Logistics simulation method and device, electronic equipment and computer readable storage medium

Abstract

The invention provides a logistics simulation method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of logistics simulation. The logistics simulation method is applied to the logistics simulation system and comprises the steps that a line prediction result is obtained at a current operation node, and the line prediction result is obtained by processing offline statistical characteristics and real-time waybill characteristics of historical waybills through a line prediction model in advance; and if the waybill flow direction corresponding to the to-be-processed waybill is determined to meet the preset condition, determining a target route from to-be-selected routes of the route prediction result so as to determine the target route as a simulation prediction route of the to-be-processed waybill. The accuracy of route planning in the logistics simulation process can be improved, and the accuracy of the logistics simulation result is improved.

Description

Technical Field

[0001] This disclosure relates to the field of logistics simulation technology, specifically to a logistics simulation method, a logistics simulation device, an electronic device, and a computer-readable storage medium. Background Art

[0002] In modern logistics, the flow of parcels involves various stages such as warehousing, sorting, transportation, and distribution, making the logistics system extremely complex. Optimizing warehouse network layout and resource allocation can significantly improve logistics service efficiency. However, the implementation of any decision can have a major impact on the complex logistics system, even causing system paralysis, resulting in extremely high trial-and-error costs. Simulation can simulate the real physical world, mimicking the operating state of actual systems and their changes over time, and can be used to estimate and infer the performance of actual systems, thereby assisting in decision-making. Due to its advantages of low cost and high computational accuracy, intelligent decision-making is widely used in modern logistics, solving to some extent the problems of complex logistics systems and the difficulty of implementing decisions.

[0003] However, when a logistics simulation system is running, package routes and other data need to be generated based on some given rules. However, in real-world operation scenarios, due to the different behavioral preferences of operators, there are certain deviations between the simulation rules and the actual situation. This leads to a large discrepancy between the simulation results and reality, which to some extent affects the simulation accuracy of the logistics simulation system.

[0004] It should be noted that the information in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this disclosure is to provide a logistics simulation method, a logistics simulation device, an electronic device, and a computer-readable storage medium, thereby improving the accuracy of route planning during logistics simulation to at least a certain extent and enhancing the accuracy of logistics simulation.

[0006] According to a first aspect of this disclosure, a logistics simulation method is provided, applied to a logistics simulation system, comprising: acquiring route prediction results at the current operation node, wherein the route prediction results are obtained in advance by processing offline statistical characteristics and real-time route characteristics of historical waybills through a route prediction model; if it is determined that the waybill flow direction corresponding to the waybill to be processed meets preset conditions, then determining a target route from the candidate routes in the route prediction results, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0007] In an exemplary embodiment, the method further includes: if the flow direction of the waybill to be processed does not meet the preset conditions, then performing simulated route planning on the waybill to be processed through the logistics simulation system to obtain the simulated predicted route of the waybill to be processed.

[0008] In an exemplary embodiment, the method further includes: obtaining historical order flow directions and misclassification values ​​of the historical order flow directions; wherein the misclassification value is used to reflect the probability of a route misjudgment in the historical order flow direction; and constructing a sampling flow direction pool based on historical order flow directions with misclassification values ​​greater than a first preset threshold.

[0009] In an exemplary embodiment, determining whether the waybill flow direction corresponding to the waybill to be processed meets the preset condition includes: if the waybill flow direction corresponding to the waybill to be processed is in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed meets the preset condition; or, if the waybill flow direction corresponding to the waybill to be processed is not in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed does not meet the preset condition.

[0010] In an exemplary embodiment, before obtaining the route prediction result at the current operation node, the method further includes: obtaining historical waybill samples covered by historical waybill flow directions in the sampling flow direction pool; obtaining candidate route samples corresponding to the historical waybill samples based on all possible routes between each operation stage of the historical waybill samples; constructing training samples based on the real-time waybill characteristics, offline statistical characteristics, and candidate route samples of the historical waybill samples, so as to train the model to be trained using the training samples to obtain the route prediction model; wherein, the sample label of the training samples is determined according to the actual selected route of the historical waybill samples.

[0011] In an exemplary embodiment, obtaining historical waybill samples covered by historical waybill flows in the sampling flow pool includes: sampling the waybills covered by each historical waybill flow for each historical waybill flow, so as to obtain historical waybill samples of the historical waybill flow based on the sampling results.

[0012] In an exemplary embodiment, obtaining the route prediction result at the current operating node includes: loading the route prediction result into the memory of the logistics simulation system, wherein the route prediction result includes the correspondence between the real-time waybill characteristics and offline statistical characteristics of the historical waybills and the route to be selected and the predicted probability of the route to be selected.

[0013] In an exemplary embodiment, the step of determining a target route from the candidate routes in the route prediction results if the flow direction corresponding to the waybill to be processed meets the preset conditions, so as to determine the target route as the simulation prediction route of the waybill to be processed, includes: determining candidate routes from the candidate routes according to the real-time waybill characteristics of the waybill to be processed and the correspondence; and determining the target route from the candidate routes according to the prediction probability corresponding to the candidate routes, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0014] In an exemplary embodiment, determining the target route from the candidate routes based on the predicted probabilities corresponding to the candidate routes, so as to determine the target route as the simulation predicted route of the waybill to be processed, includes: determining the candidate routes with predicted probabilities greater than a second preset threshold as the target routes, so as to determine the target routes as the simulation predicted routes of the waybill to be processed; or, if the predicted probabilities of all the candidate routes are less than the second preset threshold, then performing simulation route planning on the waybill to be processed through the logistics simulation system to obtain the simulation predicted route of the waybill to be processed.

[0015] According to a second aspect of this disclosure, a logistics simulation device is provided, applied to a logistics simulation system, comprising: a result acquisition module, configured to acquire route prediction results at the current operation node, wherein the route prediction results are obtained by pre-processing offline statistical features and real-time route features of historical waybills through a route prediction model; and a route determination module, configured to determine a target route from the candidate routes in the route prediction results if the direction of the waybill corresponding to the waybill to be processed meets preset conditions, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0016] According to a third aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to implement the above-described method by executing the executable instructions.

[0017] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the method described above.

[0018] The logistics simulation method provided in this disclosure is applied to a logistics simulation system. At the current operation node, route prediction results are obtained. These results are pre-processed using a route prediction model to analyze the offline statistical characteristics and real-time characteristics of historical waybills. If the flow direction of the waybill to be processed meets preset conditions, a target route is determined from the candidate routes in the route prediction results, thus establishing the target route as the simulation prediction route for the waybill to be processed. On one hand, this process in the logistics simulation system uses a route prediction model to simulate real-world route decisions. When the flow direction of the waybill to be processed meets preset conditions, a target route is determined from the route prediction results to replace the route generation rules of the logistics simulation system, guiding and improving the accuracy of route planning in the logistics simulation process, thereby enhancing the accuracy of the logistics simulation. On the other hand, the logistics simulation method of this disclosure improves the accuracy of route prediction without significantly impacting the main architecture and existing processes of the logistics simulation system, exhibiting high adaptability and broad application prospects.

[0019] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0021] Figure 1 The diagram illustrates an application scenario to which embodiments of the present disclosure can be applied.

[0022] Figure 2 The flowchart illustrating an exemplary embodiment of the present disclosure is shown.

[0023] Figure 3 This diagram illustrates an operational step in an exemplary embodiment of the present disclosure.

[0024] Figure 4 The flowchart illustrates an implementation of a sampling flow pool in an exemplary embodiment of this disclosure.

[0025] Figure 5 The flowchart illustrates an implementation of a training model training method in an exemplary embodiment of the present disclosure.

[0026] Figure 6 The illustration shows a schematic diagram of determining candidate line samples in an exemplary embodiment of the present disclosure.

[0027] Figure 7 The schematic diagram illustrates a network structure for an FM portion of an exemplary embodiment of this disclosure.

[0028] Figure 8 The schematic illustration shows a network structure for a deep portion of an exemplary embodiment of this disclosure.

[0029] Figure 9 The flowchart illustrating an exemplary embodiment of the present disclosure is shown in which a simulation-predicted route is determined based on route prediction results.

[0030] Figure 10 This illustration schematically depicts a route decision-making process in a logistics simulation system according to an exemplary embodiment of the present disclosure.

[0031] Figure 11 A schematic diagram of the composition of a logistics simulation apparatus that can be applied to exemplary embodiments of the present disclosure is shown.

[0032] Figure 12 A schematic diagram of the composition of an electronic device to which the exemplary embodiments of the present disclosure may be applied is shown. Detailed Implementation

[0033] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0034] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0035] The logistics simulation method provided in this disclosure can be applied to, for example... Figure 1 The application environment shown is illustrated. Terminal 101 communicates with server 102 via a network. A data storage system can store the data that server 102 needs to process. The data storage system can be integrated onto server 102, or it can be located in the cloud or on another network server.

[0036] In one exemplary embodiment, the logistics simulation method provided in this disclosure can be executed by server 102, and correspondingly, a logistics simulation device is disposed in server 102. Correspondingly, in this method of execution by server 102, server 102 can begin executing the steps in the technical solution of the exemplary embodiment of this disclosure in response to a triggering command, wherein the triggering command can be sent by a terminal used by a user, or can be triggered locally by the server in response to some automated event.

[0037] Server 102 can obtain the route prediction result at the current operation node. The route prediction result is obtained by processing the offline statistical characteristics and real-time characteristics of historical waybills in advance through the route prediction model. If it is determined that the waybill flow direction corresponding to the waybill to be processed meets the preset conditions, the target route is determined from the routes to be selected in the route prediction result, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0038] Server 102 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. Server 102 can execute background tasks.

[0039] Furthermore, in another exemplary embodiment, terminal 101 may also have similar functions to server 102, thereby executing the logistics simulation method provided by the exemplary embodiments of this disclosure.

[0040] The terminal 101 can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, IoT device, or portable wearable device. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. The terminal 101 can also be referred to as a mobile terminal, terminal device, mobile device, etc. The exemplary embodiments of this disclosure do not limit the type of terminal 101.

[0041] Furthermore, the technical solutions of the exemplary embodiments of this disclosure can also be executed collaboratively by terminal 101 and server 102. In this method of collaborative execution by terminal 101 and server 102, some steps in the technical solutions provided by the exemplary embodiments of this disclosure are executed by terminal 101, while other steps are executed by server 102. It should be noted that in this method of collaborative execution by terminal 101 and server 102, the steps executed by terminal 101 and server 102 respectively can be dynamically adjusted according to the actual situation, and no special restrictions are imposed on this.

[0042] The terminal 101 and the server 102 can be connected directly or indirectly via wireless communication, and the exemplary embodiments of this disclosure are not particularly limited herein.

[0043] Logistics simulation involves modeling a logistics system, developing corresponding applications on a computer, simulating the actual operation of the logistics system, and statistically analyzing the simulation results to guide the planning, design, and operation management of the actual logistics system.

[0044] In developing this disclosure, the applicant discovered that the accuracy of logistics simulation can be improved by combining machine learning models with simulation systems. In this process, the applicant found that the output of the machine learning model can be used as input to the simulation system, and the simulation system can be used to evaluate the model's output. Alternatively, in scenarios where sample acquisition is costly and difficult, a large amount of training data can be generated through the simulation system to train the machine learning model. However, the applicant also found that in the logistics field, simulation systems outperform traditional machine learning models in tasks such as hourly cargo volume prediction. How to leverage the advantages of machine learning models, based on the existing performance of simulations, to guide the output of the simulation system to approximate actual values ​​is a direction that urgently needs improvement.

[0045] Furthermore, route decisions in actual logistics operations involve interactions across multiple dimensions, such as route type, vehicle type, logistics type, and transportation time. Therefore, to address the issue of inconsistencies between the simulated logistics system process and the actual operational state, leading to unreliable simulation results, this disclosure provides a logistics simulation method applied to a logistics simulation system. This method improves the accuracy of the logistics simulation system by enhancing the accuracy of route planning during the simulation process.

[0046] Figure 2 A flowchart of a logistics simulation method according to an exemplary embodiment of the present disclosure is shown as follows: Figure 2 The logistics simulation method of this disclosure may include steps S210 and S220:

[0047] In step S210, the route prediction result is obtained at the current operation node. This route prediction result is obtained by processing the offline statistical characteristics and real-time route characteristics of historical waybills in advance through the route prediction model.

[0048] In the exemplary embodiments of this disclosure, the current operation node refers to the operation node where route planning is currently required. Typically, a waybill goes through multiple operation stages from the operation point to the destination point, and various possible routes exist between these stages. For example... Figure 3 A schematic diagram of an operational procedure is shown, such as... Figure 3If the current operation node is link 1, and there are links 2, 3 and 4 between link 1 and the destination network point, then the next step is to select which link to follow, which is to perform route prediction at each operation node.

[0049] It should be noted that, Figure 3 This is merely an illustration of the operational process. In actual operational scenarios, there may be more or more granular steps between the current operational node and the destination network point. These steps can be flexibly adjusted according to the actual situation, and there are no restrictions on them.

[0050] The route prediction model is pre-trained using training samples. It is trained based on real-time waybill characteristics, offline statistical characteristics, and candidate route samples from historical waybill samples. This embodiment does not impose specific limitations on the model results of the route prediction model; they can be selected according to actual needs. The route prediction results include the correspondence between real-time waybill characteristics, offline statistical characteristics, and candidate routes, as well as their predicted probabilities. The logistics simulation system can pre-load the route prediction results for each operation node to enable rapid route prediction during logistics simulation.

[0051] Here, "historical waybill" refers to the historical waybill used for route prediction of the waybill to be processed at the current operation node. It can be the latest historical waybill data. For example, it can be the historical waybill data of the latest day. When performing logistics simulation on the same day, the historical waybill can be the waybill data of the previous day. Alternatively, the historical waybill can be the latest day's waybill data obtained by the big data platform of the logistics simulation system through incremental data processing. The method of determining the historical waybill can be determined according to the actual logistics simulation needs, including but not limited to the above methods.

[0052] In the route model training and actual prediction of this disclosure, the features used include waybill features, route features and site features, which can be divided into real-time waybill features and offline statistical features.

[0053] Real-time waybill characteristics refer to the relevant features of the waybill status obtained in real time, including but not limited to route type, region type, product type, waybill type (such as self-operated, warehousing and distribution, pure delivery, etc.), cargo type (such as fresh produce, general cargo, etc.), and number of parcels. Offline statistical characteristics refer to the features obtained by offline statistics on waybills, including but not limited to site characteristics and route characteristics. Site characteristics include the arrival volume of the operating site at the same time of the previous day, two days, and three days, and the actual number of routes at the operating site on the previous day, two days, and three days. Route characteristics include the cargo volume of the previous day, two days, and three days in the current time period, one hour before and after, and two hours before and after, the average parcel volume of the previous day, two days, and three days in the current time period, one hour before and after, and two hours before and after, and the average parcel weight of the previous day, two days, and three days in the current time period, one hour before and after, and two hours before and after, etc. These can be extracted, statistically analyzed, and stored.

[0054] In step S220, if it is determined that the flow direction of the waybill to be processed meets the preset conditions, then the target route is determined from the routes to be selected in the route prediction results, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0055] In the exemplary embodiments of this disclosure, the waybill to be processed is the waybill at the current operation node, that is, the waybill whose flow direction needs to be determined at this time. The waybill flow direction refers to the entire process of an order from the start of processing to the completion of processing. The order to be processed can be divided into the corresponding waybill flow direction according to preset rules, such as dividing the order to be processed into different waybill flow directions according to the order's delivery address, etc. The embodiments of this disclosure do not limit the method of dividing the waybill flow direction.

[0056] To improve the accuracy of logistics simulation, the simulation prediction route can be determined by using the route prediction results provided by the route prediction model. Preset conditions can be set, and the simulation prediction route can be determined only when the flow direction of the waybill to be processed meets the preset conditions. Then, the waybill to be processed in the logistics simulation system will continue the subsequent simulation process based on the simulation prediction route.

[0057] The logistics simulation method provided in this disclosure is applied to a logistics simulation system. At the current operation node, route prediction results are obtained. These results are pre-processed using a route prediction model to analyze the offline statistical characteristics and real-time characteristics of historical waybills. If the flow direction of the waybill to be processed meets preset conditions, a target route is determined from the candidate routes in the route prediction results, thus establishing the target route as the simulation prediction route for the waybill to be processed. On one hand, this process in the logistics simulation system uses a route prediction model to simulate real-world route decisions. When the flow direction of the waybill to be processed meets preset conditions, a target route is determined from the route prediction results to replace the route generation rules of the logistics simulation system, guiding and improving the accuracy of route planning in the logistics simulation process, thereby enhancing the accuracy of the logistics simulation. On the other hand, the logistics simulation method of this disclosure improves the accuracy of route prediction without significantly impacting the main architecture and existing processes of the logistics simulation system, exhibiting high adaptability and broad application prospects.

[0058] The following is a detailed explanation of the content involved in each of the above steps.

[0059] In one exemplary embodiment, the method of this disclosure further includes:

[0060] If the flow direction of the waybill to be processed does not meet the preset conditions, the logistics simulation system will perform simulated route planning for the waybill to be processed to obtain the simulated predicted route of the waybill to be processed.

[0061] Specifically, this embodiment introduces route prediction results from a route prediction model into the logistics simulation system. The aim is to leverage the advantages of the route prediction model to improve the accuracy of route prediction, thereby guiding the simulation output of the logistics simulation system to more closely resemble the actual situation. However, this embodiment does not use route prediction results to replace the simulated route prediction for all pending waybills; rather, this replacement process is only performed when necessary. Therefore, by setting preset conditions, the pending waybills are diverted according to their corresponding flow direction. Only when the flow direction of the pending waybill meets the preset conditions is the prediction result of the route prediction model used to replace the route decision of the logistics simulation system; otherwise, the existing rules of the logistics simulation system are used to give the route decision.

[0062] It should be noted that the existing rules of the logistics simulation system for determining routes can be found in relevant existing technologies, and will not be elaborated upon here.

[0063] In one exemplary embodiment, an implementation for constructing a sampling flow pool is provided. For example... Figure 4 As shown, constructing a sampling flow pool can include:

[0064] Step S410: Obtain the historical order flow and the misclassification value of the historical order flow.

[0065] As mentioned above, the historical order flow is the entire process determined in advance based on each stage of the parcel flow. The error score of the historical order flow is used to reflect the probability of route misjudgment within the historical order flow. If the error score of the historical order flow is higher, it means that the route misjudgment within that historical order flow is more likely. Conversely, if the error score of the historical order flow is lower, it means that the route misjudgment within that historical order flow is less likely.

[0066] In some optional embodiments, the misclassification value of historical order flow can be defined as the difference between the total number of orders flowing to the destination and the maximum number of orders on the line. That is, the line with the highest number of orders is selected as the misclassified order value. Of course, other methods can also be used to determine the misclassification value of historical order flow in this embodiment of the disclosure, and there are no limitations on this.

[0067] Step S420: Construct a sampling flow pool based on historical order flows with misclassification values ​​greater than the first preset threshold.

[0068] Order flows with misclassification values ​​greater than a first preset threshold can be used to construct a sample flow pool. The first preset threshold can be flexibly selected according to actual needs and is not restricted. Alternatively, historical order flows can be sorted in reverse order of misclassification value, and a specified number of flows at the top of the sorted list can be used as a sample flow pool.

[0069] The sampling flow pool will be used in the modeling and logistics simulation process, and will be introduced below.

[0070] This embodiment of the disclosure constructs a sampling flow pool to provide a data foundation for subsequent modeling, enabling the route prediction model to be trained on samples with high misclassification costs, thereby improving the model's performance in predicting routes for orders with high misclassification costs.

[0071] In an exemplary embodiment, determining whether the flow direction corresponding to the waybill to be processed meets the preset conditions may further include:

[0072] If the waybill flow direction corresponding to the waybill to be processed is in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed meets the preset condition; or, if the waybill flow direction corresponding to the waybill to be processed is not in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed does not meet the preset condition.

[0073] Parcels typically flow along planned routes. In this embodiment, uncertain (high cost of misclassification) flow directions can be replaced by models to simulate existing rules. Therefore, at the current operation node, it can be determined whether the flow direction corresponding to the waybill to be processed is in the sampling flow direction pool. If the flow direction is in the sampling flow direction pool, it indicates that the flow direction of the waybill is likely to be misclassified. Then, the line prediction result of the line prediction model is used to assist in determining the simulated predicted line.

[0074] Conversely, if the flow direction of the waybill is not in the sampling flow direction pool, it indicates that the possibility of route misjudgment for the waybill is small. In this case, the logistics simulation system can be used to simulate route planning for the waybill to be processed in order to obtain the simulated predicted route of the waybill to be processed.

[0075] This embodiment of the present disclosure determines whether the flow direction of the waybill to be processed meets the preset conditions at the current operation node, and diverts the waybill to be processed to a route prediction method assisted by machine learning model and a route decision method based on the existing rules of the logistics simulation system. While maintaining the overall framework and processing capacity of the existing logistics simulation system, a route prediction model is used to assist waybills that may be misjudged, thereby improving the accuracy of route planning for such waybills and assisting the logistics simulation system in outputting accurate simulation results.

[0076] In one exemplary embodiment, an implementation method for training a training model is also provided. For example... Figure 5 Before obtaining the route prediction result at the current operating node, this embodiment of the disclosure further includes steps S510 to S530 to obtain the route prediction model:

[0077] Step S510: Obtain historical waybill samples covered by the historical waybill flow direction in the sampling flow direction pool.

[0078] It is possible to obtain the actual waybills covered by each historical waybill flow direction in the sampling flow pool as historical waybill samples.

[0079] Step S520: Based on all possible routes between each operational stage of the historical waybill sample, obtain the candidate route sample corresponding to the historical waybill sample.

[0080] For each historical waybill sample, its corresponding candidate route sample can be determined based on the possible routes between each operational step (i.e., all) of them.

[0081] Figure 6 A schematic diagram illustrating the determination of candidate line samples is shown, such as... Figure 6As shown, the operation links from O to D include operation links 1 to 3. There is at least one possible route between every two adjacent operation links. Therefore, by expanding by operation link, the candidate route samples corresponding to the historical waybill samples are a×b×c×d. Of course, Figure 6 This is just a simple example; the actual operation and possible circuits in real-world applications are much more complex, which will not be elaborated upon here.

[0082] Step S530: Based on the real-time waybill characteristics, offline statistical characteristics, and candidate route samples of historical waybill samples, construct training samples to train the model to be trained and obtain the route prediction model.

[0083] When determining historical waybill samples and candidate route samples, real-time waybill features and offline statistical features corresponding to the historical waybill samples can also be obtained.

[0084] In some optional embodiments, the real-time and offline statistical features of historical waybill samples specifically involve waybill features, site features, and route features. Waybill features include waybill route type, region type, and cargo type, while site features include operating outlets and outlet types. Route features include route identification, vehicle type, and route departure time. Of course, the embodiments of this disclosure can flexibly adjust each feature according to actual needs, and no specific restrictions are imposed on this.

[0085] Based on this, training samples are constructed using historical waybill samples – {real-time waybill features, offline waybill features, candidate route samples} – to train the model to be trained and obtain the route prediction model.

[0086] Specifically, the sample labels for training samples are determined based on the actual routes selected by historical waybill samples. If a historical waybill sample actually selected the candidate route, the sample is labeled as 1; otherwise, it is labeled as 0.

[0087] The route prediction model can be selected according to actual needs, choosing the appropriate type of machine learning model. For example, the deepMF (Factorization Machine) model is commonly used in click-through rate (CTR) prediction scenarios, performing well when handling high-dimensional, sparse data. Its core consists of an FM part and a deep part. The FM part captures second-order interactions of features, while the deep part extracts higher-order feature interactions. Since route prediction scenarios often involve sparse features such as location and waybill attributes, and these features exhibit strong interaction effects (e.g., waybill route type and location type, waybill volume and vehicle type), this embodiment can use the deepFM model as the route prediction model. However, the route prediction model in this embodiment is not limited to this.

[0088] The following description uses the deepFM model as an example to illustrate the line prediction model of this disclosure.

[0089] DeepFM's predictions can be expressed as:

[0090]

[0091] Among them, y FM For the FM part of the prediction, y DNN This is a prediction for the deep part.

[0092] In the input layer of the model, the input data includes discrete and continuous features. Each discrete feature is represented as a one-hot encoded vector, and each continuous feature is represented as its value. Therefore, each input data point can be transformed into (x, y), where... It is a d-dimensional vector. For example... Figure 7 The network structure of the FM part is shown. The FM part uses the clicks represented by features as the coefficients of the interaction terms, as shown in formula (2). Formula (2) shows that if a feature combination that has not appeared in the past occurs during inference, deepFM can also... <V i V j The dot product of ">" represents the interaction effect of features.

[0093]

[0094] where x∈R d Let w be the input feature vector, and w ∈ R. d V represents the weight coefficients of the first-order features. i ∈R k For the hidden representation of the features, the dimension k of the hidden vector is specified in advance.

[0095] like Figure 8 The network structure of the deep part is shown. The deep part extracts high-order feature representations through multiple fully connected layers. It's worth noting that the embedding input to the first fully connected layer is shared with the latent vector of the FM layer, thus eliminating feature engineering and achieving end-to-end training. The output of the embedding layer is as follows:

[0096] a 0 =[e1,e2,...,e m (3)

[0097] Where e i for field i The embedding vector and the hidden vector of the FM layer; m is the number of feature fields.

[0098] During model training, training samples are input into the model to be trained for iterative training, and the parameters in the model to be trained are adjusted to obtain the route prediction model.

[0099] This embodiment of the disclosure compares and predicts routes using different models. The regression prediction LGB model, FIBiNet model, and simple strategy model (prediction based on the highest route freight volume of the previous day at the same time) are compared with the deepFM model used in this embodiment. During this comparison, a total of 1.28 million waybills are obtained, broken down by operational stage. Data from the previous 7 days is used for training, and data from the last two days is used for prediction. The training set contains 970,000 data entries, with 3.27 million data entries intersecting with all possible routes. The validation set contains 310,000 data entries, with 2.07 million data entries intersecting with all possible routes. The data has 75 feature fields. Under these experimental conditions, comparing the route prediction performance of each model shows that the deepFM model has higher prediction accuracy. The route prediction accuracy is obtained using the following formula:

[0100]

[0101] The route prediction model, once trained, processes the offline statistical features and real-time features of historical waybills to obtain route prediction results. This process can improve the accuracy of route planning by fully considering the interaction effects of features in various dimensions.

[0102] In specific implementation, route statistical features and site statistical features can be updated in a pre-set manner, such as using a T+1 method to update route statistical features and site statistical features to obtain offline statistical features of historical waybills. Then, the real-time features and offline statistical features of historical waybills are correlated and input into the route prediction model to obtain the routes to be selected and the corresponding prediction probabilities.

[0103] Optionally, for historical waybills, the real-time features of the historical waybills can be enumerated. For example, continuous features can be discretized based on chi-square binning to reduce the number of enumerations and avoid unmatchable feature values ​​in future inferences. Then, the enumeration of real-time features can be associated with offline statistical features and input into the route prediction model to obtain the inference results.

[0104] In an exemplary embodiment, an implementation method for optimizing historical waybill flow directions is provided. Obtaining historical waybill samples covered by historical waybill flow directions in the sampling flow direction pool may include:

[0105] For each historical waybill flow direction, samples are taken from the waybills covered by the historical waybill flow direction to obtain historical waybill samples for the historical waybill flow direction based on the sampling results.

[0106] Specifically, a sampling ratio can be set, and the statistical significance of ensuring that the daily order volume of the route is not less than the preset number of orders can be achieved. The number of orders sampled is determined according to min(sampling ratio * flow order volume, preset number). Of course, the specific method of sampling historical order samples is not limited to this and can be flexibly selected according to the actual situation.

[0107] By sampling historical waybill samples from historical waybill flows, computational resources can be further saved and processing efficiency improved.

[0108] In an exemplary embodiment, in actual implementation, it is impractical to call the route prediction model at each operation stage of the logistics simulation system. In a GPU environment (P40 1 card), the calculation time for a single inference of the model is about 0.2ms. When this inference time is multiplied by the daily volume of tens of millions of packages and the number of operation stages, the time cost is unbearable.

[0109] Based on this, the present disclosure provides a space-for-time trade-off method. Specifically, the route prediction results obtained from the route prediction model are pre-stored, and during actual simulation in the logistics simulation system, the route prediction results are loaded into the system's memory. This allows the logistics simulation system to directly utilize existing results when needed, accelerating inference speed and avoiding the impact of the route prediction model on the system's operational efficiency.

[0110] In an optional embodiment, the route prediction results (the correspondence between real-time waybill characteristics and offline statistical characteristics of historical waybills and the candidate route and the predicted probability of the candidate route) can be stored in the form of a mapping table. Only the necessary real-time input for receiving simulation is retained in the mapping table, and the remaining statistical characteristics are hidden in the background calculation, thereby avoiding the impact of the route prediction model on the operating efficiency of the logistics simulation system.

[0111] Based on the foregoing exemplary embodiments, such as Figure 9 As shown, if the flow direction of the waybill to be processed meets the preset conditions, the target route is determined from the candidate routes in the route prediction results. The process of determining the target route as the simulation prediction route for the waybill to be processed can further include steps S910 and S920:

[0112] Step S910: Based on the real-time waybill characteristics and corresponding relationships of the waybills to be processed, determine the candidate routes from the routes to be selected.

[0113] In simulation applications, the route prediction results can be loaded into the memory of the logistics simulation system first, so that the logistics simulation system can obtain the target prediction results of feature matching based on the real-time features of the waybill to be processed, that is, determine the candidate route from the routes to be selected.

[0114] It should be noted that the next route for the real-time waybill feature input needs to be predicted and is therefore unknown, while the other features (offline statistical features) are all known. Since the correspondence includes an enumeration of all possible routes for the waybill flow, the other features can be used for indexing to determine candidate routes.

[0115] Step S920: Based on the predicted probabilities corresponding to the candidate routes, determine the target route from the candidate routes to identify the target route as the simulation prediction route for the waybill to be processed.

[0116] After obtaining all candidate routes, the candidate routes with prediction probabilities greater than the second preset threshold can be identified as target routes according to their corresponding prediction probabilities. The target routes are then used as the simulation prediction routes for the waybills to be processed, thereby replacing the route planning of the logistics simulation system with the prediction results of the route prediction model and improving the accuracy of route planning in the logistics simulation process.

[0117] The second preset threshold can be set according to actual needs, and there are no restrictions on it.

[0118] In one optional embodiment, if the predicted probabilities of all candidate routes are less than a second preset threshold, the logistics simulation system is used to perform simulated route planning for the order to be processed in order to obtain the simulated predicted route of the order to be processed.

[0119] Specifically, if the predicted probabilities of all candidate routes for the order to be processed are lower than the second preset threshold, then the existing rules of logistics simulation will be used to give the route decision, that is, to obtain the simulated predicted route of the order to be processed.

[0120] This disclosure combines machine learning models and a logistics simulation system to plan routes in the logistics simulation process, enabling the logistics simulation system to better simulate real-world scenarios and improve the accuracy of logistics simulation.

[0121] Figure 10 A schematic diagram of route decision-making in a logistics simulation system according to an embodiment of the present disclosure is shown below. Figure 10 The logistics simulation process of the embodiments of this disclosure is described in detail.

[0122] First, obtain historical data.

[0123] Among these features, historical order flow and misclassification values ​​can be obtained from the big data platform, and a sampling flow pool can be constructed based on historical order flow with misclassification values ​​greater than a first preset threshold.

[0124] Simultaneously, historical waybill samples covering the historical waybill flow direction are acquired from the sampling flow pool. Based on all possible routes between each operational stage of the historical waybill samples, candidate route samples corresponding to the historical waybill samples are obtained. Finally, training samples are constructed based on the real-time waybill characteristics, offline statistical characteristics, and candidate route samples of the historical waybill samples. The sample labels of the training samples are determined according to the actual routes selected by the historical waybill samples.

[0125] Secondly, the training samples are used to train the model to be trained, and the route prediction model is obtained.

[0126] This involves persisting the model parameters of the line prediction model.

[0127] Furthermore, relevant data from historical waybills are obtained and input into the route prediction model.

[0128] This allows for the acquisition of data from the latest day, including offline statistical features of historical waybills and real-time waybill features. These features are then input into the route prediction model to obtain the candidate route and its predicted probability.

[0129] Next, the correspondence between the real-time and offline statistical features of historical waybills and the routes to be selected and their predicted probabilities is stored, that is, the corresponding model results are displayed on the data platform.

[0130] Finally, the simulation application process.

[0131] In simulation applications, the route prediction results are obtained at the current operation node, which means that the content of the failed bids is loaded into the logistics simulation system.

[0132] If the waybill flow direction corresponding to the waybill to be processed is in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed meets the preset conditions. Then, the target route is determined from the routes to be selected in the route prediction results, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0133] If the flow direction of the waybill to be processed is not in the sampling flow direction pool, then the flow direction of the waybill to be processed does not meet the preset conditions. In this case, the logistics simulation system will perform simulated route planning for the waybill to be processed to obtain the simulated predicted route of the waybill to be processed.

[0134] At this point, the next operational step can be determined at the current operational node, and the logistics simulation process can continue.

[0135] It should be noted that the specific details involved in this section have been described in the exemplary embodiments above, and will not be repeated here.

[0136] The logistics simulation method provided in this disclosure is applied to a logistics simulation system. At the current operation node, route prediction results are obtained. These results are pre-processed using a route prediction model to analyze the offline statistical characteristics and real-time characteristics of historical waybills. If the flow direction of the waybill to be processed meets preset conditions, a target route is determined from the candidate routes in the route prediction results, thus establishing the target route as the simulation prediction route for the waybill to be processed. On one hand, this process in the logistics simulation system uses a route prediction model to simulate real-world route decisions. When the flow direction of the waybill to be processed meets preset conditions, a target route is determined from the route prediction results to replace the route generation rules of the logistics simulation system, guiding and improving the accuracy of route planning in the logistics simulation process, thereby enhancing the accuracy of the logistics simulation. On the other hand, the logistics simulation method of this disclosure improves the accuracy of route prediction without significantly impacting the main architecture and existing processes of the logistics simulation system, exhibiting high adaptability and broad application prospects.

[0137] It should be noted that the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of this disclosure, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0138] Further reference Figure 11 As shown, an exemplary embodiment of this disclosure provides a logistics simulation device 1100, including a result acquisition module 1110 and a route determination module 1120, wherein:

[0139] The result acquisition module 1110 is used to acquire the route prediction result at the current operation node. The route prediction result is obtained by processing the offline statistical features and real-time features of historical waybills in advance through the route prediction model.

[0140] The route determination module 1120 is used to determine a target route from the candidate routes in the route prediction results if the flow direction of the waybill corresponding to the waybill to be processed meets the preset conditions, so as to determine the target route as the simulation prediction route of the waybill to be processed.

[0141] In an exemplary embodiment, the route determination module 1120 is further configured to perform: if the flow direction of the waybill to be processed does not meet the preset conditions, then perform simulated route planning on the waybill to be processed through the logistics simulation system to obtain the simulated predicted route of the waybill to be processed.

[0142] In an exemplary embodiment, the result acquisition module 1110 is further configured to perform: acquiring historical order flow direction and misclassification value of the historical order flow direction; wherein the misclassification value is used to reflect the probability of a route misjudgment in the historical order flow direction;

[0143] A sampling flow pool is constructed based on the historical order flow with a misclassification value greater than a first preset threshold.

[0144] In an exemplary embodiment, the route determination module 1120 is further configured to perform: if the waybill flow direction corresponding to the waybill to be processed is in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed satisfies the preset condition; or, if the waybill flow direction corresponding to the waybill to be processed is not in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed does not satisfy the preset condition.

[0145] In an exemplary embodiment, the result acquisition module 1110 is further configured to perform: acquiring historical waybill samples covered by historical waybill flow direction in the sampling flow direction pool; obtaining candidate route samples corresponding to the historical waybill samples based on all possible routes between each operation stage of the historical waybill samples; constructing training samples based on the real-time waybill characteristics, offline statistical characteristics and candidate route samples of the historical waybill samples, so as to train the model to be trained using the training samples to obtain the route prediction model; wherein, the sample label of the training samples is determined according to the actual selected route of the historical waybill samples.

[0146] In an exemplary embodiment, the result acquisition module 1110 is further configured to perform: for each of the historical waybill flows, sampling the waybills covered by the historical waybill flows to obtain a historical waybill sample of the historical waybill flows based on the sampling results.

[0147] In an exemplary embodiment, the result acquisition module 1110 is further configured to perform: loading the route prediction result into the memory of the logistics simulation system, wherein the route prediction result includes the correspondence between the real-time waybill characteristics and offline statistical characteristics of the historical waybills and the candidate route and the predicted probability of the candidate route.

[0148] In an exemplary embodiment, the route determination module 1120 is further configured to perform: determining candidate routes from the routes to be selected based on the real-time route characteristics of the to-be-processed waybill and the correspondence; and determining the target route from the candidate routes based on the predicted probability corresponding to the candidate routes, so as to determine the target route as the simulation prediction route of the to-be-processed waybill.

[0149] In an exemplary embodiment, the route determination module 1120 is further configured to: determine the candidate routes with a predicted probability greater than a second preset threshold as the target routes, so as to determine the target routes as the simulation predicted routes of the waybill to be processed; or, if the predicted probabilities of the candidate routes are all less than the second preset threshold, then perform simulation route planning on the waybill to be processed through the logistics simulation system to obtain the simulation predicted routes of the waybill to be processed.

[0150] The specific details of each module in the above-mentioned logistics simulation device have been described in detail in the method section of the implementation. For any undisclosed details, please refer to the implementation content of the method section. That is, the explanation and beneficial effects of the logistics simulation method in the above-mentioned embodiments are also applicable to the logistics simulation device 1100 of this disclosure, and will not be elaborated here.

[0151] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0152] The following reference Figure 12 To describe an electronic device 1200 according to such an embodiment of the present disclosure. Figure 12 The electronic device 1200 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0153] like Figure 12As shown, the electronic device 1200 is manifested in the form of a general-purpose computing device. The components of the electronic device 1200 may include, but are not limited to: at least one processing unit 1210, at least one storage unit 1220, a bus 1230 connecting different system components (including storage unit 1220 and processing unit 1210), and a display unit 1240.

[0154] The storage unit stores program code that can be executed by the processing unit 1210, causing the processing unit 1210 to perform the steps described in the "Exemplary Methods" section above according to various exemplary embodiments of this disclosure.

[0155] Storage unit 1220 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 12201 and / or cache memory 12202, and may further include a read-only memory (ROM) 12203.

[0156] Storage unit 1220 may also include a program / utility 12204 having a set (at least one) of program modules 12205, such program modules 12205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0157] Bus 1230 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0158] Electronic device 1200 can also communicate with one or more external devices 1300 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 1200, and / or with any device that enables electronic device 1200 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 1250. Furthermore, electronic device 1200 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 1260. As shown, network adapter 1260 communicates with other modules of electronic device 1200 via bus 1230. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 1200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0159] Furthermore, exemplary embodiments of this disclosure also provide a computer-readable storage medium storing a program product capable of implementing the methods described above. In some possible embodiments, various aspects of this disclosure may also be implemented as a program product including program code that, when run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0160] It should be noted that the computer-readable medium disclosed herein may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0161] In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0162] Furthermore, program code for performing the operations of this disclosure can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0163] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. A logistics simulation method, characterized in that, Applications in logistics simulation systems include: Obtain the route prediction result at the current operation node. The route prediction result is obtained by processing the offline statistical characteristics and real-time route characteristics of historical waybills in advance through the route prediction model. If it is determined that the flow direction of the waybill to be processed meets the preset conditions, then the target route is determined from the candidate routes in the route prediction results, so as to determine the target route as the simulation prediction route of the waybill to be processed.

2. The method according to claim 1, characterized in that, The method further includes: If the flow direction of the waybill to be processed does not meet the preset conditions, the logistics simulation system is used to perform simulated route planning for the waybill to be processed in order to obtain the simulated predicted route of the waybill to be processed.

3. The method according to claim 1 or 2, characterized in that, The method further includes: Obtain historical order flow direction and misclassification value of the historical order flow direction; wherein, the misclassification value is used to reflect the probability of misjudging the route of the historical order flow direction; A sampling flow pool is constructed based on the historical order flow with a misclassification value greater than a first preset threshold.

4. The method according to claim 3, characterized in that, Determining whether the flow direction corresponding to the pending waybill meets the preset conditions includes: If the waybill flow direction corresponding to the waybill to be processed is in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed satisfies the preset condition; Alternatively, if the waybill flow direction corresponding to the waybill to be processed is not in the sampling flow direction pool, then the waybill flow direction corresponding to the waybill to be processed does not meet the preset condition.

5. The method according to claim 3, characterized in that, Before obtaining the route prediction result at the current operating node, the method further includes: Obtain historical waybill samples covered by the historical waybill flow direction in the sampling flow direction pool; Based on all possible routes between each operational stage of the historical waybill sample, candidate route samples corresponding to the historical waybill sample are obtained. Based on the real-time waybill characteristics and offline statistical characteristics of the historical waybill samples, as well as the candidate route samples, training samples are constructed to train the model to be trained and obtain the route prediction model. Specifically, the sample label of the training sample is determined based on the actual route selected by the historical waybill sample.

6. The method according to claim 5, characterized in that, The step of obtaining historical waybill samples covered by the historical waybill flow direction in the sampling flow direction pool includes: For each of the historical waybills, the waybills covered by the historical waybill flow are sampled to obtain a historical waybill sample for the historical waybill flow based on the sampling results.

7. The method according to claim 1, characterized in that, The step of obtaining the route prediction result at the current operating node includes: The route prediction results are loaded into the memory of the logistics simulation system. The route prediction results include the correspondence between the real-time and offline statistical features of the historical waybills and the routes to be selected and the predicted probabilities of the routes to be selected.

8. The method according to claim 7, characterized in that, If it is determined that the flow direction of the waybill to be processed meets the preset conditions, then the target route is determined from the candidate routes in the route prediction results, so as to determine the target route as the simulation prediction route of the waybill to be processed, including: Based on the real-time route characteristics of the waybill to be processed and the corresponding relationship, candidate routes are determined from the routes to be selected; Based on the predicted probabilities corresponding to the candidate routes, the target route is determined from the candidate routes, so as to determine the target route as the simulation prediction route of the waybill to be processed.

9. The method according to claim 8, characterized in that, The step of determining the target route from the candidate routes based on the predicted probabilities corresponding to the candidate routes, so as to determine the target route as the simulation prediction route for the waybill to be processed, includes: Candidate routes with a predicted probability greater than a second preset threshold are identified as target routes, and the target routes are identified as the simulated predicted routes of the waybills to be processed. Alternatively, if the predicted probabilities of all candidate routes are less than the second preset threshold, then the logistics simulation system is used to perform simulated route planning on the order to be processed in order to obtain the simulated predicted route of the order to be processed.

10. A logistics simulation device, characterized in that, Applications in logistics simulation systems include: The result acquisition module is used to acquire the route prediction result at the current operation node. The route prediction result is obtained by processing the offline statistical features and real-time features of historical waybills in advance through the route prediction model. The route determination module is used to determine a target route from the candidate routes in the route prediction results if the flow direction of the waybill to be processed meets the preset conditions, so as to determine the target route as the simulation prediction route of the waybill to be processed.

11. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 9 by executing the executable instructions.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 9.

Images