A logistics scheduling scheme generation method, device, equipment and medium

Abstract

The application discloses a logistics scheduling scheme generation method, which comprises the following steps: obtaining industrial side data and social side data and fusing the same into a multi-dimensional feature vector; inputting the multi-dimensional feature vector into a pre-trained risk calculation model after obtaining the multi-dimensional feature vector; dynamically generating adaptive risk calculation parameters according to the current specific data characteristics, thereby significantly improving the accuracy of risk value evaluation; and finally, when the delay risk value exceeds a preset risk threshold, calling a preset scheduling optimization model, solving the scheduling optimization model under the constraint condition according to the current logistics vehicle state, transportation task and path information, and taking minimizing the material transportation time as the target, so as to generate a logistics scheduling scheme. Through the above method, the efficiency and accuracy of logistics scheduling are significantly improved, the negative influence of material delay on production is reduced, and economic losses caused by the shutdown of production lines due to material shortage can be greatly reduced.

Description

Technical Field

[0001] This application relates to the field of logistics scheduling technology, specifically to a method, apparatus, equipment, and medium for generating logistics scheduling schemes. Background Technology

[0002] As the new energy vehicle industry gradually shifts from the traditional "mass production-delivery" model to order-driven production, new model launches often face the challenge of a surge in orders and backlogs reaching hundreds of thousands of vehicles. Under this model, delays in any link of the supply chain (such as supplier production or material distribution) can directly extend the delivery period for users, leading to a significant decline in customer satisfaction. To address these risks, traditional supply chain management primarily relies on two methods: firstly, continuous manual monitoring of inventory, orders, supplier status, and logistics information, with emergency measures such as coordination and communication or activation of alternative solutions upon detecting anomalies; secondly, automatic monitoring and early warning based on pre-set fixed rules in warehouse management systems (such as safety stock thresholds and material expiration dates). However, warehouse material status monitoring, early warning, and response mechanisms based on manual experience and fixed rules all have limitations and are ill-suited to the complexity and dynamism brought about by the significant increase in the number and variety of orders under the order-driven model. Summary of the Invention

[0003] This application provides a method, apparatus, equipment, and medium for generating logistics scheduling schemes to solve the above-mentioned technical problems.

[0004] This application provides a method for generating a logistics scheduling scheme, including: Acquire industry-side data and social-side data, and fuse the industry-side data and social-side data to obtain a multi-dimensional feature vector; Based on the multi-dimensional feature vector and the pre-trained risk calculation model, risk calculation parameters are generated; and based on the multi-dimensional feature vector and the risk calculation parameters, the delay risk value of the supply chain materials is calculated; and When the delay risk value exceeds the preset risk threshold, a preset scheduling optimization model is invoked. Based on the current logistics vehicle status, transportation task and route information, the scheduling optimization model is solved under the constraint condition with the goal of minimizing material transportation time, thereby generating a logistics scheduling scheme.

[0005] In one embodiment of this application, calculating the delay risk value of supply chain materials based on the multi-dimensional feature vector and the risk calculation parameters includes: Based on the industry-side data, the social-side data, and the risk calculation parameters, calculate the supply risk, logistics risk, demand risk, and quality risk respectively. Coupling the supply risk with the probability of delays represented by the logistical risk, and with the severity of the consequences represented by the demand risk, yields coupled risk; and The independent failure risk characterized by the quality risk is superimposed with the coupled risk to obtain the delay risk value of the supply chain materials.

[0006] In one embodiment of this application, the risk calculation parameter includes a first weighting coefficient, and the method for calculating the supply risk includes: Obtain supplier capacity utilization, current material production progress, ratio of material inventory level to material demand, and historical average delivery delay time; Based on the first weighting coefficient, the supplier's capacity utilization rate, current material production progress, ratio of material inventory level to material demand, and historical average delivery delay time are integrated to obtain the basic delay factor. Based on the aforementioned basic delay factor and environmental risk factor, a comprehensive delay value is calculated; and Supply risk is calculated based on the comprehensive delay value.

[0007] In one embodiment of this application, the method for calculating the demand risk includes: Acquire the line-side inventory data of the target material on the target production line at the target time and the material consumption rate of the target production line for the target material; Based on the line-side inventory data and the material consumption rate, calculate the estimated run-out time for the target material; and Calculate demand risk based on the estimated material shortage time.

[0008] In one embodiment of this application, the risk calculation parameter includes a second weighting coefficient, and the method for calculating the quality risk includes: The quality indicators of the material production batch are obtained, including: the statistical process control indicators of the target production batch corresponding to the target material issued by the supplier, the quality complaint rate determined based on similar historical batches, and the number of defects that do not meet the quality requirements found in the factory quality inspection. Based on the second weighting coefficient, the statistical process control indicator, the quality complaint rate, and the number of defects that do not meet quality requirements are integrated to obtain an intermediate risk value; and The quality risk is calculated based on the aforementioned intermediate risk value.

[0009] In one embodiment of this application, the method for calculating the logistics risk includes: Obtain the historical average speed of logistics vehicles, the instantaneous speed at the current moment, and the length of each segment in the logistics route; For each road segment, the historical average speed is adjusted based on traffic impact factors, weather impact factors, event impact factors, and current speed inertia factors to obtain the predicted speed for each road segment; wherein, the traffic impact factors are determined based on the degree of traffic impact, the weather impact factors are determined based on the degree of meteorological impact, the event impact factors are determined based on the degree of event impact, and the current speed inertia factor is determined based on the ratio of the instantaneous speed at the current moment to the historical average speed; Based on the length of each road segment and the corresponding predicted speed, the predicted arrival time of the logistics vehicle is calculated. Based on the departure time of the logistics vehicle, the length of each road segment, and the planned speed of the logistics vehicle, the planned arrival time of the logistics vehicle is calculated; and The logistics delay value is calculated based on the predicted arrival time and the planned arrival time; Logistics risk is calculated based on the aforementioned logistics delay value.

[0010] In one embodiment of this application, the risk calculation parameters include nonlinear adjustment parameters, and the method for obtaining the traffic impact degree value includes: The average vehicle speed of the target road segment at the target time, the free-flow vehicle speed of the target road segment, and the congestion level value of the target road segment at the target time are obtained. Based on the ratio of the average vehicle speed to the free-flow vehicle speed, and calculated using the nonlinear adjustment parameters, the baseline delay bias of the target road segment at the target time is obtained; and Based on the baseline delay bias and the congestion level value, the traffic impact level value is calculated.

[0011] In one embodiment of this application, the risk calculation parameter includes a third weighting coefficient, and the method for obtaining the meteorological impact degree value includes: Acquire meteorological data for the target area during the target time period, wherein the data includes at least the weather type and warning level; The basic weight of the meteorological type is determined based on the meteorological type, and the multiplier of the warning level is determined based on the warning level; Based on the basic weight of the meteorological type and the multiplier of the warning level, the meteorological type component is calculated; The intensity component is calculated by fusing wind speed, rainfall, and visibility based on the aforementioned third weighting coefficient; and The meteorological impact degree value is calculated based on the meteorological type component and the intensity component.

[0012] In one embodiment of this application, the risk calculation parameters include a fourth weighting coefficient and a fifth weighting coefficient, and the method for obtaining the event impact value includes: Acquire data for the target event, including event type, restriction level multiplier, event impact scale value, importance coefficient of involved entities, information source credibility coefficient, and verification status coefficient; Based on the basic weight of the event type and the restriction level multiplier corresponding to the event type, the basic event type component is calculated; The event intensity component is calculated by fusing the ratio of the event impact scale value to the preset maximum event scale reference value and the importance coefficient of the involved entities with the fourth weighting coefficient. The credibility coefficient of the information source and the verification status coefficient are fused using the fifth weighting coefficient to calculate the event credibility component; and Based on the basic event type component, the event intensity component, and the credibility component, the event impact value is calculated.

[0013] In one embodiment of this application, the risk calculation parameters include adjustment parameters, and the method for obtaining the current velocity inertia factor includes: Obtain the target vehicle's historical average speed and its instantaneous speed at the current moment; and The ratio of the historical average velocity to the instantaneous velocity is adjusted based on the adjustment parameters to obtain the velocity inertia factor.

[0014] In one embodiment of this application, generating risk calculation parameters based on the multi-dimensional feature vector and the pre-trained risk calculation model includes: The multidimensional independent variables are input into a pre-trained risk calculation model, and the risk calculation parameters are output through the risk calculation model. The risk calculation parameters are obtained by training a neural network with the historical values ​​of the multidimensional independent variables as input, the historical material delay risk values ​​as output, and the objective of minimizing the binary cross-entropy loss based on risk classification.

[0015] This application provides a logistics scheduling scheme generation device, comprising: The data fusion module is used to acquire industry-side data and social-side data, and to fuse the industry-side data and social-side data to obtain a multi-dimensional feature vector; The delay risk calculation module is used to generate risk calculation parameters based on the multi-dimensional feature vector and the pre-trained risk calculation model; and to calculate the delay risk value of the supply chain materials based on the multi-dimensional feature vector and the risk calculation parameters; and The scheme optimization module is used to call a preset scheduling optimization model when the delay risk value exceeds a preset risk threshold. Based on the current logistics vehicle status, transportation task and route information, the module solves the scheduling optimization model with the goal of minimizing material transportation time under the constraints, thereby generating a logistics scheduling scheme.

[0016] This application provides a logistics scheduling scheme generation device, comprising: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, enable the logistics scheduling scheme generation device to implement the logistics scheduling scheme generation method.

[0017] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform a logistics scheduling scheme generation method.

[0018] The beneficial effects of this application are as follows: By acquiring and fusing industry-side and social-side data into a multi-dimensional feature vector, this application can simultaneously capture the internal operating status of the supply chain and external environmental disturbances. This avoids inaccurate risk predictions caused by relying on a single data source and enhances the ability to quantify potential delay risks in complex environments. Furthermore, by inputting the multi-dimensional feature vector into a pre-trained risk calculation model, adaptive risk calculation parameters can be dynamically generated based on the specific data characteristics, significantly improving the accuracy of risk assessment. Through these methods, the efficiency and accuracy of logistics scheduling are significantly improved, the negative impact of material delays on production is reduced, and economic losses caused by production line shutdowns due to material shortages are greatly minimized.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

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

[0021] In the attached diagram: Figure 1 This is a flowchart illustrating a logistics scheduling scheme generation method according to an embodiment of this application; Figure 2 This is a schematic flowchart of a genetic algorithm according to an embodiment of this application; Figure 3 This is a schematic diagram of a coding scheme for logistics scheduling according to an embodiment of this application; Figure 4 This is a schematic diagram illustrating the process of calculating the delay risk value of supply chain materials according to an embodiment of this application; Figure 5 This is a flowchart illustrating a method for determining supply risk according to an embodiment of this application; Figure 6 This is a flowchart illustrating a method for calculating demand risk according to an embodiment of this application; Figure 7 This is a flowchart illustrating a method for calculating quality risk according to an embodiment of this application; Figure 8 This is a flowchart illustrating a method for calculating logistics risk according to an embodiment of this application. Figure 9 This is a schematic flowchart of a method for obtaining traffic impact level values ​​according to an embodiment of this application; Figure 10 This is a flowchart illustrating a method for obtaining weather influencing factors according to an embodiment of this application; Figure 11 This is a flowchart illustrating a method for obtaining the degree of impact of an event according to an embodiment of this application; Figure 12 This is a flowchart illustrating a method for obtaining the current velocity inertia factor according to an embodiment of this application; Figure 13 This is a schematic diagram of the process for generating risk calculation parameters according to an embodiment of this application; Figure 14 This is a schematic diagram of a logistics scheduling scheme generation device according to an embodiment of this application; Figure 15 A schematic diagram of a computer system suitable for implementing the memory of the embodiments of this application is shown. Detailed Implementation

[0022] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0023] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the shape, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0024] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0025] Please see Figure 1 , Figure 1 This is a flowchart illustrating a logistics scheduling scheme generation method according to an embodiment of this application. The logistics scheduling scheme generation method includes at least steps S110 to S130, which are described in detail below: Step S110: Obtain industry-side data and social-side data, and fuse the industry-side data and social-side data to obtain a multi-dimensional feature vector; Industry-side data refers to operational data from enterprises and their supply chains, which can be obtained from the industry brain database via API (Application Programming Interface). Specifically, industry-side data can include supply chain information and logistics vehicle information. Supply chain information includes supplier lists, product technology levels, real-time production capacity data, inventory levels, quality certification information, and precise geographic coordinates. Logistics vehicle information includes basic vehicle information, real-time location and trajectory, stops along the route, vehicle maintenance status (cabin temperature, engine status, door open / closed status), and order plans. Social-side data refers to event information from the external public environment, which can be obtained from the social-side database via API. Specifically, social-side data can include real-time collection of meteorological data (earthquakes, typhoons, heavy rain and snow), public health event data (regional lockdowns), traffic event data (highway closures, port congestion), and public opinion data (fires, strikes, accidents, etc. extracted from news and social media).

[0026] The integration of social and industry data involves standardizing heterogeneous social and industry data through methods such as special character removal, space standardization, missing value imputation, timestamp unification, and geocoding, thereby transforming multi-source heterogeneous data into structured objects with a unified time reference.

[0027] Step S120: Generate risk calculation parameters based on multi-dimensional feature vectors and pre-trained risk calculation models; and calculate the delay risk value of supply chain materials based on multi-dimensional feature vectors and risk calculation parameters. The risk calculation model is a model that can dynamically output risk calculation parameters based on the current supply chain status, and simultaneously output a delay risk value based on these parameters. The risk calculation model can be a deep learning model, which is a computational model capable of learning complex patterns and nonlinear relationships from large amounts of data. Examples include Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks.

[0028] In this step, the risk calculation model receives multi-dimensional feature vectors as input, extracts high-order interaction relationships between features through hidden layers, and finally outputs parameters such as risk weights, coupling coefficients, and threshold parameters applicable to the current scenario. Then, based on the risk calculation parameters and coupling rules output by the risk calculation model, risk calculation is performed to obtain the delay risk value of supply chain materials.

[0029] Step S130: When the delay risk value exceeds the preset risk threshold, the preset scheduling optimization model is invoked. Based on the current logistics vehicle status, transportation task and route information, the scheduling optimization model is solved under the constraint condition with the goal of minimizing the material transportation time, thereby generating a logistics scheduling scheme.

[0030] The scheduling optimization model can be solved using genetic algorithms or reinforcement learning methods, with the goal of minimizing the latest completion time among all transportation tasks. The optimal matching scheme of vehicle-task-path is output by solving the scheduling optimization model.

[0031] The status of logistics vehicles can include the vehicle's current location, available load capacity, and estimated idle time; the transportation task can include the type and quantity of materials to be transported, the pickup point, the delivery point, and the expected delivery time; the route information can include the distance of different routes and the estimated travel time.

[0032] In this step, the delay risk value is compared with a preset risk threshold. When the delay risk value exceeds the preset risk threshold, the system automatically invokes a pre-set scheduling optimization model. This model takes the current logistics vehicle status, transportation task, and route information as input, aims to minimize material transportation time, and solves the problem under corresponding constraints to generate an optimal logistics scheduling scheme, thereby achieving proactive response to delay risks and dynamic allocation of transportation resources.

[0033] This application acquires and merges industry-side and social-side data into a multi-dimensional feature vector, simultaneously capturing the internal operating status of the supply chain and external environmental disturbances. This avoids inaccurate risk predictions caused by relying on a single data source and enhances the ability to quantify potential delay risks in complex environments. By inputting the multi-dimensional feature vector into a pre-trained risk calculation model, adaptive risk calculation parameters can be dynamically generated based on the specific data characteristics, significantly improving the accuracy of risk assessment. Through this method, the efficiency and accuracy of logistics scheduling are significantly improved, the negative impact of material delays on production is reduced, and the economic losses caused by production line shutdowns due to material shortages are greatly minimized.

[0034] For social data, it can include unstructured text, JSON data returned by APIs, geolocation information, CSV files (Comma-Separated Values), etc. Therefore, it is necessary to convert unstructured text into structured objects. In one embodiment, converting unstructured text into structured objects includes: The LSTM-GCN joint model (Long Short-Term Memory - Graph Convolutional Network Joint Model) is used to model the graph structure of text data and extract complex relationships between entities. The LSTM-GCN joint model outputs entity label sequences and entity relationship triples simultaneously. Finally, the extraction results are output in a structured manner according to a predefined event template through a post-processing module.

[0035] Relation extraction includes: sequence encoding, graph structure modeling and graph convolutional network (GCN) encoding, and multi-task decoding.

[0036] In sequence encoding, the original text data (unstructured text) is first segmented into individual word units. For example, inputting a news sentence... The word sequence is decomposed using a Byte Pair Encoding Tokenizer (BPE), and then each word is mapped to a fixed-dimensional real-valued vector, i.e., a word vector, using a pre-trained word embedding model (such as Word2Vec, GloVe, or FastText). ,in , The word embedding dimension is then used. Next, these word vectors are fed into a Bidirectional Long Short-Term Memory (Bi-LSTM) network. The Bi-LSTM consists of a forward LSTM and a backward LSTM; the forward LSTM uses the formula... Processing information from the beginning to the end of the sequence, the backward LSTM uses the formula... Information is processed from the end of the sequence to the beginning. Through this bidirectional processing, Bi-LSTM can fully capture the forward and backward contextual information of each word in the sentence. Finally, the final contextual representations of each word are concatenated to generate the contextual semantic representation of each word. . ,in, The hidden layer dimension of a unidirectional LSTM.

[0037] Graph structure construction and graph convolutional network (GCN) encoding include: using words in the word vector sequence as nodes, and constructing a relation graph based on the syntactic dependencies and sequence adjacency relationships between words; In the process of constructing the relationship graph, after obtaining the contextual semantic representation of each word, each word in the text is treated as an independent node in the graph. The relationship graph uses...G express, ,node V For each word in the sentence, edge E includes syntactic dependency edges and sequential adjacency edges. Syntactic dependency edges connect word pairs with direct dependencies by parsing the sentence using a dependency parser. Sequential adjacency edges connect adjacent words in the text. In sequential adjacency edges, each word is connected to itself (self-loop) and to the next word in the sequence to ensure the flow of local contextual information. Defined as: if node I and nodes J If there is an edge between them, then Otherwise, it is 0. A self-loop is usually also included, i.e. ,in M It is an identity matrix.

[0038] Multi-task decoding includes: using contextual semantic representation as the initial node features of the graph convolutional network, performing message propagation and aggregation on the relation graph through the graph convolutional network, and obtaining node representations with syntactic structural information.

[0039] In the multi-task decoding process, the contextual semantic representation of each word is used as the initial feature vector of the corresponding node in the Graph Convolutional Network (GCN). A Graph Convolutional Network is a neural network model for processing graph-structured data. It learns node representations by performing multi-layer iterative message propagation and aggregation operations on the graph.

[0040] The output of LSTM is the context semantic representation. As the initial input to the Graph Convolutional Network (GCN). l The graph convolutional network (GCN) with multiple layers performs message propagation and aggregation using the following formula:

[0041] in, It is the first graph convolutional network (GCN) network. l Layer node representation, This represents the dimension of the vectors represented by the nodes in a graph convolutional network. The dimension of the nodes in the first layer of a graph convolutional network (GCN) is equal to the dimension of the LSTM output vector. . yes The degree matrix is ​​a diagonal matrix. . It is a symmetric normalized Laplacian matrix used for stable training. yes l The layer's trainable weight matrix. ReLU is the activation function. After... l After passing through a layered graph convolutional network (GCN), node representations rich in syntactic structure information are obtained. .

[0042] In the process of identifying entities and classifying entity relationships, based on the final node representation, the identification of entities and the classification of entity relationships are performed to obtain entity label sequences and entity relationship triples.

[0043] Represent the output nodes of the Graph Convolutional Network (GCN) Given a fully connected layer and a Softmax function (normalized exponential function), perform BIOES (Begin, Inside, Outside, End, Single) sequence entity recognition. The recognition formula is as follows:

[0044] in, Indicates the first I The final representation of each node, This represents the weight matrix of the fully connected layer for entity recognition. This represents the bias vector of the fully connected layer for entity recognition. and These are trainable parameters; Indicates the first I Entity tags for each word Indicates the first I Each word is labeled with a probability distribution for each entity label, and the model is trained by maximizing the likelihood probability of the correct label sequence.

[0045] For each entity pair Construct a comprehensive representation of the relationships between entities. r To obtain an entity representation, the average of the representations of all words in the entity is typically used, as shown in the following expression:

[0046] v s Indicates the subject entity, This represents the set of indices of all words corresponding to the subject entity. Indicates the number of words contained in the subject entity. v 0 represents the object entity. This represents the set of indices of all words corresponding to the subject entity. Indicates the number of words contained in the object entity.

[0047] Concatenate the two entity representations, the element-wise product of the two entity representations, and the sentence-level representation (such as [CLS] tags or average pooling of the entire sentence): .

[0048] The entity pair feature vector representing relation classification. Sentence-level representation, which can be a vector labeled [CLS]. Relationship classification, classification formula: .

[0049] in, This represents the weight matrix of the fully connected layer for relation classification. This represents the bias vector of the fully connected layer for relational classification. and These are trainable parameters; Represents entity pairs The predicted probability distribution of various relation labels. Predicted relationship labels, including the "no relationship" category.

[0050] After obtaining the entity label sequence and entity relation triples, the entity label sequence and entity relation triples are integrated in a structured manner according to a predefined event template to obtain structured information for calculating the delay risk value.

[0051] The predefined event templates are structured frameworks used to describe specific types of events (such as traffic congestion events, severe weather events, supply chain disruption events, etc.). Each template contains a series of pre-defined slots to populate specific information extracted from entity label sequences and entity relationship triples. For example, a traffic congestion event template might contain slots for "event type (traffic congestion)," "location," "time of occurrence," "impact level," and "duration." By mapping the entities identified in the entity label sequence and the entity relationships extracted from the entity relationship triples to these template slots, the textual information can be structurally integrated to obtain structured objects.

[0052] In one embodiment, the scheduling optimization model includes a set of available logistics vehicles, a set of transportation tasks, decision variables, an optimization objective function, and constraints, as detailed below: Available logistics vehicle collection V , .in, Indicates the first k The vehicle number is the serial number of the logistics vehicle. K This represents the total number of logistics vehicles; express The start time for the next transportation task; express Maximum load-bearing capacity.

[0053] Transportation task set , .in, Indicates the first j One transportation task,J This indicates the total number of transportation tasks. express The task attribute is a binary variable, representing... Is it a delayed task? express The weight of the shipment Indicates material j A set of optional transportation routes. In this implementation example, the logistics vehicles have the same maximum load capacity, and the weight of the materials transported is an integer multiple of the maximum load capacity of the logistics vehicle. ,in, express The Middle m A number of optional paths express The number of available paths. , , ,in, express The first in u Each section of road, express The number of road sections in the middle, express The set formed by their locations Indicates the road sections passed by each logistics vehicle. The collection of time spent, express pass The time spent. The time can be obtained from historical data.

[0054] The decision variables of the scheduling optimization model are and . This is a binary variable indicating whether to... Assigned to . A binary variable, representing Choose .

[0055] The optimization objective of the scheduling optimization model is to minimize the material transportation time, and its expression is as follows:

[0056] in, This coefficient represents the importance of a normal transportation task. When... When the value is 0, the model does not consider the transportation time of non-delayed materials.

[0057] The constraints of the scheduling optimization model include task allocation constraints, vehicle selection constraints, and capacity constraints. Their expressions are as follows: Task allocation constraints

[0058] Vehicle selection constraints

[0059] Capacity constraints ' Among these constraints, the allocation constraint indicates that a vehicle can only transport one type of material task. The vehicle selection constraint indicates that a vehicle can only choose one route, and the capacity constraint indicates that the capacity of the selected vehicle must exceed the weight of the material to be transported.

[0060] In one embodiment, a genetic algorithm is used to solve the scheduling optimization model. The genetic algorithm is one of the classic swarm intelligence algorithms. A flowchart of the genetic algorithm is shown below. Figure 2 As shown, this algorithm simulates gene mutation and crossover behavior, and through continuous iterative optimization, outputs the optimal logistics vehicle scheduling scheme. To achieve interaction between the algorithm and the model, the design and coding scheme of this implementation example is as follows: Figure 3 As shown in the diagram, the scheme consists of two parts: a vehicle selection layer and a route selection layer. In the vehicle selection layer, each column corresponds to a transportation task, and its encoded value is an integer representing the nth vehicle selected sequentially from the currently available vehicle candidate set. Figure 3 Taking column 3 as an example, a code value of "1" indicates that the vehicle ranked first among the remaining vehicles is selected for the third transportation task. In the route selection layer, each column also corresponds one-to-one with a transportation task, and its code value represents the specific route number selected by the vehicle from the set of available routes after assignment. Again, using column 3 as an example, a code value of "3" indicates that after assigning a vehicle to this task, the third of the four available routes was selected as the actual transportation route.

[0061] Please see Figure 4 , Figure 4 This is a schematic diagram illustrating the process of calculating the delay risk value of supply chain materials according to an embodiment of this application. Figure 4 In this process, based on multi-dimensional feature vectors and risk calculation parameters, the delay risk value of supply chain materials is calculated, including at least steps S410 to S430, which are detailed below: Step S410: Calculate supply risk, logistics risk, demand risk, and quality risk based on multi-dimensional feature vectors and risk calculation parameters, respectively. Among them, supply risk represents the risk that the materials supplied will not be delivered on time; logistics risk represents the risk that the materials will be delayed during transportation; demand risk represents the risk that the production consumption rate exceeds expectations or that the production plan changes, resulting in a more urgent material demand in terms of time than originally planned; and quality risk represents the risk of material quality deterioration in transit after passing the supplier's quality inspection. Step S420: Couple the probability of delay in the supply risk and logistics risk representation with the severity of the consequences of the demand risk representation to obtain coupled risk; In this step, the probability of delay occurring can be determined by... To express, , Indicates the weighting coefficient. Indicates supply risk, Indicates logistics risks, It indicates the severity of the consequences.

[0062] In this step, during the calculation of the probability of delay, supply risk and logistical risk are weighted by weighting coefficients β1 and β2 respectively to obtain the probability of delay. Coupling specifically refers to combining the probability of delay with the severity of its consequences through multiplication to obtain the coupling risk. .

[0063] Step S430: The independent failure risk and coupled risk of the quality risk characterization are superimposed to obtain the delay risk value of the supply chain materials.

[0064] In this step, the quality risk value is weighted by a weighting coefficient β3, and then added to the aforementioned coupling risk. The resulting delay risk value is a dimensionless scalar ranging from 0 to 1, which can be directly compared with a preset risk threshold.

[0065] The delay risk value can be expressed by the following formula:

[0066] in, Indicates in t Time, materials j From suppliers i To the production unit m During the distribution process, by the transportation unit k The resulting delay risk value. , , , These represent supply risk, logistics risk, demand risk, and quality risk, respectively. Indicates the probability of delay in the arrival of materials. This indicates the severity of the consequences of the delay. Multiplying the two means that if the production line has sufficient inventory (low demand risk), the overall risk will not be amplified; conversely, once the production line is about to run out of materials (high demand risk), any slight delay may cause huge losses, and the overall risk will increase sharply. Quality risk is treated as a separate item because it could lead to complete material failure. , , The weighting coefficients can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees and neural networks). After the supplier completes material production and delivery, .

[0067] Please see Figure 5 , Figure 5 This is a flowchart illustrating a method for determining supply risk according to an embodiment of this application. The risk calculation parameters include a first weighting coefficient. Figure 5 The method for calculating this supply risk includes at least steps S510 to S540, which are detailed below: Step S510: Obtain supplier capacity utilization rate, current material production progress, ratio of material inventory level to material demand, and historical average delivery delay time. Among these metrics, supplier capacity utilization reflects the saturation level of a supplier's current production capacity. Current material production progress directly reflects the completion status of target material production at the supplier. The ratio of material inventory level to material demand measures the supplier's ability to cover future material needs with its current inventory. Historical average delivery delay time, based on the supplier's past delivery performance, reflects the average delay time in historical orders.

[0068] Step S520: Based on the first weighting coefficient, the supplier's capacity utilization rate, the current material production progress, the ratio of material inventory level to material demand, and the historical average delivery delay time are integrated to obtain the basic delay factor. The fusion process can employ a linear weighted summation method, where the basic delay factor equals the sum of the products of each indicator value and its corresponding first weight coefficient. The first weight coefficient adjusts the contribution of three indicators—supplier capacity utilization, current material production progress, and the ratio of material inventory level to material demand—to the basic delay factor. The first weight coefficient can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees and neural networks).

[0069] Step S530: Calculate the comprehensive delay value based on the basic delay factor and the environmental risk factor; In this step, the impact of external environmental factors on supply risk is further considered on top of the basic delay factor, making the assessment more comprehensive. Environmental risk factors refer to uncontrollable external factors that may affect supplier production or logistics, such as natural disasters, policy changes, public health emergencies, and geopolitical conflicts. These factors may lead to production disruptions or logistical obstacles for suppliers, thereby exacerbating supply delays. Environmental risk factors can be obtained and assessed in real time based on external data sources (such as meteorological bureaus, news media, and government announcements) and quantified into a numerical value. For example, when a typhoon warning is issued for a region, the environmental risk factor can be set to a higher value.

[0070] Step S540: Calculate supply risk based on the comprehensive delay value.

[0071] In this step, when determining supply risk based on the comprehensive delay value, the comprehensive delay value is mapped to a supply risk level or a specific risk value. This can be determined using a preset mapping function. For example, the comprehensive delay value can be normalized and mapped to the range of 0-1.

[0072] This application assesses the delay risk of supply chain materials by acquiring multi-dimensional indicators such as supplier capacity utilization, current material production progress, the ratio of material inventory level to material demand, and historical average delivery delay time. These indicators are then integrated based on a first weighting coefficient to accurately capture the impact of the supplier's internal operational status on material delivery. Furthermore, environmental risk factors are incorporated to obtain a comprehensive delay value. This comprehensive delay value considers not only the supplier's internal situation but also the uncertainties of the external environment, making the identified supply risk more comprehensive and realistic, and effectively improving the accuracy of delay risk calculation.

[0073] In one embodiment, supply risk It can be calculated using the following formula:

[0074]

[0075]

[0076] in, The delay is mapped to the interval [0,1] to represent the probability or degree of risk. Indicates environmental risk factors, Indicates supplier i Geographical location Indicates supplier i exist t Capacity utilization rate at any given moment (between 0 and 1). Indicates supplier iAt time t, the production progress of the current material j order is (completed quantity / total order quantity). Indicates supplier i exist t Materials of the moment j Inventory levels. Indicates supplier i Order materials j The quantity required. Indicates supplier i materials j Historical average delivery delay time. , , This represents the first weighting coefficient.

[0077] Please see Figure 6 , Figure 6 This is a flowchart illustrating a method for calculating demand risk according to an embodiment of this application. Figure 6 The method for calculating demand risk includes at least steps S610 to S630, which are detailed below: Step S610: Obtain the line-side inventory data of the target material on the target production line at the target time and the material consumption rate of the target production line for the target material. Line-side inventory refers to the quantity of materials stored alongside the production line to meet immediate production needs. Material consumption rate refers to the quantity of target materials consumed by the target production line per unit of time.

[0078] Step S620: Calculate the estimated material shortage time for the target material based on the line-side inventory data and material consumption rate; The estimated material shortage time is used to measure the urgency of material supply. It is usually calculated by dividing the current lineside inventory data by the material consumption rate.

[0079] Step S630: Calculate demand risk based on the expected material shortage time.

[0080] In this step, when determining demand risk based on the estimated material shortage time, the estimated material shortage time is mapped to a specific risk value. This can be determined using a preset mapping function. For example, the estimated material shortage time can be normalized and mapped to a range of 0-1.

[0081] This application acquires real-time data on line-side inventory and material consumption rates from the production line, and calculates the estimated material shortage time accordingly. This allows demand risk assessment to move beyond vague, empirical judgments and instead be based on the actual operational status of the production site. This risk determination method based on estimated material shortage time can intuitively reflect the urgency of material supply, providing a more accurate and reliable input for calculating the delay risk value of supply chain materials.

[0082] In one embodiment, demand risk It can be calculated using the following formula:

[0083]

[0084] in, Indicates material j On the production line m The estimated time of material shortage. express t Time production line m materials j Off-line inventory. Indicates production line m materials j The consumption rate is usually the production cycle time. BOM (Bill of Materials) usage.

[0085] Please see Figure 7 , Figure 7 This is a flowchart illustrating a method for calculating quality risk according to an embodiment of this application. The risk calculation parameters include a second weighting coefficient. Figure 7 The method for calculating quality risk includes at least steps S710 to S730, which are detailed below: Step S710: Obtain the quality indicators of the material production batch. The quality indicators of the material production batch include: the statistical process control indicators of the target production batch corresponding to the target material issued by the supplier, the quality complaint rate determined based on similar historical batches, and the number of defects that do not meet the quality requirements found in the factory quality inspection. Statistical process control indicators (SPCIs) refer to the indicators generated by the statistical tools and methods used by suppliers to monitor and control product quality during the production process. If SPCIs show that the process is out of control, it indicates that the batch of materials has a high quality risk.

[0086] The quality complaint rate, determined based on similar historical batches, refers to the average proportion of complaints calculated by analyzing past batches of the target material that share similarities with the current batch in terms of production processes, raw materials, suppliers, and production equipment. The quality complaint rate reflects the probability of quality problems occurring in similar batches during actual use or subsequent stages, thus indirectly predicting the potential quality risks of the current batch.

[0087] The number of defects found during factory quality inspection refers to the specific number of defective or non-conforming products that are detected in the final quality inspection stage before the materials leave the supplier's factory, indicating that they do not meet the preset quality standards or technical specifications. The number of defects reflects the actual quality level of the batch of materials at the time of shipment; the higher the number of defects, the higher the quality risk of the batch of materials.

[0088] Step S720: Based on the second weighting coefficient, the statistical process control indicators, the quality complaint rate, and the number of defects that do not meet quality requirements are fused to obtain the intermediate risk value; In this step, after obtaining these multi-dimensional quality indicators, the statistical process control indicators, quality complaint rate, and the number of defects that do not meet quality requirements are further integrated based on a second weighting coefficient to obtain an intermediate risk value. The integration process can employ a linear weighted summation method, meaning the intermediate risk value equals the sum of the products of each indicator and its corresponding second weighting coefficient. The second weighting coefficient is used to adjust the contribution of the statistical process control indicators, quality complaint rate, and the number of defects that do not meet quality requirements to quality risk. The second weighting coefficient can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees and neural networks).

[0089] Step S730: Calculate quality risk based on intermediate risk value.

[0090] In this step, when determining quality risk based on intermediate risk values, the intermediate risk values ​​are mapped to specific risk values. This mapping can be determined using a pre-defined mapping function. For example, the intermediate risk values ​​can be normalized and mapped to the range of 0-1.

[0091] This application comprehensively considers multiple quality indicators, such as statistical process control indicators of material production batches, quality complaint rates determined based on similar historical batches, and the number of defects that do not meet quality requirements found in factory quality inspection, and integrates them based on a second weighting coefficient to obtain a more comprehensive and accurate intermediate risk value, thereby accurately determining quality risks.

[0092] In one embodiment, quality risk The expression is as follows:

[0093] in, Indicates supplier i Material issuance j The corresponding production batch. express Statistical process control indicators; for example, the number of times key process parameters (such as pressure and temperature) exceed the limit during material generation, with higher values ​​indicating poorer stability. Indicates batch-based The end-customer quality complaint rate of similar historical batches; for example, the average quality level of historical batches using the same equipment and the same raw material suppliers. This indicates the number of minor defects found during the supplier's factory quality inspection; , , , This represents the second weighting coefficient.

[0094] Please see Figure 8 , Figure 8 This is a flowchart illustrating a method for calculating logistics risk according to an embodiment of this application. Figure 8 The method for calculating this logistics risk includes at least steps S810 to S860, which are detailed below: Step S810: Obtain the historical average speed of the logistics vehicle, the instantaneous speed at the current moment, and the length of each segment in the logistics route. Historical average speed reflects a vehicle's typical driving performance on a specific road segment. The instantaneous speed at the current moment reflects the vehicle's current real-time operating status. The length of each segment in the logistics route is a fundamental parameter for calculating travel time, ensuring the accuracy of time calculations.

[0095] Step S820: For each road segment, adjust the historical average speed based on traffic impact factor, weather impact factor, event impact factor, and current speed inertia factor to obtain the predicted speed for each road segment; wherein, the traffic impact factor is determined based on the traffic impact degree value, the weather impact factor is determined based on the meteorological impact degree value, the event impact factor is determined based on the event impact degree value, and the current speed inertia factor is determined based on the ratio of the instantaneous speed at the current moment to the historical average speed. The traffic impact factor quantifies the effect of traffic congestion on vehicle speed. It is determined based on the traffic impact level of the road segment; for example, the traffic impact factor increases when a high level of congestion is detected. The weather impact factor quantifies the effect of weather conditions on vehicle speed. It is determined based on the weather impact level; for example, the weather impact factor increases during severe weather events such as rain, snow, fog, or strong winds, leading to a downward adjustment in predicted speed. The event impact factor quantifies the impact of unforeseen events on road capacity. It is determined based on the event impact level; for example, events such as traffic accidents, road construction, and temporary traffic control will reduce the predicted speed of a road segment through this factor. The current speed inertia factor is used to quantify the inertial effect of the vehicle's current driving state on its future short-term speed. It is determined based on the ratio of the instantaneous speed at the current moment to the historical average speed. For example, if the instantaneous speed is significantly higher than the historical average speed, the inertia factor is larger, causing the predicted speed to be slightly adjusted upward; conversely, if the instantaneous speed is lower, the inertia factor is smaller and positive, causing the predicted speed to be slightly adjusted downward, so as to reflect the trend of vehicle speed change.

[0096] Step S830: Based on the length of each road segment and the corresponding predicted speed, calculate the predicted arrival time of the logistics vehicle. After obtaining the predicted speed for each road segment, the predicted time required for a vehicle to travel through that segment can be calculated by dividing the length of each segment by its corresponding predicted speed. By summing the predicted times for all road segments, the total predicted arrival time of the logistics vehicle from its origin to its destination can be obtained.

[0097] Step S840: Based on the departure time of the logistics vehicle, the length of each road segment, and the planned speed of the logistics vehicle, calculate the planned arrival time of the logistics vehicle. The planned arrival time is calculated based on a preset, ideal transportation plan. It is typically calculated using the planned departure time of the logistics vehicles, the length of each route segment, and the preset planned speed.

[0098] Step S850: Calculate the logistics delay value based on the predicted arrival time and the planned arrival time; The logistics delay value is the difference between the predicted arrival time and the planned arrival time. If the predicted arrival time is later than the planned arrival time, it indicates a delay; if it is earlier than or equal to the planned arrival time, the delay value is zero or negative (indicating early arrival).

[0099] Step S860: Calculate logistics risk based on logistics delay value.

[0100] In this step, when determining logistics risk based on logistics delay values, the logistics delay values ​​are mapped to specific risk values. This can be determined using a preset mapping function. For example, the logistics delay values ​​can be normalized and mapped to the range of 0-1.

[0101] This application obtains basic data such as the historical average speed of logistics vehicles, the instantaneous speed at the current moment, and the length of each segment in the logistics route. Based on this, for each segment, the historical average speed is dynamically adjusted by comprehensively considering traffic impact factors, weather impact factors, event impact factors, and current speed inertia factors, thereby obtaining a more accurate predicted speed for each segment. Based on the predicted speed, the predicted arrival time of logistics vehicles can be accurately calculated and compared with the planned arrival time to obtain a quantified logistics delay value. Finally, logistics risk is determined based on the logistics delay value, significantly improving the accuracy of logistics risk assessment.

[0102] In one embodiment, logistical risks It can be calculated using the following formula:

[0103]

[0104]

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111] in, Indicates logistics risks, This represents the material delay for logistics vehicle k at time t. Indicates logistics vehicle k exist t The predicted arrival time of a moment. Indicates logistics vehicle k The planned arrival time N k This indicates the number of road segments on the logistics route of logistics vehicle k. Indicates vehicle k In the n The length of each road segment Indicates vehiclek In the n The predicted speed of each road segment Indicates vehicle k Historical average speed Indicates vehicle k exist t Instantaneous velocity at a given moment Indicates traffic impact factors, Indicates weather influencing factors, Indicates the event's impact factor. Indicates the current velocity inertia factor. Indicates the degree of traffic impact. Indicates the degree of meteorological impact. Indicates the degree of impact of the event. This indicates the adjustment parameter.

[0112] Please see Figure 9 , Figure 9 This is a schematic flowchart illustrating a method for obtaining traffic impact level values ​​according to an embodiment of this application. The risk calculation parameters include nonlinear adjustment parameters. Figure 9 The method for obtaining the traffic impact level value includes at least steps S910 to S930, which are described in detail below: Step S910: Obtain the average vehicle speed of the target road segment at the target time, the free-flow speed of the vehicle on the target road segment, and the congestion level value of the target road segment at the target time. The free-flow speed of vehicles on a target road segment refers to the speed that vehicles would naturally choose on that segment under ideal traffic conditions (i.e., no congestion, no traffic lights, no interference from other vehicles, etc.). Congestion levels are usually divided into multiple levels, and different levels can be represented by different values.

[0113] Step S920: Based on the ratio of the average vehicle speed to the free-flow vehicle speed, and using nonlinear adjustment parameters, the basic delay bias of the target road segment at the target time is calculated. The nonlinear adjustment parameters can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees and neural networks).

[0114] Step S930: Based on the basic delay bias and congestion level value, calculate the traffic impact value.

[0115] In this step, by further integrating the baseline delay bias with the congestion level value, a more comprehensive and accurate traffic impact value can be obtained.

[0116] This application calculates the baseline delay bias by comprehensively considering the average vehicle speed, free-flow vehicle speed, and congestion level of the target road segment at the target time, and by combining nonlinear adjustment parameters. This can more accurately quantify the impact of traffic conditions on the speed of logistics vehicles.

[0117] During implementation, the degree of traffic impact, : ,

[0118] in, This represents the baseline delay bias of road segment seg at time t. This represents the average speed of vehicles on the road segment at time t. This represents the free-flow speed of vehicles on the road segment, and α represents the nonlinear adjustment parameter (usually 0.5 ≤ α ≤ 0.8). Please see Figure 10 , Figure 10 This is a flowchart illustrating a method for obtaining weather influencing factors according to an embodiment of this application. The risk calculation parameters include a third weighting coefficient. Figure 10 The method for obtaining the meteorological impact level value includes at least steps S1010 to S1050, which are described in detail below: Step S1010: Obtain meteorological data for the target area during the target time period. The data shall include at least the meteorological type and warning level. Meteorological types can include typhoons, heavy rain, heavy snow, dense fog, and other weather phenomena. Warning levels indicate the severity of severe weather, such as red, orange, yellow, and blue warnings.

[0119] Step S1020: Determine the basic weight of the meteorological type based on the meteorological type, and determine the warning level multiplier based on the warning level; The basic weight of weather type is used to quantify the impact of each weather type on logistics speed. Different weather types have different degrees of impact on logistics transportation; for example, heavy snow usually has a much greater impact on transportation than light rain. The basic weight of weather type can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees and neural networks).

[0120] The warning level multiplier is used to determine the extent of the impact of the warning level on the baseline; for example, a red warning usually means the most severe traffic disruptions.

[0121] Step S1030: Calculate the meteorological type component based on the basic weight of the meteorological type and the warning level multiplier; In this step, by multiplication, the weather type and warning level are combined to obtain a preliminary assessment that comprehensively reflects the impact of weather phenomena and their warning levels on logistics and transportation, thus providing a more accurate indicator than a single weather type or warning level.

[0122] Step S1040: Based on the third weighting coefficient, wind speed, rainfall and visibility are fused to calculate the intensity component; The fusion process can employ a linear weighted summation method, where the intensity component equals the sum of the products of wind speed, rainfall, and visibility, and their corresponding third weighting coefficients. The third weighting coefficients are used to adjust the contribution of wind speed, rainfall, and visibility to the intensity component. These coefficients can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees or neural networks).

[0123] Step S1050: Calculate the meteorological impact degree value based on the meteorological type component and intensity component.

[0124] In this step, by further integrating the meteorological type component and the intensity component, a more comprehensive and accurate value of the meteorological impact can be obtained.

[0125] This application obtains meteorological data of the target area during the target time period and comprehensively considers multiple dimensions such as weather type, warning level, wind speed, rainfall and visibility, thereby more comprehensively quantifying the impact of weather on logistics and transportation.

[0126] In one embodiment, the meteorological impact level value :

[0127]

[0128]

[0129] in, This indicates the meteorological type component, which includes typhoon, rainstorm, heavy snow, dense fog, and others. Indicates intensity component, The basic weight for meteorological types is 0.5 to 2.0. This is a multiplier for the warning level, which includes red, orange, yellow, and blue warnings. Indicates wind speed. Indicates rainfall amount, Indicates visibility. , , This represents the weighting coefficient, which is configured according to the weather type.

[0130] Please see Figure 11 , Figure 11 This is a flowchart illustrating a method for obtaining the impact value of an event according to an embodiment of this application. The risk calculation parameters include a fourth weighting coefficient and a fifth weighting coefficient. Figure 11 The method for obtaining the impact value of an event includes at least steps S1110 to S1150, which are described in detail below: Step S1110: Obtain data for the target event, including event type, restriction level multiplier, event impact scale value, importance coefficient of involved entities, information source credibility coefficient, and verification status coefficient; The types of events can include, but are not limited to, regional lockdowns, social activities, natural disasters (such as floods, earthquakes, blizzards, fires, etc.), and public health incidents. Each type of event may have a different degree of impact on logistics.

[0131] The restriction level multiplier is used to quantify the degree to which an event restricts traffic flow. For example, the restriction level multiplier for a completely closed road will be higher than that for a partially closed lane or a speed limit.

[0132] The event impact scale value indicates the scope or duration of the event's impact, such as the length of the affected road segment, the area of ​​the affected region, or the expected duration of the event.

[0133] The importance coefficient of the entities involved is used to assess the criticality of the logistics infrastructure (such as key bridges, main roads, and important logistics nodes) affected by the event in the entire logistics network.

[0134] The information source credibility coefficient is used to assess the reliability of the source of event information. For example, information released by official transportation departments is generally more credible, while unverified information on social media is less credible.

[0135] The verification status coefficient indicates whether event information has been verified. For example, verified event information has a higher verification status coefficient.

[0136] Step S1120: Calculate the basic event type component based on the basic weight of the event type corresponding to the event type and the restriction level multiplier; In this step, when calculating the basic event type components, a basic event type weight can be preset for different event types. Then, the basic event type weight is multiplied by the event's restriction level multiplier to obtain the basic event type components.

[0137] Step S1130: The event intensity component is calculated by fusing the ratio of the event impact scale value to the preset maximum event scale reference value and the importance coefficient of the involved entities with the fourth weighting coefficient. In this step, when calculating the event intensity component, the ratio of the event impact scale to a preset maximum event scale reference value is calculated. This ratio is then combined with the importance coefficient of the involved entities, and finally fused using a fourth weighting coefficient. The fusion process can employ a linear weighted summation, meaning the event intensity component equals the sum of the ratio of the event impact scale to the preset maximum event scale reference value and the product of the importance coefficient of the involved entities and its corresponding fourth weighting coefficient. The fourth weighting coefficient adjusts the contribution of the ratio of the event impact scale to the preset maximum event scale reference value and the importance coefficient of the involved entities to the event intensity component. The fourth weighting coefficient can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees or neural networks).

[0138] Step S1140: The credibility coefficient of the information source and the verification status coefficient are fused using the fifth weighting coefficient to calculate the event credibility component. In this step, when calculating the event credibility component, the information source credibility coefficient and the verification state coefficient can be fused using a fifth weighting coefficient. The fusion process can employ a linear weighted summation, meaning the event credibility component equals the sum of the products of the information source credibility coefficient and the verification state coefficient, along with their corresponding fourth weighting coefficients. The fifth weighting coefficient is used to adjust the contribution of the information source credibility coefficient and the verification state coefficient to the event's credibility component. This fifth weighting coefficient can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees or neural networks).

[0139] Step S1150: Calculate the event impact value based on the basic event type component, event intensity component, and credibility component.

[0140] In this step, the calculated basic event type component, event intensity component, and event credibility component are combined to calculate the event impact value.

[0141] This application conducts a multi-dimensional quantitative assessment of emergencies affecting logistics and transportation, taking into account not only the nature and severity of the event itself, but also the actual scale of the event's impact, the importance of the affected entities, and the credibility of the event information, thereby making the calculation of the event's impact value more accurate.

[0142] In one embodiment, the event impact value :

[0143]

[0144]

[0145]

[0146] in, loc Indicates the location where the event occurred; This indicates the basic event type component, which includes regional lockdowns, protests, fires, public health emergencies, and others. Indicates intensity component; Represents the credibility component; Indicates the basic weight of the event type; This represents the restriction level multiplier, with a value range of (0.2-1). Restriction levels include full block, partial block, and advisory block. Indicates the scale of the event's impact; Indicates the maximum reference value for the event; This represents the importance coefficient of the entity involved; if it is a hub in the material logistics and transportation system, the maximum value is 1. This indicates the credibility coefficient of the information source, which includes official announcements, mainstream media, social media, and unverified information. This represents the verification status coefficient, which includes verification statuses such as multi-party confirmation, single source, and pending verification. , Indicates the fifth weighting coefficient, , This represents the fourth weighting coefficient.

[0147] Please see Figure 12 , Figure 12 This is a flowchart illustrating a method for obtaining the current velocity inertia factor according to an embodiment of this application. Figure 12 The method for obtaining the current velocity inertia factor includes at least steps S1210 to S1220, which are described in detail below: Step S1210: Obtain the historical average speed of the target vehicle and its instantaneous speed at the current moment; Historical average speed refers to the average speed of the target vehicle over a period of time, such as the past week, month, or on a specific road segment. Instantaneous speed refers to the real-time speed of the target vehicle at the current moment.

[0148] Step S1220: Adjust the ratio of historical average velocity to instantaneous velocity based on the adjustment parameters to obtain the velocity inertia factor.

[0149] The adjustment parameter can be a constant less than or equal to 1. By multiplying the ratio of the instantaneous speed to the historical average speed by this adjustment parameter, a speed inertia factor that reflects the vehicle's current speed trend and its impact on future speed can be obtained. The adjustment parameter can be determined based on expert experience, historical data analysis (such as regression analysis), or machine learning models (such as decision trees and neural networks).

[0150] This application obtains the historical average speed and the instantaneous speed of the target vehicle, and adjusts the ratio of the two based on adjustment parameters to obtain a factor that can accurately reflect the vehicle's current speed inertia. This makes the predicted speed in logistics risk assessment closer to reality, thereby improving the accuracy of logistics delay value calculation.

[0151] In one embodiment, the risk calculation parameters are generated based on the multi-dimensional feature vector and the pre-trained risk calculation model, including: Multidimensional independent variables are input into a pre-trained risk calculation model, and risk calculation parameters are output through the risk calculation model. The risk calculation parameters are obtained by training a neural network with the historical values ​​of the multidimensional independent variables as input, the historical material delay risk value as output, and the goal of minimizing the binary cross-entropy loss based on risk classification.

[0152] During the training phase of the risk calculation model, a large amount of historical supply chain operation data (i.e., historical values ​​of multi-dimensional independent variables) and corresponding actual material delay risk values ​​are collected. By learning from this historical data, a mapping relationship between independent variables and delay risks is established. Minimizing the binary cross-entropy loss based on risk classification is used as the training objective, enabling the model to effectively classify material delay risks into two categories: "delayed" or "no delay." In this process, its internal parameters are optimized, thereby accurately estimating the risk calculation parameters that affect risk calculation.

[0153] Please see Figure 13 , Figure 13 This is a schematic diagram of the structure of a risk calculation model according to an embodiment of this application. Figure 13 In this model, the risk calculation process consists of an input layer, a hidden layer, and an output layer. The input layer receives a feature vector composed of multi-dimensional independent variables from the supply chain. A： The average speed of vehicles on the road segment at time t Vehicle free-flow speed , meteorological impact level , Rainfall ,visibility Impact of the event Basic event type components The scale of the event's impact Maximum reference value of the event Entity importance coefficient , , Environmental risk factors , ,supplier i At time t, the production progress of the current material j order is... ,supplier i Order materials j Demand quantity ,supplier i materials j Historical average delivery delay time , ,vehicle k Historical average speed , Weather Influencing Factors Event impact factor ,vehicle k exist t instantaneous velocity at a moment Logistics vehicles k Planned arrival time Serious consequences , Statistical process control indicators Based on batch Final client quality complaint rate for similar historical batches The number of minor defects found during the supplier's factory quality inspection eigenvectors A Before input, min-max normalization is performed to improve training stability and mitigate gradient explosion. The hidden layers employ a three-layer fully connected structure, with 128, 64, and 19 neurons respectively, all using ReLU as the activation function. The final layer generates a vector with respect to the target parameters. K Parameter estimation vectors of the same dimension K' (Risk calculation parameters), and normalize the output parameters to ensure that the weight coefficients in the same formula satisfy the constraint that they sum to 1. Among them, K and K' It includes the following elements: nonlinear adjustment parameter α, third weighting coefficient. , , , , Fifth weighting coefficient , Weighting coefficient , , First weighting coefficient , , Adjust parameters Second weighting coefficient , , , .

[0154] Output layer with K' Vector and original input feature vector A As input, let the risk calculation function in the supply chain domain be used. Real-value risk prediction is calculated. Y' It is then compared with a preset risk threshold and converted into a final Boolean value output, which represents... Y' Whether the threshold is exceeded is used to characterize whether the risk exceeds the safety limit.

[0155] The training objective of this risk calculation model is to minimize the binary cross-entropy loss in risk threshold judgment. When the model achieves near-perfect classification accuracy on sufficiently diverse datasets, it indicates that the risk function Function(A,K') formed by the parameter estimation vector K' is highly consistent with the true function Function(A,K) on the decision boundary. Therefore, K' generated internally by the network after training convergence can be regarded as an effective estimate of the true unknown parameter vector K.

[0156] Before training the risk calculation model, the risk categories of the training data need to be labeled. The principle for determining the risk categories is as follows: The process of each batch of materials from supplier production to transportation to the automobile factory is divided into multiple time periods. The risk data of each time period constitutes a set of data. Therefore, a batch of materials corresponds to multiple sets of training data. First, the risk calculation parameters of the risk calculation formula are preset. When a disturbance event causes a material delay accident, the risk value corresponding to the time period after the disturbance event will change abruptly. Therefore, for the data where a material delay accident occurs, the minimum risk value in the data group where the risk value changes abruptly is taken as the risk value threshold for the batch of materials to cause material delay risk, and the data in the time period of risk change is marked as "causing delay risk".

[0157] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0158] Please see Figure 14 , Figure 14 This is a schematic diagram illustrating the principle of a logistics scheduling scheme generation device according to an embodiment of this application. Figure 14The logistics scheduling plan generation device includes: The data fusion module 1410 is used to acquire industry-side data and social-side data, and to fuse the industry-side data and social-side data to obtain a multi-dimensional feature vector; The delay risk calculation module 1420 is used to generate risk calculation parameters based on multi-dimensional feature vectors and a pre-trained risk calculation model; and to calculate the delay risk value of supply chain materials based on multi-dimensional feature vectors and risk calculation parameters. The scheme optimization module 1430 is used to call a preset scheduling optimization model when the delay risk value exceeds the preset risk threshold. Based on the current logistics vehicle status, transportation task and route information, the scheduling optimization model is solved under the constraint condition with the goal of minimizing material transportation time, thereby generating a logistics scheduling scheme.

[0159] It should be noted that the logistics scheduling scheme generation device and the logistics scheduling scheme generation method provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the logistics scheduling scheme generation device provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0160] Embodiments of this application also provide a logistics scheduling scheme generation device, including: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by one or more processors, the memory implements the logistics scheduling scheme generation method in the above embodiments.

[0161] Embodiments of this application also provide one or more machine-readable media having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the logistics scheduling scheme generation method described in the above embodiments.

[0162] Figure 15 A schematic diagram of a computer system suitable for implementing the memory of embodiments of this application is shown. It should be noted that... Figure 15 The computer system with the memory shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0163] like Figure 15As shown, the computer system 1500 includes a Central Processing Unit (CPU) 1501, which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, based on programs stored in Read-Only Memory (ROM) 1502 or programs loaded from storage into Random Access Memory (RAM) 1503. The RAM also stores various programs and data required for system operation. The CPU 1501, ROM 1502, and RAM 1503 are interconnected via a bus 1504. An input / output (I / O) interface 1505 is also connected to the bus 1504.

[0164] The following components are connected to I / O interface 1505: an input section 1506 including a keyboard, mouse, etc.; an output section 1507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1508 including a hard disk, etc.; and a communication section 1509 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 1509 performs communication processing via a network such as the Internet. A drive 1510 is also connected to I / O interface 1505 as needed. Removable media 1511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1510 as needed so that computer programs read from them can be installed into storage section 1508 as needed.

[0165] Specifically, according to embodiments of this application, the processes described in the above-described flowchart can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the logistics scheduling scheme generation method of the aforementioned embodiments. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from removable medium 1511. When the computer program is executed by a central processing unit (CPU) 1501, it performs various functions defined in the system of this application.

[0166] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, 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) 1503, read-only memory (ROM) 1502, erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0167] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in the flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block or combination of blocks in the block diagram or flowchart may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0168] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0169] Another aspect of this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform the aforementioned logistics scheduling scheme generation method. This computer-readable storage medium may be included in the memory described in the above embodiments, or it may exist independently and not incorporated into that memory.

[0170] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the logistics scheduling scheme generation method provided in the various embodiments described above.

[0171] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for generating a logistics scheduling scheme, characterized in that, The method for generating the logistics scheduling scheme includes: Acquire industry-side data and social-side data, and fuse the industry-side data and social-side data to obtain a multi-dimensional feature vector; Based on the multi-dimensional feature vector and the pre-trained risk calculation model, risk calculation parameters are generated; and based on the multi-dimensional feature vector and the risk calculation parameters, the delay risk value of the supply chain materials is calculated; and When the delay risk value exceeds the preset risk threshold, a preset scheduling optimization model is invoked. Based on the current logistics vehicle status, transportation task and route information, the scheduling optimization model is solved under the constraint condition with the goal of minimizing material transportation time, thereby generating a logistics scheduling scheme.

2. The logistics scheduling scheme generation method according to claim 1, characterized in that, The calculation of the delay risk value of supply chain materials based on the multi-dimensional feature vector and the risk calculation parameters includes: The supply risk, logistics risk, demand risk, and quality risk are calculated based on the multi-dimensional feature vector and the risk calculation parameters, respectively. Coupling the supply risk with the probability of delays represented by the logistical risk, and with the severity of the consequences represented by the demand risk, yields coupled risk; and The independent failure risk characterized by the quality risk is superimposed with the coupled risk to obtain the delay risk value of the supply chain materials.

3. The logistics scheduling scheme generation method according to claim 2, characterized in that, The risk calculation parameters include a first weighting coefficient, and the method for calculating supply risk includes: Obtain supplier capacity utilization, current material production progress, ratio of material inventory level to material demand, and historical average delivery delay time; Based on the first weighting coefficient, the supplier's capacity utilization rate, current material production progress, ratio of material inventory level to material demand, and historical average delivery delay time are integrated to obtain the basic delay factor. Based on the aforementioned basic delay factor and environmental risk factor, a comprehensive delay value is calculated; and Supply risk is calculated based on the comprehensive delay value.

4. The logistics scheduling scheme generation method according to claim 2, characterized in that, The calculation method for the demand risk includes: Acquire the line-side inventory data of the target material on the target production line at the target time and the material consumption rate of the target production line for the target material; Based on the line-side inventory data and the material consumption rate, calculate the estimated run-out time for the target material; and Calculate demand risk based on the estimated material shortage time.

5. The logistics scheduling scheme generation method according to claim 2, characterized in that, The risk calculation parameters include a second weighting coefficient, and the method for calculating the quality risk includes: The quality indicators of the material production batch are obtained, including: the statistical process control indicators of the target production batch corresponding to the target material issued by the supplier, the quality complaint rate determined based on similar historical batches, and the number of defects that do not meet the quality requirements found in the factory quality inspection. Based on the second weighting coefficient, the statistical process control indicator, the quality complaint rate, and the number of defects that do not meet quality requirements are integrated to obtain an intermediate risk value; and The quality risk is calculated based on the aforementioned intermediate risk value.

6. The logistics scheduling scheme generation method according to claim 2, characterized in that, The methods for calculating logistics risks include: Obtain the historical average speed of logistics vehicles, the instantaneous speed at the current moment, and the length of each segment in the logistics route; For each road segment, the historical average speed is adjusted based on traffic impact factors, weather impact factors, event impact factors, and current speed inertia factors to obtain the predicted speed for each road segment; wherein, the traffic impact factors are determined based on the degree of traffic impact, the weather impact factors are determined based on the degree of meteorological impact, the event impact factors are determined based on the degree of event impact, and the current speed inertia factor is determined based on the ratio of the instantaneous speed at the current moment to the historical average speed; Based on the length of each road segment and the corresponding predicted speed, the predicted arrival time of the logistics vehicle is calculated. Based on the departure time of the logistics vehicle, the length of each road segment, and the planned speed of the logistics vehicle, the planned arrival time of the logistics vehicle is calculated; and The logistics delay value is calculated based on the predicted arrival time and the planned arrival time; Logistics risk is calculated based on the aforementioned logistics delay value.

7. The logistics scheduling scheme generation method according to claim 6, characterized in that, The risk calculation parameters include nonlinear adjustment parameters, and the method for obtaining the traffic impact level value includes: The average vehicle speed of the target road segment at the target time, the free-flow vehicle speed of the target road segment, and the congestion level value of the target road segment at the target time are obtained. Based on the ratio of the average vehicle speed to the free-flow vehicle speed, and calculated using the nonlinear adjustment parameters, the baseline delay bias of the target road segment at the target time is obtained; and Based on the baseline delay bias and the congestion level value, the traffic impact level value is calculated.

8. The logistics scheduling scheme generation method according to claim 6, characterized in that, The risk calculation parameters include a third weighting coefficient, and the method for obtaining the meteorological impact degree value includes: Acquire meteorological data for the target area during the target time period, wherein the data includes at least the weather type and warning level; The basic weight of the meteorological type is determined based on the meteorological type, and the multiplier of the warning level is determined based on the warning level; Based on the basic weight of the meteorological type and the multiplier of the warning level, the meteorological type component is calculated; The intensity component is calculated by fusing wind speed, rainfall, and visibility based on the aforementioned third weighting coefficient; and The meteorological impact degree value is calculated based on the meteorological type component and the intensity component.

9. The logistics scheduling scheme generation method according to claim 6, characterized in that, The risk calculation parameters include a fourth weighting coefficient and a fifth weighting coefficient, and the method for obtaining the event impact value includes: Acquire data for the target event, including event type, restriction level multiplier, event impact scale value, importance coefficient of involved entities, information source credibility coefficient, and verification status coefficient; Based on the basic weight of the event type and the restriction level multiplier corresponding to the event type, the basic event type component is calculated; The event intensity component is calculated by fusing the ratio of the event impact scale value to the preset maximum event scale reference value and the importance coefficient of the involved entities with the fourth weighting coefficient. The credibility coefficient of the information source and the verification status coefficient are fused using the fifth weighting coefficient to calculate the event credibility component; and Based on the basic event type component, the event intensity component, and the credibility component, the event impact value is calculated.

10. The logistics scheduling scheme generation method according to claim 6, characterized in that, The risk calculation parameters include adjustment parameters, and the method for obtaining the current velocity inertia factor includes: Obtain the target vehicle's historical average speed and its instantaneous speed at the current moment; and The ratio of the historical average velocity to the instantaneous velocity is adjusted based on the adjustment parameters to obtain the velocity inertia factor.

11. The method for generating a logistics scheduling scheme according to any one of claims 1-10, characterized in that, The risk calculation parameters generated based on the multi-dimensional feature vector and the pre-trained risk calculation model include: The multidimensional independent variables are input into a pre-trained risk calculation model, and the risk calculation parameters are output through the risk calculation model. The risk calculation parameters are obtained by training a neural network with the historical values ​​of the multidimensional independent variables as input, the historical material delay risk values ​​as output, and the objective of minimizing the binary cross-entropy loss based on risk classification.

12. A logistics scheduling scheme generation device, characterized in that, include: The data fusion module is used to acquire industry-side data and social-side data, and to fuse the industry-side data and social-side data to obtain a multi-dimensional feature vector; The delay risk calculation module is used to generate risk calculation parameters based on the multi-dimensional feature vector and the pre-trained risk calculation model; and to calculate the delay risk value of the supply chain materials based on the multi-dimensional feature vector and the risk calculation parameters. as well as The scheme optimization module is used to call a preset scheduling optimization model when the delay risk value exceeds a preset risk threshold. Based on the current logistics vehicle status, transportation task and route information, the module solves the scheduling optimization model with the goal of minimizing material transportation time under the constraints, thereby generating a logistics scheduling scheme.

13. A logistics scheduling scheme generation device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, cause the logistics scheduling scheme generation device to implement the logistics scheduling scheme generation method as described in any one of claims 1 to 12.

14. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by the computer's processor, causes the computer to perform the logistics scheduling scheme generation method according to any one of claims 1 to 12.

Images