer the travel time, the lower the pheromone concentration; the traffic flow is proportional to the pheromone concentration. The larger the flow, the higher the pheromone concentration. After the initialization is completed, the ant colony algorithm is used to dynamically plan the vehicle driving path. The algorithm simulates 100 ants walking in the road network. Each ant uses the roulette selection method to determine the direction of the next step according to the pheromone concentration of the edge connected to the current node. In each iteration, the algorithm updates the pheromone concentration of each node and edge on the path according to the path length and travel time of the ant. Among them, the shorter the travel time and the smaller the path length, the greater the increase in the pheromone concentration, and the increase is 10% of the original pheromone concentration. At the same time, the algorithm introduces a pheromone evaporation mechanism. If a node or edge has not been passed by ants for three consecutive iterations, its pheromone concentration will gradually decrease by 5% each time. After 50 iterations, the path with a pheromone concentration higher than 8 is obtained, and it is used as the preferred path under the current traffic conditions and recommended to the vehicle as a driving path. In order to adapt to changes in real-time traffic conditions, road traffic conditions are continuously monitored. When the travel time or traffic flow changes by more than 20%, the ant colony algorithm is triggered to re-execute and update the vehicle driving path to ensure the real-time and effectiveness of path planning.

[0068] Furthermore, a genetic algorithm is used to optimize the vehicle loading and unloading routes to obtain the optimal loading and unloading routes; Figure 3 Shown; including:

[0069] Based on the loading and unloading locations and routes, construct a loading and unloading route map for the shared warehouse;

[0070] Designing a fitness function of a genetic algorithm for the loading and unloading route map; wherein the fitness function combines the distance, time, and congestion level of the loading and unloading route;

[0071] According to the fitness function, an initial population is randomly generated, wherein each individual in the initial population corresponds to a complete loading and unloading route;

[0072] Using a roulette wheel selection algorithm, the initial population is evaluated according to the fitness function, and selection, crossover and mutation operations are performed to obtain a mutated loading and unloading route;

[0073] According to the preset termination conditions, the selection, crossover and mutation operations are repeated until the optimal loading and unloading route is obtained.

[0074] The loading and unloading route map for constructing the shared warehouse includes:

[0075] Constructing a digital warehouse model of the shared warehouse, and mapping loading and unloading locations and routes to the digital warehouse model;

[0076] Based on the digital model of the warehouse, a graph theory algorithm is used to define the loading and unloading locations as nodes of a graph, and the routes as edges of a graph, so as to construct a loading and unloading route map of the shared warehouse.

[0077] Perform selection, crossover, and mutation operations, including:

[0078] Randomly select two individual fragments to exchange and combine to generate new loading and unloading routes;

[0079] A mutation operation is performed on the new loading and unloading route, nodes in the new loading and unloading route are randomly replaced or deleted, a new path is introduced, and a mutated loading and unloading route is obtained.

[0080] Specifically, in this embodiment, in order to optimize the loading and unloading routes of the shared warehouse, a three-dimensional digital model is first established based on the actual layout and distribution of goods in the warehouse. Through laser scanning and visual recognition technology, the internal structure of the warehouse, shelf location, channel width and other information are accurately collected, and the loading and unloading locations and routes are mapped to the model. Then, the loading and unloading locations are abstracted as nodes of the graph using the graph theory algorithm, and the routes connecting these locations are abstracted as edges of the graph to construct a complete loading and unloading route map. On this basis, the fitness function of the genetic algorithm is designed to comprehensively consider factors such as the distance, time and congestion degree of the loading and unloading routes. Among them, the distance weight is 4, the time weight is 4, and the congestion degree weight is 2. According to the fitness function, 100 initial populations are randomly generated, and each individual corresponds to a complete loading and unloading route. The roulette selection algorithm is used to select individuals with higher fitness with a probability of 8, and individuals with lower fitness with a probability of 2. The fragments of two individuals are randomly selected for crossover operation to generate a new loading and unloading route. The new route is mutated, and the nodes in the route are randomly replaced or deleted with a probability of 0.5 to introduce a new path. Repeat the selection, crossover and mutation operations 500 times until the average fitness of the population stabilizes and the optimal loading and unloading route is obtained. Finally, the optimal route is transmitted to the AGV and loading and unloading robots to guide them to complete the loading and unloading operations efficiently, thereby improving the operational efficiency of the shared warehouse.

[0081] Furthermore, using a graph theory algorithm, the loading and unloading locations are defined as nodes of a graph, and the routes are defined as edges of the graph. Constructing a loading and unloading route map of the shared warehouse includes:

[0082] According to the coordinates of the loading and unloading locations, a graph theory algorithm is used to abstract the locations into nodes of a graph to form a node set; according to the loading and unloading routes, the routes are abstracted into edges of a graph to form an edge set; a loading and unloading route map is constructed based on the node set and the edge set; a depth-first search algorithm is used to traverse the loading and unloading route map to determine the connectivity of the graph; if the graph is connected, the Dijkstra algorithm is used to calculate the shortest path between the nodes in the loading and unloading route map; based on the calculated shortest path, combined with the loading and unloading time window constraints, the optimal loading and unloading route is determined; the optimal loading and unloading route is issued to logistics vehicles to guide actual loading and unloading operations and improve logistics distribution efficiency.

[0083] Specifically, in a shared warehouse, first, based on the coordinate information of the loading and unloading locations, a graph theory algorithm is used to abstract each location into a node of the graph, such as A(1,2), B(3,4), C(5,6), etc., to form a node set V = {A, B, C, ...}. Then, based on the loading and unloading routes, the routes are abstracted into edges of the graph, such as routes AB, BC, AC, etc., to form an edge set E = {(A, B), (B, C), (A, C), ...}. Based on the node set V and the edge set E, an undirected graph of loading and unloading routes G = (V, E) is constructed. Then, a depth-first search algorithm is used to traverse the entire graph G from any node to determine the connectivity of the graph. If the graph G is connected, the Dijkstra algorithm is used to calculate the shortest path between any two nodes in the graph G. For example, the shortest path from node A to node C is ABC, with a length of 10 meters. Finally, combined with the loading and unloading time window constraints, such as 8:00-10:00, determine the optimal loading and unloading route, such as ABCDE, with a total length of at least 50 meters to meet the time window requirements. Automatically send the optimal route to logistics vehicles, such as AGV vehicles, to guide them to perform loading and unloading operations according to the optimal route, improve logistics distribution efficiency, and save time costs. Through graph theory algorithms and depth-first search, shortest path and other algorithms, it is possible to scientifically and rationally plan loading and unloading routes, reduce repeated movement and mutual interference of vehicles within the warehouse, and improve the operational efficiency of shared warehouses.

[0084] Furthermore, the scheduling strategy is updated in real time using online learning algorithms, such as Figure 4 As shown, including:

[0085] Obtain real-time feedback information based on changes in cargo distribution as input data for online learning algorithms;

[0086] An online learning algorithm is used to process the input feedback information to obtain the characteristic representation of the change in cargo distribution;

[0087] According to the characteristic representation of cargo distribution changes, determine whether the current scheduling strategy needs to be optimized and adjusted; if optimization and adjustment are needed, generate a new scheduling strategy through online learning algorithm to replace the original scheduling strategy;

[0088] The new scheduling strategy is applied to the execution of scheduling tasks to schedule goods in real time.

[0089] Using online learning algorithms to update scheduling strategies in real time also includes:

[0090] During the execution of the scheduling task, feedback information on changes in cargo distribution is continuously obtained as input data for the next online learning.

[0091] Specifically, in this embodiment, according to the change in cargo distribution, real-time feedback information is obtained as input data for the online learning algorithm. The input feedback information is processed by the online learning algorithm to obtain a characteristic representation of the change in cargo distribution. According to the characteristic representation of the change in cargo distribution, it is determined whether the current scheduling strategy needs to be optimized and adjusted. If optimization and adjustment are required, a new scheduling strategy is generated by the online learning algorithm to replace the original scheduling strategy. The new scheduling strategy is applied to the execution of the scheduling task to schedule the cargo in real time. During the execution of the scheduling task, feedback information on the change in cargo distribution is continuously obtained as input data for the next online learning. Repeat the above steps, continuously iterate and optimize the scheduling strategy, so that the scheduling method can adapt to the dynamically changing cargo distribution and improve the real-time and accuracy of the scheduling.

[0092] More specifically, in the shared warehouse, infrared sensors are installed in the cargo distribution area to obtain the changes in cargo distribution in real time. The sensors collect data every 5 seconds and transmit the data to the data processing center. The data processing center uses the online learning algorithm FTRL to process the input feedback information and extract the feature vector of cargo distribution changes, including cargo quantity, location, volume and other attributes. According to the extracted feature vector, the decision tree algorithm is used to determine whether the current scheduling strategy needs to be optimized. If the output probability of the classifier is greater than 8, the optimization process of the scheduling strategy is triggered. In the optimization process, the reinforcement learning algorithm DQN is used to generate a new scheduling strategy, and the optimal scheduling path and strategy are learned by continuously interacting with the environment. The reward function of the DQN algorithm comprehensively considers factors such as cargo distribution, scheduling efficiency, and time cost, and updates the parameters of the Q value network through the gradient descent method to obtain the optimal strategy. The generated new strategy is loaded into the scheduling method, and a scheduling instruction is issued to the AGV car to transport cargo according to the new strategy. During the scheduling process, the changes in cargo distribution are continuously monitored, and the feedback information is input into the online learning module every 10 minutes for the next round of strategy iteration optimization. Through the close integration of online learning and scheduling execution, the scheduling method can quickly respond to dynamic changes in cargo distribution, improve scheduling efficiency and reduce logistics costs.

[0093] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A collaborative vehicle scheduling optimization method based on a shared warehouse, characterized in that: include: Based on the shared warehouse information, a clustering algorithm is used to group warehouses to obtain several warehouse clusters; The shared warehouse information includes: geographical location distribution and cargo storage volume information of the shared warehouse; In each warehouse cluster, the ant colony algorithm is used to dynamically plan the vehicle's driving path based on real-time road traffic information; Inside the shared warehouse, according to the layout characteristics of the warehouse and the distribution of goods, a genetic algorithm is used to optimize the loading and unloading routes of vehicles to obtain the optimal loading and unloading routes; During the execution of the dispatching task, the dispatching strategy is updated in real time using the online learning algorithm according to the changes in the distribution of goods.

2. The collaborative vehicle scheduling optimization method based on shared warehouse according to claim 1 is characterized in that: Get several warehouse clusters including: Collect the longitude and latitude coordinate information of shared warehouses and count the cargo storage data of each warehouse; The K-means clustering algorithm is used to set the number of clusters. The Euclidean distance between the longitude and latitude coordinates of the warehouses and the difference in cargo storage volume are calculated, and the cluster centers are iteratively optimized. Finally, warehouses that meet the preset distance and the preset cargo storage volume are divided into the same cluster.

3. The collaborative vehicle scheduling optimization method based on shared warehouse according to claim 1 is characterized in that: The dynamic planning of vehicle driving paths using ant colony algorithm includes: Acquire real-time traffic status information of the road network within the warehouse cluster, wherein the real-time traffic status information includes the travel time and traffic flow data of each node and edge; Initialize the pheromone concentration of each node and edge in the road network within the warehouse cluster according to the real-time traffic status information, wherein the pheromone concentration is inversely proportional to the travel time and directly proportional to the traffic flow; The ant colony algorithm is used to dynamically plan the vehicle's driving path, simulating ants walking in the road network, and determining the next direction of each ant's movement based on the pheromone concentration of the nodes and edges; In each iteration of the ant colony algorithm, the pheromone concentration of each node and edge on the path is updated according to the path length and travel time of the ants, wherein the shorter the travel time and the smaller the path length, the greater the increase in pheromone concentration; In the iterative process of the ant colony algorithm, a pheromone evaporation mechanism is introduced. If a node or edge is not passed by ants, its pheromone concentration will gradually decrease over time. After several iterations, the path with pheromone concentration higher than the preset threshold is obtained as the recommended driving path for the vehicle under the current traffic conditions.

4. The collaborative vehicle scheduling optimization method based on a shared warehouse according to claim 1 is characterized in that: Genetic algorithm is used to optimize the loading and unloading routes of vehicles. The optimal loading and unloading routes include: Based on the loading and unloading locations and routes, construct a loading and unloading route map for the shared warehouse; Designing a fitness function of a genetic algorithm for the loading and unloading route map; wherein the fitness function combines the distance, time, and congestion level of the loading and unloading route; According to the fitness function, an initial population is randomly generated, wherein each individual in the initial population corresponds to a complete loading and unloading route; Using a roulette wheel selection algorithm, the initial population is evaluated according to the fitness function, and selection, crossover and mutation operations are performed to obtain a mutated loading and unloading route; According to the preset termination conditions, the selection, crossover and mutation operations are repeated until the optimal loading and unloading route is obtained.

5. The collaborative vehicle scheduling optimization method based on shared warehouse according to claim 4 is characterized in that: The loading and unloading route map for constructing the shared warehouse includes: Constructing a digital warehouse model of the shared warehouse, and mapping loading and unloading locations and routes to the digital warehouse model; Based on the digital model of the warehouse, a graph theory algorithm is used to define the loading and unloading locations as nodes of a graph, and the routes as edges of a graph, so as to construct a loading and unloading route map of the shared warehouse.

6. The collaborative vehicle scheduling optimization method based on shared warehouse according to claim 4 is characterized in that: Perform selection, crossover, and mutation operations, including: Randomly select two individual fragments to exchange and combine to generate new loading and unloading routes; A mutation operation is performed on the new loading and unloading route, nodes in the new loading and unloading route are randomly replaced or deleted, a new path is introduced, and a mutated loading and unloading route is obtained.

7. The collaborative vehicle scheduling optimization method based on shared warehouse according to claim 1 is characterized in that: Using online learning algorithms to update scheduling strategies in real time includes: Obtain real-time feedback information based on changes in cargo distribution as input data for online learning algorithms; An online learning algorithm is used to process the input feedback information to obtain the characteristic representation of the change in cargo distribution; According to the characteristic representation of cargo distribution changes, determine whether the current scheduling strategy needs to be optimized and adjusted; if optimization and adjustment are needed, generate a new scheduling strategy through online learning algorithm to replace the original scheduling strategy; The new scheduling strategy is applied to the execution of scheduling tasks to schedule goods in real time.

8. The collaborative vehicle scheduling optimization method based on shared warehouse according to claim 7 is characterized in that: Using online learning algorithms to update scheduling strategies in real time also includes: During the execution of the scheduling task, feedback information on changes in cargo distribution is continuously obtained as input data for the next online learning.

Images