Warehouse logistics scheduling method and system based on cloud computing

Abstract

The invention relates to a warehouse logistics scheduling method and system based on cloud computing, and relates to the technical field of warehouse logistics scheduling, and the method comprises the steps: jointly building a logistics scheduling model through a plurality of edge cloud computing nodes, and the logistics scheduling model comprises an edge scheduling scheme generation model and a comprehensive scheduling scheme generation model; determining at least one first optimal edge cloud computing node based on the computing power resource demand of the warehouse logistics scheduling task; each first optimal edge cloud computing node generates a model through an edge scheduling scheme, generates a plurality of candidate warehouse logistics scheduling schemes based on the warehouse logistics scheduling task, the warehouse node state information and the logistics vehicle state information, screens the plurality of candidate warehouse logistics scheduling schemes, and determines a plurality of target warehouse logistics scheduling schemes; through the comprehensive scheduling scheme generation model, the optimal warehouse logistics scheduling scheme is generated based on the multiple target warehouse logistics scheduling schemes, warehouse logistics scheduling is carried out, and the method has the advantage of improving the intelligent level of warehouse logistics scheduling.

Description

Technical Field

[0001] The present invention relates to the technical field of warehousing logistics scheduling, and in particular to a warehousing logistics scheduling method and system based on cloud computing. Background Art

[0002] In the supply chain of the circulation industry, storage, custody of goods and logistics distribution are the main business forms of warehousing and logistics, which are composed of a series of supply and demand. In the prior art, during the operation, the goods stored in the warehouse at each distribution point are loaded into the corresponding logistics vehicles and arranged to be transported to the corresponding distribution points, which is called the dispatching of logistics vehicles. At present, the vehicle dispatching of warehousing and logistics mainly relies on manual dispatching, that is, manually specifying which goods are assembled into which vehicles and transported to which nodes according to the current inventory of goods and the status of the vehicle. The main defects of manual warehousing and logistics dispatching are as follows: low dispatching efficiency. During the dispatching process, only manual dispatching is performed according to the storage status and vehicle status of the goods, and the actual transportation status of the goods and the status of the vehicle are not considered, resulting in low dispatching efficiency and failure to make loading decisions in time; high logistics costs. There is a certain correlation between the size, weight, destination of the goods and the size, load, and transportation route of the vehicle. A better match can greatly reduce the logistics transportation cost of the vehicle, reduce the transportation route and number of vehicles, and thus reduce the logistics transportation cost. Manual dispatching results in a smaller transport volume with the same number of vehicles due to unreasonable vehicle assembly, such as inability to fully load and overly complex vehicle planning routes, which in turn leads to a high storage rate in warehouses. For this reason, logistics companies have to increase the capacity of warehouses, which in turn increases the operating costs of logistics companies. Unreasonable assembly increases the number of loading and unloading times for vehicle assembly, which in turn increases loading and unloading costs.

[0003] Therefore, it is necessary to provide a warehouse logistics scheduling method and system based on cloud computing to improve the intelligence level of warehouse logistics scheduling. Summary of the invention

[0004] The present invention provides a warehousing logistics scheduling method based on cloud computing, comprising: jointly establishing a logistics scheduling model through multiple edge cloud computing nodes, wherein the logistics scheduling model includes an edge scheduling solution generation model and a comprehensive scheduling solution generation model; deploying the logistics scheduling model at each edge cloud computing node; obtaining warehousing logistics scheduling tasks; calculating the computing power resource requirements of the warehousing logistics scheduling tasks; based on the computing power resource requirements of the warehousing logistics scheduling tasks, determining at least one first optimal edge cloud computing node from the multiple edge cloud computing nodes; obtaining warehousing node status information and logistics vehicle status information; each of the first optimal edge cloud computing nodes generates the edge scheduling solution model based on the computing power resource requirements of the warehousing logistics scheduling tasks; The storage logistics scheduling task, storage node status information and logistics vehicle status information are used to generate multiple candidate storage logistics scheduling schemes; based on the device status information of each of the first optimal edge cloud computing nodes, a second optimal edge cloud computing node is determined from the at least one first optimal edge cloud computing node; the second optimal edge cloud computing node screens the multiple candidate storage logistics scheduling schemes to determine multiple target storage logistics scheduling schemes; the second optimal edge cloud computing node generates an optimal storage logistics scheduling scheme based on the multiple target storage logistics scheduling schemes through the comprehensive scheduling scheme generation model; based on the optimal storage logistics scheduling scheme, storage logistics scheduling is performed to complete the storage logistics scheduling task.

[0005] Furthermore, a logistics scheduling model is jointly established through multiple edge cloud computing nodes, including: establishing an initial logistics scheduling model; deploying the initial logistics scheduling model on each of the edge cloud computing nodes; and jointly training the initial logistics scheduling model through the multiple edge cloud computing nodes to obtain the logistics scheduling model.

[0006] Furthermore, the initial logistics scheduling model is jointly trained by the multiple edge cloud computing nodes to obtain the logistics scheduling model, including: S11, each of the edge cloud computing nodes trains the initial logistics scheduling model to obtain the trained initial logistics scheduling model, tests the trained initial logistics scheduling model to obtain the model performance test data of the trained initial logistics scheduling model; S12, based on the model performance test data of the trained initial logistics scheduling model, judge whether the training is completed, if so, generate the logistics scheduling model based on the model parameters of the trained initial logistics scheduling model on each of the edge cloud computing nodes, if not, execute S13; S13, based on the model parameters of the trained initial logistics scheduling model on each of the edge cloud computing nodes, generate the intermediate logistics scheduling model degree model; S14, each of the edge cloud computing nodes trains the intermediate logistics scheduling model, obtains the trained intermediate logistics scheduling model, tests the trained intermediate logistics scheduling model, and obtains the model performance test data of the trained intermediate logistics scheduling model; S15, based on the model performance test data of the trained intermediate logistics scheduling model, determines whether the training is completed, if so, generates the logistics scheduling model based on the model parameters of the trained intermediate logistics scheduling model on each of the edge cloud computing nodes, if not, executes S16; S16, based on the model parameters of the trained intermediate logistics scheduling model on each of the edge cloud computing nodes, updates the parameters of the intermediate logistics scheduling model, deploys the updated parameters of the intermediate logistics scheduling model on each of the edge cloud computing nodes, and executes S14.

[0007] Furthermore, calculating the computing power resource requirements of the warehousing and logistics scheduling task includes: establishing a sample demand database, wherein the sample demand database is used to store the computing power resource requirements corresponding to multiple sample warehousing and logistics scheduling tasks; calculating the similarity between the warehousing and logistics scheduling task and the sample warehousing and logistics scheduling task, and determining similar sample warehousing and logistics scheduling tasks; and calculating the computing power resource requirements of the warehousing and logistics scheduling task based on the computing power resource requirements of the similar sample warehousing and logistics scheduling tasks.

[0008] Furthermore, the second optimal edge cloud computing node screens the multiple candidate warehousing and logistics scheduling plans to determine multiple target warehousing and logistics scheduling plans, including: calculating the plan similarity between any two of the candidate warehousing and logistics scheduling plans; clustering the multiple target warehousing and logistics scheduling plans based on the plan similarity between any two of the candidate warehousing and logistics scheduling plans to obtain multiple plan clustering clusters; the second optimal edge cloud computing node screens the multiple plan clustering clusters through multiple plan screening indicators to determine the optimal plan clustering cluster; the second optimal edge cloud computing node screens the multiple candidate warehousing and logistics scheduling plans included in the optimal plan clustering cluster through multiple plan screening indicators to determine multiple target warehousing and logistics scheduling plans.

[0009] Furthermore, the second optimal edge cloud computing node screens multiple candidate warehousing and logistics scheduling schemes included in the optimal solution cluster cluster through multiple solution screening indicators to determine multiple target warehousing and logistics scheduling schemes, including: predicting the warehousing node status information and logistics vehicle status information at multiple future time points based on the warehousing node status information and logistics vehicle status information at multiple historical time points through a state prediction model; obtaining weather forecast information and traffic flow information of the target area at multiple historical points; predicting the traffic flow information of the target area at multiple future time points through a road condition prediction model based on weather forecast information and traffic flow information of the target area at multiple historical points; for each candidate warehousing and logistics scheduling scheme included in the optimal solution cluster cluster, based on the predicted warehousing node status information, logistics vehicle status information and traffic flow information of the target area at multiple future time points, determining the score of the candidate warehousing and logistics scheduling scheme in multiple solution screening indicators; based on the score of each candidate warehousing and logistics scheduling scheme included in the optimal solution cluster cluster in multiple solution screening indicators, screening the multiple candidate warehousing and logistics scheduling schemes included in the optimal solution cluster cluster to determine multiple target warehousing and logistics scheduling schemes.

[0010] Furthermore, the fusion weight of each target warehousing logistics scheduling plan is calculated through the comprehensive scheduling plan generation model; the optimal warehousing logistics scheduling plan is generated based on the multiple target warehousing logistics scheduling plans and the fusion weight of each target warehousing logistics scheduling plan through the comprehensive scheduling plan generation model.

[0011] The present invention provides a cloud computing-based warehousing and logistics scheduling system, which is used to execute the above-mentioned cloud computing-based warehousing and logistics scheduling method, including: a model building module, which is used to jointly establish a logistics scheduling model through multiple edge cloud computing nodes, wherein the logistics scheduling model includes an edge scheduling solution generation model and a comprehensive scheduling solution generation model; a model deployment module, which is used to deploy the logistics scheduling model on each edge cloud computing node; a task acquisition module, which is used to acquire warehousing and logistics scheduling tasks; a demand determination module, which is used to calculate the computing power resource requirements based on the warehousing and logistics scheduling tasks; a primary screening module, which is used to determine at least one first optimal edge cloud computing node from the multiple edge cloud computing nodes based on the computing power resource requirements; an information acquisition module, which is used to acquire warehousing node status information and logistics vehicle status information; a solution generation module, which is used to obtain the first optimal edge cloud computing node based on each of the first optimal edge cloud computing nodes. The computing node generates a plurality of candidate warehousing and logistics scheduling schemes based on the warehousing and logistics scheduling tasks, warehousing node status information and logistics vehicle status information through the edge scheduling scheme generation model; a secondary screening module is used to determine the second optimal edge cloud computing node from the at least one first optimal edge cloud computing node based on the device status information of each of the first optimal edge cloud computing nodes; a scheme screening module is used to screen the plurality of candidate warehousing and logistics scheduling schemes based on the second optimal edge cloud computing node, and determine a plurality of target warehousing and logistics scheduling schemes; a scheme optimization module is used to generate an optimal warehousing and logistics scheduling scheme based on the plurality of target warehousing and logistics scheduling schemes through the comprehensive scheduling scheme generation model based on the second optimal edge cloud computing node; a logistics scheduling module is used to perform warehousing and logistics scheduling based on the optimal warehousing and logistics scheduling scheme to complete the warehousing and logistics scheduling task.

[0012] Compared with the prior art, the cloud computing-based warehouse logistics scheduling method and system provided in this specification has at least the following beneficial effects:

[0013] 1. Through the collaborative work of multiple edge cloud computing nodes, distributed processing of warehousing logistics scheduling tasks can be achieved, and the utilization rate of computing resources can be improved. According to the equipment status information and computing resource requirements of each node, tasks can be reasonably allocated to avoid overloading of a single node, achieve load balancing, and improve overall efficiency. Deploying the same logistics scheduling model on multiple nodes can increase the fault tolerance of the system. Even if a node fails, other nodes can continue to work to ensure the normal progress of scheduling tasks. Each first optimal edge cloud computing node will generate multiple candidate scheduling schemes. Through comprehensive comparison and screening, the optimal scheduling scheme can be selected to optimize the scheduling cost;

[0014] 2. Using the computing power of multiple edge cloud computing nodes for distributed training can accelerate the training process of the logistics scheduling model and improve training efficiency. Each node may process warehousing and logistics data from different regions. This data diversity helps the logistics scheduling model learn more comprehensive features, thereby improving the generalization ability and accuracy of the logistics scheduling model. Through multiple iterative training, the model parameters can be gradually optimized, the model error can be reduced, and the model performance can be improved. The trained logistics scheduling model is deployed on each node to form model redundancy. When the logistics scheduling model of an edge cloud computing node fails, it can be quickly switched to the logistics scheduling model of other edge cloud computing nodes to ensure service continuity;

[0015] 3. Through the state prediction model and the road condition prediction model, the storage node status, logistics vehicle status and traffic flow information of the target area at future time points can be predicted based on data from multiple historical time points. This enables the scheduling plan to more accurately reflect the actual situation in the future and improve the accuracy and reliability of the plan. Based on the real-time predicted information, the scheduling plan can be dynamically adjusted to adapt to changes in the storage and logistics environment, such as weather changes, traffic flow changes, etc., to ensure the flexibility and adaptability of the scheduling plan. By comprehensively considering the predicted storage node status, logistics vehicle status and traffic flow information in the target area, the scheduling plan can be formulated more accurately, reducing invalid transportation and waiting time, and improving scheduling efficiency. Accurate scheduling plans help reduce transportation costs and reduce the waste of human and material resources, thereby achieving cost control and maximizing benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0017] Figure 1 It is a flow chart of a warehousing logistics scheduling method based on cloud computing shown in an embodiment of the present application;

[0018] Figure 2 is a flowchart of a joint training initial logistics scheduling model shown in an embodiment of the present application;

[0019] Figure 3 It is a module diagram of a cloud computing-based warehousing logistics scheduling system shown in one embodiment of the present application. DETAILED DESCRIPTION

[0020] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly introduces the drawings required for describing the embodiments.

[0021] Figure 1is a flow chart of a warehousing logistics scheduling method based on cloud computing shown in an embodiment of the present application, such as Figure 1 As shown, a warehouse logistics scheduling method based on cloud computing may include the following steps.

[0022] Step 111, jointly establish a logistics scheduling model through multiple edge cloud computing nodes.

[0023] Among them, the logistics scheduling model includes an edge scheduling plan generation model and a comprehensive scheduling plan generation model.

[0024] Specifically include:

[0025] Through multiple edge cloud computing nodes, a logistics scheduling model is jointly established, including:

[0026] Establishing an initial logistics scheduling model, wherein the initial logistics scheduling model can be a Convolutional Neural Networks (CNN) model;

[0027] An initial logistics scheduling model is deployed on each edge cloud computing node;

[0028] The initial logistics scheduling model is jointly trained by multiple edge cloud computing nodes to obtain the logistics scheduling model.

[0029] Figure 2 is a flowchart of the joint training initial logistics scheduling model shown in an embodiment of the present application, such as Figure 2 As shown in the figure, the initial logistics scheduling model is jointly trained by multiple edge cloud computing nodes to obtain a logistics scheduling model, including:

[0030] S11. Each edge cloud computing node trains the initial logistics scheduling model, obtains the trained initial logistics scheduling model, tests the trained initial logistics scheduling model, and obtains model performance test data of the trained initial logistics scheduling model;

[0031] S12, based on the model performance test data of the trained initial logistics scheduling model, determine whether the training is completed, if so, generate a logistics scheduling model based on the model parameters of the trained initial logistics scheduling model on each edge cloud computing node, if not, execute S13;

[0032] S13, generating an intermediate logistics scheduling model based on the model parameters of the trained initial logistics scheduling model on each edge cloud computing node;

[0033] S14. Each edge cloud computing node trains the intermediate logistics scheduling model, obtains the trained intermediate logistics scheduling model, tests the trained intermediate logistics scheduling model, and obtains model performance test data of the trained intermediate logistics scheduling model;

[0034] S15, based on the model performance test data of the trained intermediate logistics scheduling model, determine whether the training is completed, if so, generate a logistics scheduling model based on the model parameters of the trained intermediate logistics scheduling model on each edge cloud computing node, if not, execute S16;

[0035] S16. Based on the model parameters of the trained intermediate logistics scheduling model on each edge cloud computing node, update the parameters of the intermediate logistics scheduling model, deploy the updated parameters of the intermediate logistics scheduling model on each edge cloud computing node, and execute S14.

[0036] Specifically, the model performance test data may include scores of the trained logistics scheduling model on multiple model performance indicators. Based on the model performance test data of the trained logistics scheduling model on multiple edge cloud computing nodes, a first model performance mean and a first model performance standard deviation may be calculated, and a training progress evaluation score may be calculated based on the model performance mean and the model performance standard deviation. When the training progress evaluation score is less than the training progress evaluation score threshold, the training is determined to be completed.

[0037] Preferably, the training progress evaluation score can be calculated based on the following formula:

[0038]

[0039] Among them, S t is the training progress evaluation score of the tth training, μ t is the mean performance value of the first model in the t-th training, μ0 is the mean performance value of the preset model, σ t is the first model performance standard deviation of the t-th training, and σ0 is the preset model performance standard deviation.

[0040] In some embodiments, the loss function used to train the logistics scheduling model is:

[0041] L total =β 11 ×L G +β 12 ×L F

[0042] Among them, L total For 11 and β 12 are all preset weights, and β 11 and β 12 are greater than 0, β11 +β 12 =1,L G Generate loss for the solution, L F is the scheme fusion loss.

[0043] The solution generation loss and solution fusion loss are calculated based on the following formula:

[0044]

[0045] Among them, N is the total number of training samples included in a training batch, β 21 and β 22 are all preset weights, and β 21 and β 22 are greater than 0, β 21 +β 22 =1, for,σ (n,1) is the standard deviation of the similarity of multiple candidate warehouse logistics scheduling solutions generated corresponding to the nth training sample, σ (n,2) is the standard deviation of the similarity of the actual multiple candidate warehouse logistics scheduling solutions corresponding to the nth training sample, P (n,1) is the coverage rate of the multiple candidate warehousing logistics scheduling schemes generated corresponding to the nth training sample to the multiple real candidate warehousing logistics scheduling schemes, that is, the intersection of the multiple candidate warehousing logistics scheduling schemes generated corresponding to the nth training sample and the multiple real candidate warehousing logistics scheduling schemes, P0 is the preset coverage rate, β 31 and β 32 are all preset weights, and β 31 and β 32 are greater than 0, β 31 +β 32 =1,w (n,1) is the weight matrix of multiple target warehousing logistics scheduling solutions generated corresponding to the nth training sample, w (n,2) is the weight matrix of the actual multiple target warehouse logistics scheduling schemes corresponding to the nth training sample, R (n,1) is the optimal warehouse logistics scheduling solution generated for the nth training sample, R (n,2) is the actual optimal warehouse logistics scheduling solution corresponding to the nth training sample.

[0046] For each edge cloud computing node, the parameter adjustment weight corresponding to the edge cloud computing node can be calculated based on the model performance test data and the model performance mean of the intermediate logistics scheduling model after each training of the edge cloud computing node, and the model parameters of the intermediate logistics scheduling model can be updated based on the parameter adjustment weights corresponding to each edge cloud computing node.

[0047] Preferably, the parameter adjustment weight corresponding to the edge cloud computing node can be calculated based on the following formula:

[0048]

[0049] Among them, w i The parameter adjustment weight corresponding to the i-th edge cloud computing node, p (i,t) is the model performance test data of the logistics scheduling model after the training of the ith edge cloud computing node in the tth training, T is the total number of completed trainings, J is the total number of edge cloud computing nodes, and p (j,t) It is the model performance test data of the logistics scheduling model of the j-th edge cloud computing node after the t-th training.

[0050] Preferably, the model parameters can be updated based on the following formula:

[0051]

[0052] Among them, η t+1 is the updated parameter of the intermediate logistics scheduling model for the t+1th training, η (i,t) are the parameters of the intermediate logistics scheduling model of the i-th edge cloud computing node trained for the t-th time.

[0053] Step 112: deploy a logistics scheduling model on each edge cloud computing node.

[0054] Step 113, obtaining warehousing logistics scheduling tasks.

[0055] Warehouse logistics scheduling tasks may include the starting point, destination, transportation time requirements, transportation volume, type of goods, size, weight, packaging method, and whether special handling is required (such as refrigeration, shockproof, etc.).

[0056] Step 114, calculate the computing resource requirements of the warehousing logistics scheduling task.

[0057] Specifically include:

[0058] Establish a sample demand database, where the sample demand database is used to store computing resource requirements corresponding to multiple sample warehousing and logistics scheduling tasks;

[0059] Calculate the similarity between the warehousing logistics scheduling task and the sample warehousing logistics scheduling task, and determine the similar sample warehousing logistics scheduling task;

[0060] Based on the computing resource requirements of similar sample warehousing and logistics scheduling tasks, calculate the computing resource requirements of warehousing and logistics scheduling tasks.

[0061] Specifically, the similarity between the warehousing logistics scheduling task and the sample warehousing logistics scheduling task can be calculated by a similarity determination model, wherein the similarity determination model can be a convolutional neural network (CNN) model. The sample warehousing logistics scheduling task with a similarity greater than a similarity threshold is regarded as a similar sample warehousing logistics scheduling task.

[0062] The computing power resource requirements of similar sample warehousing and logistics scheduling tasks are weighted and summed to calculate the computing power resource requirements of the warehousing and logistics scheduling tasks, where the weight corresponding to the computing power resources of the similar sample warehousing and logistics scheduling tasks is the similarity between the similar sample warehousing and logistics scheduling tasks and the warehousing and logistics scheduling tasks.

[0063] Step 115: Determine at least one first optimal edge cloud computing node from multiple edge cloud computing nodes based on the computing resource requirements of the warehousing logistics scheduling task.

[0064] Specifically, an edge cloud computing node that meets the computing resource requirements can be selected as the first optimal edge cloud computing node.

[0065] Step 116, obtaining storage node status information and logistics vehicle status information.

[0066] Specifically, the storage node status information may include the storage situation of the goods in the warehouse, including the type, quantity, storage location and status of the goods in the node (such as waiting to be shipped out, shipped out, waiting to be shipped in, etc.), and may also include the status information of the warehouse equipment (such as shelves, forklifts, conveyor belts, etc.) (such as whether it is idle, whether it is faulty, whether it needs maintenance, etc.), and may also include weather information, ongoing storage logistics scheduling tasks and pending storage logistics scheduling tasks.

[0067] Logistics vehicle status information may include information such as the vehicle model, license plate number, load capacity, volume capacity, warehousing logistics scheduling tasks being executed, and warehousing logistics scheduling tasks to be executed.

[0068] Step 117, each first optimal edge cloud computing node generates a plurality of candidate warehousing and logistics scheduling schemes based on the warehousing and logistics scheduling tasks, warehousing node status information and logistics vehicle status information through an edge scheduling scheme generation model.

[0069] Step 118: Determine a second optimal edge cloud computing node from at least one first optimal edge cloud computing node based on the device status information of each first optimal edge cloud computing node.

[0070] Specifically, based on the device status information of each first optimal edge cloud computing node, the first optimal edge cloud computing node with the best device performance may be used as the second optimal edge cloud computing node.

[0071] Step 119: the second optimal edge cloud computing node screens multiple candidate warehousing logistics scheduling plans to determine multiple target warehousing logistics scheduling plans.

[0072] Specifically include:

[0073] Calculate the similarity between any two candidate warehouse logistics scheduling solutions;

[0074] Based on the similarity of any two candidate warehousing logistics scheduling solutions, multiple target warehousing logistics scheduling solutions are clustered to obtain multiple solution clustering clusters. For example, based on the similarity of any two candidate warehousing logistics scheduling solutions, multiple target warehousing logistics scheduling solutions are clustered by a K-means clustering algorithm to obtain multiple solution clustering clusters.

[0075] The second optimal edge cloud computing node screens multiple solution clusters through multiple solution screening indicators to determine the optimal solution cluster. For example, for each solution cluster, the scores of the candidate warehousing logistics scheduling solutions corresponding to the cluster center of the solution cluster in multiple solution screening indicators are calculated, and the solution cluster with the highest score in multiple solution screening indicators is used as the optimal solution cluster;

[0076] The second optimal edge cloud computing node screens multiple candidate warehousing and logistics scheduling solutions included in the optimal solution clustering cluster through multiple solution screening indicators, and determines multiple target warehousing and logistics scheduling solutions.

[0077] In some embodiments, the second optimal edge cloud computing node screens multiple candidate warehousing logistics scheduling solutions included in the optimal solution cluster cluster through multiple solution screening indicators, and determines multiple target warehousing logistics scheduling solutions, including:

[0078] Predicting the storage node status information and logistics vehicle status information at multiple future time points based on the storage node status information and logistics vehicle status information at multiple historical time points through a status prediction model, wherein the status prediction model may be a long short-term memory network (LSTM, Long Short-Term Memory) model;

[0079] Obtain weather forecast information and traffic flow information for target areas at multiple historical points;

[0080] Predicting the traffic flow information of the target area at multiple future time points through a traffic condition prediction model based on weather forecast information and traffic flow information of the target area at multiple historical points, wherein the traffic condition prediction model may be a long short-term memory network (LSTM, Long Short-Term Memory) model;

[0081] For each candidate warehousing logistics scheduling scheme included in the optimal solution cluster, based on the predicted storage node status information, logistics vehicle status information and vehicle flow information of the target area at multiple future time points, the scores of the candidate warehousing logistics scheduling scheme in multiple scheme screening indicators are determined, wherein the multiple scheme screening indicators may at least include a transportation time consumption indicator, a transportation route length indicator, a transportation safety indicator, an average vehicle load factor indicator and a warehouse loading and unloading pressure indicator, etc.;

[0082] Based on the scores of each candidate warehousing logistics scheduling solution included in the optimal solution clustering cluster in multiple solution screening indicators, multiple candidate warehousing logistics scheduling solutions included in the optimal solution clustering cluster are screened to determine multiple target warehousing logistics scheduling solutions.

[0083] For example, the scores of each candidate warehousing logistics scheduling solution included in the optimal solution clustering cluster in multiple solution screening indicators are weighted and summed, and the comprehensive score of the candidate warehousing logistics scheduling solution is calculated, and the candidate warehousing logistics scheduling solution whose comprehensive score is greater than the comprehensive score threshold is used as the target warehousing logistics scheduling solution.

[0084] Step 120: The second optimal edge cloud computing node generates an optimal warehousing and logistics scheduling plan based on a plurality of target warehousing and logistics scheduling plans through a comprehensive scheduling plan generation model.

[0085] Specifically include:

[0086] The fusion weight of each target warehousing logistics scheduling plan is calculated through the comprehensive scheduling plan generation model;

[0087] The optimal warehousing logistics scheduling plan is generated through the comprehensive scheduling plan generation model based on multiple target warehousing logistics scheduling plans and the fusion weight of each target warehousing logistics scheduling plan.

[0088] Step 121, based on the optimal warehousing logistics scheduling plan, perform warehousing logistics scheduling to complete the warehousing logistics scheduling task.

[0089] Figure 3 is a module diagram of a cloud computing-based warehousing logistics scheduling system shown in an embodiment of the present application, such as Figure 3As shown, a cloud computing-based warehousing logistics scheduling system may include a model building module, a model deployment module, a task acquisition module, a demand determination module, a primary screening module, an information acquisition module, a solution generation module, a secondary screening module, a solution screening module, a solution optimization module and a logistics scheduling module.

[0090] The model building module can be used to jointly establish a logistics scheduling model through multiple edge cloud computing nodes, wherein the logistics scheduling model includes an edge scheduling solution generation model and a comprehensive scheduling solution generation model;

[0091] The model deployment module can be used to deploy the logistics scheduling model on each edge cloud computing node;

[0092] The task acquisition module can be used to obtain warehouse logistics scheduling tasks;

[0093] The demand determination module can be used to calculate computing resource requirements based on warehouse logistics scheduling tasks;

[0094] The primary screening module may be used to determine at least one first optimal edge cloud computing node from a plurality of edge cloud computing nodes based on computing power resource requirements;

[0095] The information acquisition module can be used to obtain storage node status information and logistics vehicle status information;

[0096] The solution generation module can be used to generate multiple candidate warehousing logistics scheduling solutions based on each first optimal edge cloud computing node through an edge scheduling solution generation model, based on warehousing logistics scheduling tasks, warehousing node status information and logistics vehicle status information;

[0097] The secondary screening module may be used to determine a second optimal edge cloud computing node from at least one first optimal edge cloud computing node based on device status information of each first optimal edge cloud computing node;

[0098] The solution screening module can be used to screen multiple candidate warehousing logistics scheduling solutions based on the second optimal edge cloud computing node to determine multiple target warehousing logistics scheduling solutions;

[0099] The solution optimization module can be used to generate an optimal warehousing and logistics scheduling solution based on a second optimal edge cloud computing node through a comprehensive scheduling solution generation model and multiple target warehousing and logistics scheduling solutions;

[0100] The logistics scheduling module can be used to perform warehousing logistics scheduling based on the optimal warehousing logistics scheduling plan and complete the warehousing logistics scheduling tasks.

[0101] A cloud computing-based warehousing logistics scheduling system can be used to execute a cloud computing-based warehousing logistics scheduling method. For more descriptions of a cloud computing-based warehousing logistics scheduling system, please refer to the relevant descriptions of a cloud computing-based warehousing logistics scheduling method, which will not be repeated here.

[0102] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A warehouse logistics scheduling method and system based on cloud computing, characterized in that: include: A logistics scheduling model is jointly established through multiple edge cloud computing nodes, wherein the logistics scheduling model includes an edge scheduling solution generation model and a comprehensive scheduling solution generation model; The logistics scheduling model is deployed on each edge cloud computing node; Obtain warehousing logistics scheduling tasks; Calculate the computing resource requirements for the warehousing logistics scheduling task; Based on the computing resource requirements of the warehousing logistics scheduling task, determining at least one first optimal edge cloud computing node from the multiple edge cloud computing nodes; Obtain storage node status information and logistics vehicle status information; Each of the first optimal edge cloud computing nodes generates a plurality of candidate warehousing and logistics scheduling schemes based on the warehousing and logistics scheduling tasks, warehousing node status information, and logistics vehicle status information through the edge scheduling scheme generation model; Based on the device status information of each of the first optimal edge cloud computing nodes, determining a second optimal edge cloud computing node from the at least one first optimal edge cloud computing node; The second optimal edge cloud computing node screens the multiple candidate warehousing logistics scheduling solutions to determine multiple target warehousing logistics scheduling solutions; The second optimal edge cloud computing node generates an optimal warehousing and logistics scheduling plan based on the multiple target warehousing and logistics scheduling plans through the comprehensive scheduling plan generation model; Based on the optimal warehousing logistics scheduling plan, warehousing logistics scheduling is carried out to complete the warehousing logistics scheduling task.

2. A cloud computing-based warehousing logistics scheduling method and system according to claim 1, characterized in that: Through multiple edge cloud computing nodes, a logistics scheduling model is jointly established, including: Establish an initial logistics scheduling model; The initial logistics scheduling model is deployed on each edge cloud computing node; The initial logistics scheduling model is jointly trained by the multiple edge cloud computing nodes to obtain the logistics scheduling model.

3. A cloud computing-based warehousing logistics scheduling method and system according to claim 2, characterized in that: The initial logistics scheduling model is jointly trained by the multiple edge cloud computing nodes to obtain the logistics scheduling model, including: S11. Each of the edge cloud computing nodes trains the initial logistics scheduling model to obtain the trained initial logistics scheduling model, tests the trained initial logistics scheduling model, and obtains model performance test data of the trained initial logistics scheduling model; S12. Based on the model performance test data of the trained initial logistics scheduling model, determine whether the training is completed. If so, generate the logistics scheduling model based on the model parameters of the trained initial logistics scheduling model on each edge cloud computing node. If not, execute S13.

4. A cloud computing-based warehousing logistics scheduling method and system according to claim 3, characterized in that: The logistics scheduling model is obtained by jointly training the initial logistics scheduling model through the multiple edge cloud computing nodes, and further includes: S13, generating an intermediate logistics scheduling model based on the model parameters of the trained initial logistics scheduling model on each edge cloud computing node; S14. Each of the edge cloud computing nodes trains the intermediate logistics scheduling model to obtain the trained intermediate logistics scheduling model, tests the trained intermediate logistics scheduling model, and obtains model performance test data of the trained intermediate logistics scheduling model.

5. A cloud computing-based warehousing logistics scheduling method and system according to claim 4, characterized in that: The initial logistics scheduling model is jointly trained by the multiple edge cloud computing nodes to obtain the logistics scheduling model, further comprising: S15, judging whether the training is completed based on the model performance test data of the trained intermediate logistics scheduling model, if so, generating the logistics scheduling model based on the model parameters of the trained intermediate logistics scheduling model on each edge cloud computing node, if not, executing S16; S16. Based on the model parameters of the trained intermediate logistics scheduling model on each of the edge cloud computing nodes, update the parameters of the intermediate logistics scheduling model, deploy the updated parameters of the intermediate logistics scheduling model on each of the edge cloud computing nodes, and execute S14.

6. A cloud computing-based warehousing logistics scheduling method and system according to claim 5, characterized in that: Calculate the computing resource requirements for the warehousing logistics scheduling task, including: Establishing a sample demand database, wherein the sample demand database is used to store computing resource requirements corresponding to multiple sample warehousing logistics scheduling tasks; Calculating the similarity between the warehousing logistics scheduling task and the sample warehousing logistics scheduling task, and determining a similar sample warehousing logistics scheduling task; Based on the computing power resource requirements of the similar sample warehousing and logistics scheduling tasks, the computing power resource requirements of the warehousing and logistics scheduling tasks are calculated.

7. According to the cloud computing-based warehousing logistics scheduling method and system described in claim 5, it is characterized in that: The second optimal edge cloud computing node screens the multiple candidate warehousing logistics scheduling solutions to determine multiple target warehousing logistics scheduling solutions, including: Calculate the similarity between any two candidate warehousing logistics scheduling solutions; Based on the similarity between any two candidate warehousing logistics scheduling solutions, clustering the multiple target warehousing logistics scheduling solutions to obtain multiple solution clusters; The second optimal edge cloud computing node screens the multiple solution clusters through multiple solution screening indicators to determine the optimal solution cluster; The second optimal edge cloud computing node screens multiple candidate warehousing and logistics scheduling solutions included in the optimal solution clustering cluster through multiple solution screening indicators, and determines multiple target warehousing and logistics scheduling solutions.

8. A cloud computing-based warehousing logistics scheduling method and system according to claim 7, characterized in that: The second optimal edge cloud computing node screens multiple candidate warehousing logistics scheduling solutions included in the optimal solution clustering cluster through multiple solution screening indicators, and determines multiple target warehousing logistics scheduling solutions, including: The state prediction model is used to predict the state information of storage nodes and logistics vehicles at multiple future time points based on the state information of storage nodes and logistics vehicles at multiple historical time points. Obtain weather forecast information and traffic flow information for target areas at multiple historical points; Predicting the traffic flow information of the target area at multiple future time points by using a road condition prediction model based on weather forecast information and traffic flow information of the target area at multiple historical points; For each candidate warehousing logistics scheduling solution included in the optimal solution cluster, based on the predicted storage node status information, logistics vehicle status information and vehicle flow information of the target area at multiple future time points, determine the score of the candidate warehousing logistics scheduling solution in multiple solution screening indicators; Based on the scores of each candidate warehousing logistics scheduling solution included in the optimal solution clustering cluster in multiple solution screening indicators, multiple candidate warehousing logistics scheduling solutions included in the optimal solution clustering cluster are screened to determine multiple target warehousing logistics scheduling solutions.

9. A cloud computing-based warehousing logistics scheduling method and system according to any one of claims 1 to 5, characterized in that: The second optimal edge cloud computing node generates an optimal warehousing and logistics scheduling plan based on the multiple target warehousing and logistics scheduling plans through the comprehensive scheduling plan generation model, including: Calculate the fusion weight of each target warehousing logistics scheduling plan through the comprehensive scheduling plan generation model; The comprehensive scheduling scheme generation model generates an optimal warehousing logistics scheduling scheme based on the multiple target warehousing logistics scheduling schemes and the fusion weight of each target warehousing logistics scheduling scheme.

10. A cloud computing-based warehouse logistics scheduling system, characterized in that: A cloud computing-based warehousing logistics scheduling method for executing any one of claims 1 to 9, comprising: A model building module, used to jointly establish a logistics scheduling model through multiple edge cloud computing nodes, wherein the logistics scheduling model includes an edge scheduling solution generation model and a comprehensive scheduling solution generation model; A model deployment module, used to deploy the logistics scheduling model on each edge cloud computing node; Task acquisition module, used to obtain warehousing logistics scheduling tasks; A demand determination module, used to calculate the computing power resource demand based on the warehousing logistics scheduling task; A primary screening module, configured to determine at least one first optimal edge cloud computing node from the plurality of edge cloud computing nodes based on the computing resource requirement; Information acquisition module, used to obtain storage node status information and logistics vehicle status information; A solution generation module, configured to generate a plurality of candidate warehousing and logistics scheduling solutions based on each of the first optimal edge cloud computing nodes through the edge scheduling solution generation model, based on the warehousing and logistics scheduling tasks, warehousing node status information, and logistics vehicle status information; a secondary screening module, configured to determine a second optimal edge cloud computing node from the at least one first optimal edge cloud computing node based on device status information of each of the first optimal edge cloud computing nodes; A solution screening module, used to screen the multiple candidate warehousing logistics scheduling solutions based on the second optimal edge cloud computing node, and determine multiple target warehousing logistics scheduling solutions; A solution optimization module, configured to generate an optimal warehousing and logistics scheduling solution based on the second optimal edge cloud computing node through the comprehensive scheduling solution generation model and based on the multiple target warehousing and logistics scheduling solutions; The logistics scheduling module is used to perform warehousing logistics scheduling based on the optimal warehousing logistics scheduling plan and complete the warehousing logistics scheduling task.

Images