Intelligent logistics dynamic allocation method and system based on big data analysis

Abstract

This invention discloses an intelligent logistics dynamic allocation method and system based on big data analysis, relating to the fields of logistics management and the Internet of Things (IoT). The method includes: collecting multi-source logistics data and preprocessing it to construct a set of logistics state feature vectors; generating implicit state vectors and identifying potential risks through spatiotemporal and historical correlation analysis; establishing a risk correlation network to calculate propagation results; constructing a collaborative resource pool based on risk results to complete resource and task matching; constructing a path model integrating time, distance, and risk to determine the target path; generating an allocation scheme and dynamically adjusting the execution status in real time. This invention solves the problems of insufficient risk perception, fragmented resource path planning, and lagging dynamic response in existing technologies, thereby improving the operational efficiency of logistics systems. The advantages of this invention are: it can predict potential risks in advance, achieve collaborative optimization of resource paths, significantly reduce vehicle empty load rates and order delay rates, and improve the response speed to abnormal events.

Description

Technical Field

[0001] This invention relates to the fields of logistics management and Internet of Things (IoT) technology, specifically to an intelligent logistics dynamic allocation method and system based on big data analysis. Background Technology

[0002] With the rapid development of e-commerce and the same-city instant delivery industry, logistics order volume has exploded, and delivery scenarios are becoming increasingly complex and varied. Traditional static scheduling models can no longer meet the demands for high-efficiency and high-reliability logistics services. Intelligent logistics dynamic allocation technology based on big data analysis has become a core means to improve system operating efficiency, reduce operating costs, and optimize resource allocation, which is of great significance for promoting the digital transformation of the logistics industry.

[0003] Existing intelligent logistics dynamic allocation methods mostly focus only on the real-time load status of equipment, failing to accurately perceive the distribution patterns of order flow and potential operational risks from a spatiotemporal perspective. They lack quantitative analysis of the risk propagation and diffusion process, and the resource matching and route planning stages are disconnected from each other, resulting in a lag in dynamic adjustment response. This can easily lead to high vehicle empty load rates, frequent order delays, and insufficient ability to handle abnormal events, making it difficult to achieve globally optimal resource scheduling. Therefore, it is necessary to provide intelligent logistics dynamic allocation methods and systems based on big data analysis to solve the problems mentioned above. Summary of the Invention

[0004] To address the aforementioned technical issues, this paper provides an intelligent logistics dynamic allocation method and system based on big data analysis. This technical solution solves the problems of existing intelligent logistics dynamic allocation methods mentioned in the background, which only focus on the real-time load of equipment, lack risk perception and propagation quantitative analysis in the spatiotemporal dimensions, have fragmented resource matching and path planning, and have lagging dynamic adjustments, resulting in low efficiency, many delays, poor anomaly handling, and difficulty in achieving global optimal scheduling.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The intelligent logistics dynamic allocation method based on big data analysis includes the following steps: S1. Collect multi-source data from the logistics system and preprocess it to extract feature parameters that reflect the logistics operation status in order to construct a set of logistics status feature vectors. The multi-source data includes order data, transportation resource data, traffic environment data and historical delivery data. S2. Obtain the changing trend, spatiotemporal correlation, and historical behavior correlation of the feature parameters within a preset time window and perform correlation analysis to determine the correlation strength between each feature parameter in order to generate a corresponding implicit logistics state vector. The implicit logistics state vector is used to characterize the potential operating state of the logistics system and to determine the potential delivery delay risk, potential route congestion risk, and potential capacity imbalance risk based on the changes in correlation strength. S3. Establish a logistics risk association network, and determine the risk propagation weight between each risk node based on the association strength and state propagation relationship between risk nodes, so as to obtain the logistics risk propagation results. S4. Obtain the real-time status of logistics resources and construct a collaborative resource pool based on the results of logistics risk propagation. Determine the dynamic adaptability of each logistics resource to the logistics task based on the correlation between the status of logistics resources and the requirements of logistics tasks, so as to generate logistics resource matching relationships. S5. Obtain the real-time path status and construct a dynamic path cost model by combining the logistics resource matching relationship and the logistics risk propagation result. By predicting the status change trend of candidate paths within a future preset time window, determine the dynamic path cost value corresponding to each candidate path in order to determine the target delivery path. S6. Generate a dynamic allocation plan for logistics resources based on the target delivery route and the matching relationship of logistics resources, and allocate logistics tasks to the corresponding logistics resources.

[0006] In an optional embodiment, the step of acquiring the changing trends, spatiotemporal correlations, and historical behavioral correlations of the feature parameters within a preset time window and performing correlation analysis to determine the correlation strength between each feature parameter specifically includes: S201. Based on the set of logistics status feature vectors, determine the window length and sliding step of the preset sliding time window, and generate a continuous time window sequence; S202. Based on the sequence of logistics status feature vectors within each time window, obtain the spatiotemporal correlation coefficient between any two feature parameters i and j; S203. Obtain the historical delivery database to obtain the historical behavior correlation coefficient between any two feature parameters i and j; S204. Determine the comprehensive correlation strength between feature parameters i and j based on the spatiotemporal correlation coefficient and the historical behavior correlation coefficient:

[0007] In the formula, The comprehensive correlation strength between feature parameter i and feature parameter j within the k-th time window. The dynamic spatiotemporal correlation weight coefficient for the k-th time window. The spatiotemporal correlation coefficient between feature parameter i and feature parameter j within the k-th time window. The correlation coefficient between feature parameter i and feature parameter j is the historical behavior correlation coefficient. The preset time decay factor, This is the window number of the most recent major anomaly event; if no major anomaly event occurred, then... .

[0008] In an optional embodiment, generating the corresponding implicit logistics state vector specifically includes: S205. Based on the comprehensive correlation strength among all feature parameters, construct an asymmetric correlation strength matrix: ; In the formula, d represents the total number of characteristic parameters. To reflect the unidirectional causal relationship between characteristic parameters; S206. Based on the historical abnormal event database, determine the contribution of each characteristic parameter to various logistics risks. Construct the feature weight matrix The elements on the diagonal of the matrix represent the risk contribution of each feature parameter. S207, Correlation Strength Matrix After weighting, the weighted correlation strength matrix is ​​obtained: ; S208. Perform principal component analysis on the weighted correlation strength matrix to obtain the eigenvalues ​​and eigenvectors of the weighted correlation strength matrix; S209, Preset cumulative variance contribution rate threshold and risk explanatory power threshold Determine the number of principal components K, i.e., select those whose cumulative variance contribution rate is greater than or equal to... And the cumulative risk explanatory power is greater than or equal to The minimum value of K; S210. Extract the feature vectors corresponding to the first K principal components and construct a K-row, d-column feature vector matrix, where each row of the matrix corresponds to the feature vector of a principal component. S211. Based on the eigenvector matrix and the average logistics status eigenvector of the current time window, determine the score of each principal component: ; In the formula, Let m be the score of the m-th principal component in the k-th time window, where m = 1, 2, ..., K. Let m be the eigenvector corresponding to the m-th principal component. It is the average value of all logistics status feature vectors within the k-th time window; S212. Based on the scores of all principal components, generate a latent logistics state vector of dimension K. S213. Establish a principal component-risk type mapping table, and map each principal component to one or more specific logistics risk types. The first principal component corresponds to global system risk, the second principal component corresponds to path congestion risk, the third principal component corresponds to capacity imbalance risk, and the fourth principal component corresponds to delivery delay risk. S214. For each principal component m, set a three-level risk threshold: low risk threshold... Medium risk threshold and high risk threshold ,in ; S215. Compare the scores of each principal component in the implicit logistics state vector with the corresponding risk thresholds to determine the risk level corresponding to each principal component. S216. Obtain the prediction accuracy of each principal component for the same type of risk in historical data, and determine the confidence level of the corresponding risk identification. When the score of any principal component exceeds the corresponding medium risk threshold and the confidence level is greater than or equal to the preset confidence threshold, it is determined that there is a potential logistics risk of the corresponding type. S217. Based on the risk level and confidence level of each principal component, generate a logistics risk feature fingerprint containing K elements, where the m-th element is the product of the score of the m-th principal component in the k-th time window and the corresponding risk identification confidence level.

[0009] In an optional embodiment, step S3 specifically includes: S301. Based on the various potential risks identified by the implicit logistics state vector, define a set of risk nodes in the logistics risk association network, containing N risk nodes, each risk node... For each specific type of potential risk, p = 1, 2, ..., N; S302. Based on the comprehensive correlation strength between the feature parameters corresponding to each risk node, construct an N-row, N-column risk association network adjacency matrix: ; In the formula, Risk Node and The weight of the associated edges between them, when When it exceeds the preset association threshold, ,otherwise ; S303. Based on the values ​​of each principal component in the implicit logistics state vector, determine the initial risk value for each risk node: ; in, To be with risk nodes The corresponding number The principal component in the first... The score for each time window, Preset risk node weights; S304. Obtain the state propagation relationship between risk nodes and determine the risk propagation direction matrix: ; in, Indicates risk from nodes propagation to nodes , This indicates that there is no transmission relationship; S305. Based on the adjacency matrix and propagation direction matrix of the risk association network, determine the risk nodes. arrive Risk propagation weight: ; in, This is a preset propagation probability coefficient; S306. Construct an iterative formula for risk propagation to calculate the risk diffusion process in the network: ; In the formula, Risk node after the t-th iteration The risk value, Risk nodes after the (t-1)th iteration The risk value, Risk nodes after the (t-1)th iteration The risk value, Risk from nodes propagation to nodes The propagation weight; S307. Iterate the calculation until the change in risk value of all risk nodes is less than the preset convergence threshold to obtain the final logistics risk propagation result.

[0010] In an optional embodiment, step S4 specifically includes: S401. Based on the results of logistics risk propagation and real-time logistics resource status data, select logistics resources that are in normal and available status, and build an initial collaborative resource pool. S402. Based on the real-time status data of each available logistics resource, construct a resource feature vector of dimension L, where the l-th element of the vector is the value of the l-th resource feature parameter. S403. Based on the demand data of each logistics task to be assigned, construct a task demand feature vector of dimension L, where the l-th element of the vector is the value of the l-th task demand parameter, and l=1,2,...,L; S404. Based on resource feature vectors and task requirement feature vectors, determine the dynamic adaptability of logistics resources to logistics tasks: ; In the formula, The degree of dynamic adaptation of logistics resource a to logistics task b. Let be the resource feature vector of logistics resource 'a'. Let be the feature vector representing the task requirements of logistics task b. This is a diagonal weight matrix, where the diagonal elements are the weights of each feature parameter. It is the weighted L2 norm; S405. Based on the dynamic adaptability between all resources and tasks, construct an adaptability matrix with A rows and B columns. The element in the a-th row and b-th column of the matrix is ​​the dynamic adaptability of logistics resource a to logistics task b, where A is the total number of available logistics resources and B is the total number of logistics tasks to be assigned. S406. Based on the fitness matrix, the Hungarian algorithm is used to solve the optimal resource matching problem, maximize the total fitness, obtain the initial logistics resource matching relationship, and verify whether the initial matching relationship meets the resource load constraints, volume constraints, and time constraints. Matching relationships that do not meet the constraints are adjusted to obtain the final logistics resource matching relationship.

[0011] In an optional embodiment, step S5 specifically includes: S501. Based on logistics resource matching relationships, logistics risk propagation results, and real-time path status data, define a dynamic path cost function: ; In the formula, Value for the path For the cost of time, For the cost of distance, For the cost of risk, , , The preset cost weighting coefficients, and ; S502. Generate a set of candidate delivery routes containing Y candidate routes based on the latitude and longitude coordinates of the starting point and the ending point. S503. Based on historical and real-time path status data, a time series prediction model is used to predict the road traffic speed, congestion index and risk value of each candidate path within a preset time window in the future. S504. Based on the predicted road speed and path length, determine the time cost of the candidate path: ; in, Assign index numbers to candidate paths. Preset the time window length for future path prediction. Candidate paths Total length, Candidate paths The predicted road traffic speed at time t, Candidate paths The predicted congestion index at time t; S505. Determine the distance cost based on the path length and unit distance transportation cost of the candidate paths: ; In the formula, Candidate paths The cost of transportation per unit distance; S506. Based on the results of logistics risk propagation and the predicted risk values ​​of candidate routes, determine the risk cost: ; In the formula, For path At time t, the node is at risk. The extent of the impact, Let be the final risk value of the p-th risk node; S507. Substitute each cost component into the dynamic path cost function to determine each candidate path. Dynamic path value ; S508. Sort the dynamic path cost values ​​of all candidate paths in ascending order and select the candidate path with the lowest cost value as the target delivery path.

[0012] In an optional embodiment, step S6 specifically includes: S601. Based on the matching relationship of logistics resources, allocate a set of corresponding delivery tasks to each logistics resource; S602. Based on the target delivery route and the latitude and longitude of the delivery address for each task, determine the task delivery order for each logistics resource and generate an initial delivery sequence; S603. Determine the load rate of each logistics resource based on its load capacity and volume parameters, as well as the total weight and volume of the assigned tasks. S604. Determine the average load factor of all logistics resources. and standard deviation To achieve load balancing: ; S605. When the load balancing degree LB is greater than the preset balancing threshold, the task allocation of each resource is adjusted, and some tasks of the resource with excessive load rate are transferred to the resource with excessive load rate until the load balancing degree meets the requirements. S606. Verify whether the adjusted delivery plan meets the timeliness requirements of all tasks and the resource endurance requirements, and make secondary adjustments to the parts that do not meet the requirements. S607. Generate a final dynamic allocation scheme for logistics resources, including resource ID, task list, delivery order, target route, and estimated arrival time, and send the final dynamic allocation scheme for logistics resources to the terminal devices of the corresponding logistics resources through the wireless communication network.

[0013] In an optional embodiment, a dynamic adjustment step is also included: S701. Based on the real-time data transmission interface of the logistics resource terminal equipment, the status change information during the execution of logistics tasks is obtained at a preset sampling frequency f, specifically including the real-time location coordinates of the vehicle, driving speed, number of completed tasks, number of remaining tasks, and temperature and humidity status of the goods. S702. Based on the real-time state change information at the current time t, extract the corresponding real-time feature parameters and construct a real-time state feature vector of dimension d. The i-th element in the vector is the value of the i-th real-time feature parameter, where i = 1, 2, ..., d. S703, Obtain the current implicit logistics state vector The corresponding predicted state feature vector is calculated using the following formula. : ,in, This is the K-row, d-column eigenvector matrix constructed in step S210; S704. Determine the real-time state feature vector according to the following formula. With the predicted state feature vector The degree of relative deviation between : In the formula, It is an L2 norm; S705, Degree of relative deviation Deviation threshold from preset The comparison is performed, and the system also checks for pre-set abnormal events such as vehicle malfunctions, road closures, and customer relocations. when Or, if a preset abnormal event is detected, it is determined that the dynamic adjustment conditions are met; When the dynamic adjustment conditions are met, the old data at the corresponding moment in the logistics status feature vector set is replaced with real-time status change information to obtain the updated logistics status feature vector set. S706. Based on the updated set of logistics status feature vectors, re-execute steps S2 to S6 to obtain the updated dynamic allocation scheme for logistics resources, and send it to the terminal devices of the corresponding logistics resources.

[0014] Furthermore, a smart logistics dynamic allocation system based on big data analysis is proposed to execute the allocation method described in any of the preceding items, specifically including: The data acquisition and preprocessing unit is used to collect various types of data from the logistics system and preprocess them to construct a set of logistics status feature vectors. The correlation analysis unit is used to perform correlation analysis on the logistics status feature vector, generate implicit logistics status vectors, and identify potential risks. The risk propagation determination unit is used to establish a logistics risk association network, determine the risk propagation weights, and obtain the logistics risk propagation results. The resource matching unit is used to build a collaborative resource pool, determine the dynamic adaptability of resources and tasks, and generate logistics resource matching relationships. The route planning unit is used to build a dynamic route cost model, determine the dynamic cost of each candidate route, and determine the target delivery route. The allocation and execution unit is used to generate dynamic allocation schemes for logistics resources and assign tasks to corresponding logistics resources; The dynamic adjustment unit is used to acquire information on changes in task execution status in real time, determine whether the dynamic adjustment conditions are met, and update the allocation scheme.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. The intelligent logistics dynamic allocation method based on big data analysis proposed in this solution collects multi-source heterogeneous logistics data and performs standardized preprocessing. It combines spatiotemporal correlation analysis and weighted principal component analysis to mine implicit logistics status and constructs a risk correlation network to simulate the propagation and diffusion process. This enables the early identification and accurate quantification of potential logistics risks, effectively improving the system's risk prediction capability and anti-interference ability. 2. The intelligent logistics dynamic allocation method proposed in this solution, based on big data analysis, constructs a collaborative resource pool based on risk propagation results, uses weighted cosine similarity and Hungarian algorithm to achieve optimal matching of resources and tasks, and constructs a three-dimensional dynamic path cost model that integrates time, distance and risk, thereby realizing the collaborative optimization of resource allocation and path planning, and significantly reducing vehicle empty load rate and overall delivery cost. 3. The intelligent logistics dynamic allocation method based on big data analysis proposed in this solution collects task execution status data in real time and calculates status deviation. It introduces an abnormal event triggering mechanism to realize dynamic rescheduling and constructs a closed loop of the entire process from data perception and risk prediction to scheduling execution. This enables dynamic adaptive adjustment of logistics resources and significantly improves order on-time rate and abnormal event response speed. Attached Figure Description

[0016] Figure 1 This is an overall flowchart of the intelligent logistics dynamic allocation method based on big data analysis proposed in this invention; Figure 2 This is a system framework diagram of the intelligent logistics dynamic allocation system based on big data analysis proposed in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1

[0019] Please see Figure 1 As shown in the figure, this embodiment presents an intelligent logistics dynamic allocation method based on big data analysis. The method includes: Step S1: Collect and preprocess multi-source data from the logistics system to extract feature parameters reflecting the logistics operation status, thereby constructing a set of logistics status feature vectors. Multi-source data includes order data, transportation resource data, traffic environment data, and historical delivery data. Methods for collecting multi-source data include: real-time collection of order data and transportation resource data through the logistics management platform interface; real-time collection of traffic environment data through the traffic information service interface; and acquisition of historical delivery data from a historical database. Specifically, order data includes order ID, delivery address latitude and longitude, cargo weight, cargo volume, delivery time requirements, order placement time, customer contact information, and cargo type; transportation resource data includes vehicle ID, vehicle type, rated load capacity, rated volume, current location latitude and longitude, remaining battery power, vehicle status, driver ID, driver experience, and number of orders completed that day; traffic environment data includes real-time traffic speed, congestion index, road construction information, traffic accident information, weather conditions, visibility, and temperature for each road segment; and historical delivery data includes delivery time, actual driving route, delay status, vehicle trajectory, abnormal event records, and customer reviews for all orders within the past 12 months.

[0020] The preprocessing methods for the collected raw data include: data cleaning, deleting duplicate records and outliers, and using linear interpolation to fill in missing location and status data. Outliers include data with a load exceeding 120% of the vehicle's rated load capacity, a driving speed exceeding 120 km / h, and location coordinates outside the target logistics area; data normalization, using the Z-score normalization method to map all feature parameters to a standard normal distribution interval with a mean of 0 and a variance of 1, eliminating the influence of different units on subsequent analysis; and feature extraction, extracting d feature parameters reflecting the logistics operation status from the preprocessed data to construct a logistics status feature vector set X. For example, d=24, the characteristic parameters include order density, vehicle empty load rate, average driving speed, road congestion index, order timeliness fulfillment rate, average vehicle load rate, delivery station platform utilization rate, average vehicle range, average order waiting time, average road segment travel time, historical delay rate, vehicle failure rate, average driver delivery time, average cargo weight, average cargo volume, proportion of expedited orders, proportion of fresh produce orders, proportion of fragile orders, regional order growth rate, average vehicle mileage, average delivery station processing time, traffic incident rate, weather impact index, and customer complaint rate.

[0021] Step S2: Obtain the changing trends, spatiotemporal correlations, and historical behavioral correlations of the feature parameters within a preset time window, and perform correlation analysis to determine the correlation strength between each feature parameter. This generates a corresponding implicit logistics state vector, which characterizes the potential operating state of the logistics system. Based on changes in correlation strength, potential delivery delay risks, potential route congestion risks, and potential capacity imbalance risks are determined. The method for obtaining the comprehensive correlation strength between feature parameters includes: S201, determining the window length and sliding step size of the preset sliding time window based on the set of logistics state feature vectors, and generating a continuous time window sequence. Specifically, the window length T... w The sliding step size Δt is preset by those skilled in the art according to the real-time requirements of the logistics system. For example, in a same-city instant delivery scenario, the window length T... w =20 minutes, sliding step Δt=3 minutes, generating a continuous time window sequence. Where k is the window number, S202: Based on the logistics status feature vector sequence within each time window, obtain the spatiotemporal correlation coefficient between any two feature parameters i and j. Specifically, the Pearson correlation coefficient is used to calculate the spatiotemporal correlation coefficient. The calculation formula is as follows: In the formula, Let i be the value of the feature parameter i at time t within the time window. Let j be the value of the feature parameter j at time t within the time window. Let i be the average value of the feature parameter i within the time window. Let j be the average value of feature parameter j within the time window, and n be the number of sampling points within the time window; S203, obtain the historical delivery database to obtain the historical behavior correlation coefficient between any two feature parameters i and j. Specifically, extract feature parameter data from the past 6 months from the historical delivery database, and use the same Pearson correlation coefficient calculation method to obtain the historical behavior correlation coefficient. S204. Based on the spatiotemporal correlation coefficient and the historical behavior correlation coefficient, determine the comprehensive correlation strength between feature parameters i and j: In the formula, The comprehensive correlation strength between feature parameter i and feature parameter j within the k-th time window. The dynamic spatiotemporal correlation weight coefficient for the k-th time window. The spatiotemporal correlation coefficient between feature parameter i and feature parameter j within the k-th time window. The correlation coefficient between feature parameter i and feature parameter j is the historical behavior correlation coefficient. The preset time decay factor, This is the window number of the most recent major anomaly event; if no major anomaly event occurred, then... Among them, the dynamic spatiotemporal correlation weight coefficient The value range is 0.6-0.8, for example, Time decay factor The value range is 0.05-0.2, for example, Major abnormal events include large-scale road closures, extreme weather, and large-scale vehicle breakdowns.

[0022] Methods for generating the corresponding implicit logistics state vectors include: S205, constructing an asymmetric correlation strength matrix based on the comprehensive correlation strength among all feature parameters. In the formula, d represents the total number of characteristic parameters. This reflects the unidirectional causal relationship between characteristic parameters. For example, traffic congestion can cause delivery delays, but delivery delays do not cause traffic congestion. and S206. Based on the historical abnormal event database, determine the contribution of each characteristic parameter to various logistics risks. Construct the feature weight matrix The diagonal elements of the matrix represent the risk contribution of each characteristic parameter. Specifically, the changes in each characteristic parameter before each historical logistics risk event are statistically analyzed, the correlation coefficient between each characteristic parameter and the risk event is calculated, and the normalized correlation coefficient is used as the risk contribution of that characteristic parameter; S207, regarding the correlation strength matrix... After weighting, the weighted correlation strength matrix is ​​obtained: S208, Weighted Correlation Strength Matrix Perform principal component analysis to obtain the eigenvalues ​​and eigenvectors of the weighted correlation strength matrix; S209, preset cumulative variance contribution rate threshold. and risk explanatory power threshold Determine the number of principal components K, i.e., select those whose cumulative variance contribution rate is greater than or equal to... And the cumulative risk explanatory power is greater than or equal to The minimum K value. Among them, the cumulative variance contribution rate threshold. The value range is 0.8-0.9, for example, Risk Explanation Threshold The value range is 0.75-0.9, for example, For example, the minimum number of principal components K satisfying the two threshold conditions is calculated to be 4; S210, extract the feature vectors corresponding to the first K principal components and construct a K-row, d-column feature vector matrix. Each row of the matrix corresponds to an eigenvector of a principal component; S211. Based on the eigenvector matrix and the average logistics state eigenvector of the current time window, determine the score of each principal component: In the formula, Let m be the score of the m-th principal component in the k-th time window, where m = 1, 2, ..., K. Let m be the eigenvector corresponding to the m-th principal component. S212: Based on the scores of all principal components, generate a latent logistics state vector of dimension K. S213. Establish a principal component-risk type mapping table, mapping each principal component to one or more specific logistics risk types. For example, the first principal component corresponds to global system risk, the second principal component to path congestion risk, the third principal component to capacity imbalance risk, and the fourth principal component to delivery delay risk. S214. For each principal component m, set three levels of risk thresholds: low risk threshold... Medium risk threshold and high risk threshold ,in For example, , , S215. Compare the scores of each principal component in the implicit logistics state vector with the corresponding risk thresholds to determine the risk level corresponding to each principal component; S216. Obtain the prediction accuracy of each principal component for similar risks in historical data to determine the confidence level of the corresponding risk identification. Specifically, the number of times the corresponding risk actually occurred when the score of a principal component exceeded the medium-risk threshold is statistically analyzed, and the proportion of this number to the total number of occurrences is used as the prediction accuracy, i.e., the confidence level, of that principal component. For example, the preset confidence threshold is 0.8. When the score of any principal component exceeds the corresponding medium-risk threshold and the confidence level is greater than or equal to the preset confidence threshold, a potential logistics risk of the corresponding type is identified. S217: Based on the risk level and confidence level of each principal component, a logistics risk feature fingerprint is generated, containing K elements. The m-th element is the product of the score of the m-th principal component in the k-th time window and the corresponding risk identification confidence level, i.e., .

[0023] Step S3: Establish a logistics risk association network, and determine the risk propagation weight between each risk node based on the association strength and state propagation relationship between risk nodes, so as to obtain the logistics risk propagation results.

[0024] Specifically, step S3 includes: S301, based on the various potential risks identified by the implicit logistics state vector, defining a set of risk nodes in the logistics risk association network, containing N risk nodes, each risk node... For each specific potential risk type, p = 1, 2, ..., N. For example, N = 4, and the risk node set R = {r1, r2, r3, r4}, where r1 corresponds to global system risk, r2 corresponds to path congestion risk, r3 corresponds to capacity imbalance risk, and r4 corresponds to delivery delay risk; S302, based on the comprehensive correlation strength between the characteristic parameters corresponding to each risk node, construct an N-row N-column risk association network adjacency matrix: In the formula, Risk Node and The weight of the associated edges between them, when When it exceeds the preset association threshold, ,otherwise The preset association threshold ranges from 0.4 to 0.6; for example, the preset association threshold is 0.5. S303: Based on the values ​​of each principal component in the implicit logistics state vector, determine the initial risk value for each risk node. In the formula, To be with risk nodes The score of the corresponding m-th principal component in the k-th time window. The risk node weights are preset. For example, (Global system risk) (Route congestion risk) (Risk of capacity imbalance) (Delivery delay risk); S304, Obtain the state propagation relationship between risk nodes and determine the risk propagation direction matrix: In the formula, Indicates risk from nodes propagation to nodes , This indicates no propagation relationship. For example, propagation relationships are: global system risk → path congestion risk, capacity imbalance risk, delivery delay risk; path congestion risk → delivery delay risk; capacity imbalance risk → delivery delay risk. Therefore, the propagation direction matrix D is:

[0025] S305. Based on the adjacency matrix and propagation direction matrix of the risk association network, determine the risk nodes. arrive Risk propagation weight:

[0026] In the formula, This is a preset propagation probability coefficient. The value of the propagation probability coefficient ranges from 0.5 to 0.9. For example, , , , , ,the remaining S306. Construct an iterative formula for risk propagation to calculate the risk diffusion process in the network:

[0027] In the formula, Risk node after the t-th iteration The risk value, Risk nodes after the (t-1)th iteration The risk value, Risk nodes after the (t-1)th iteration The risk value, Risk from nodes propagation to nodes The propagation weights are determined; S307, iterative calculations are performed until the change in risk value at all risk nodes is less than the preset convergence threshold, thus obtaining the final logistics risk propagation result. The preset convergence threshold ranges from [value range missing]. - For example, the preset convergence threshold is In this embodiment, convergence is typically achieved in no more than 15 iterations.

[0028] Step S4: Obtain the real-time status of logistics resources and construct a collaborative resource pool based on the results of logistics risk propagation. Determine the dynamic adaptability of each logistics resource to the logistics task based on the correlation between the status of logistics resources and the requirements of logistics tasks, so as to generate logistics resource matching relationships.

[0029] Specifically, step S4 includes: S401, based on the logistics risk propagation results and real-time logistics resource status data, selecting logistics resources that are in normal and available status, and constructing an initial collaborative resource pool. Specifically, resources with vehicle status of maintenance or out of service are excluded, resources located in high-risk areas (risk value greater than 0.8) are excluded, and resources with insufficient remaining power to complete basic delivery tasks are excluded; S402, based on the real-time status data of each available logistics resource a, constructing a resource feature vector of dimension L. The l-th element in the vector represents the value of the l-th resource feature parameter. For example, L=6, and the resource feature vector... The elements are as follows: remaining load capacity, remaining volume, distance from current location to delivery center, vehicle type, remaining battery power, and driver experience level; S403, based on the demand data of each logistics task to be assigned b, construct a task demand feature vector of dimension L. The l-th element in the vector represents the value of the l-th task requirement parameter, where l = 1, 2, ..., L. For example, a task requirement feature vector... The elements are, in order: cargo weight, cargo volume, distance from the delivery address to the distribution center, required vehicle type, timeliness requirements, and cargo type; S404. Based on resource feature vectors and task requirement feature vectors, determine the dynamic adaptability of logistics resources to logistics tasks:

[0030] In the formula, The degree of dynamic adaptation of logistics resource a to logistics task b. Let be the resource feature vector of logistics resource 'a'. Let be the feature vector representing the task requirements of logistics task b. This is a diagonal weight matrix, where the diagonal elements are the weights of each feature parameter. This is the weighted L2 norm. For example, The diagonal elements are [0.25, 0.2, 0.15, 0.15, 0.15, 0.1]. S405. Based on the dynamic adaptability between all resources and tasks, construct an A-row, B-column adaptability matrix, Score. The element in the a-row, b-column matrix represents the dynamic adaptability of logistics resource a to logistics task b, where A is the total number of available logistics resources and B is the total number of logistics tasks to be assigned. S406. Based on the adaptability matrix, use the Hungarian algorithm to solve the optimal resource matching problem, maximizing the total adaptability to obtain the initial logistics resource matching relationship. Verify whether the initial matching relationship satisfies the resource load constraint, volume constraint, and time constraint. Adjust matching relationships that do not meet the constraints to obtain the final logistics resource matching relationship. Specifically, the load constraint is that the total weight of the assigned tasks ≤ the vehicle's rated load; the volume constraint is that the total volume of the assigned tasks ≤ the vehicle's rated volume; and the time constraint is that the estimated delivery time ≤ the task's required delivery time.

[0031] Step S5: Obtain the real-time path status and construct a dynamic path cost model by combining the logistics resource matching relationship and the logistics risk propagation result. By predicting the status change trend of candidate paths within a future preset time window, determine the dynamic path cost value corresponding to each candidate path, so as to determine the target delivery path.

[0032] Specifically, step S5 includes: S501, defining a dynamic path cost function based on logistics resource matching relationships, logistics risk propagation results, and real-time path status data:

[0033] In the formula, C represents the path cost. For the cost of time, For the cost of distance, For the cost of risk, , , The preset cost weighting coefficients, and For example, , , S502. Based on the latitude and longitude coordinates of the starting and ending points, generate a set of candidate delivery routes containing Y candidate routes. For example, Y=5. S503. Based on historical and real-time route status data, use a time series prediction model to predict the road speed, congestion index, and risk value of each candidate route within a preset time window. Specifically, an LSTM time series prediction model is used, which contains two hidden layers, each with 64 neurons, trained using the Adam optimizer with a learning rate of 0.001. Preset route prediction time window. The value range is 30-60 minutes, for example, Minutes; S504. Determine the time cost of candidate routes based on predicted road speeds and path lengths:

[0034] In the formula, y is the candidate path index number. Candidate paths Total length, Candidate paths The predicted road traffic speed at time t, Candidate paths The predicted congestion index at time t (range 0 to 1, where 0 indicates no congestion and 1 indicates complete congestion); S505, determine the distance cost based on the path length of the candidate path and the unit distance transportation cost:

[0035] In the formula, Candidate paths The cost of transportation per unit distance. For example, Yuan / km; S506, based on the logistics risk propagation results and the predicted risk values ​​of candidate routes, determine the risk cost:

[0036] In the formula, For path At time t, the node is at risk. The extent of the impact, Let the final risk value be the p-th risk node; S507, substitute each cost component into the dynamic path cost function to determine each candidate path. Dynamic path value S508. Sort the dynamic path values ​​of all candidate paths in ascending order and select the candidate path with the lowest value as the target delivery path.

[0037] Step S6: Generate a dynamic allocation plan for logistics resources based on the target delivery route and the matching relationship of logistics resources, and allocate the logistics tasks to the corresponding logistics resources.

[0038] Specifically, step S6 includes: S601, assigning a set of tasks to be delivered to each logistics resource based on the matching relationship of logistics resources; S602, determining the task delivery order of each logistics resource using an improved nearest neighbor algorithm based on the target delivery route and the latitude and longitude of the delivery address of each task, and generating an initial delivery sequence; S603, determining the load rate of each logistics resource based on its load capacity and volume parameters, as well as the total weight and total volume of the assigned tasks. Specifically, load rate = (total weight / rated load capacity + total volume / rated volume) / 2; S604, determining the average load rate of all logistics resources. and standard deviation To achieve load balancing:

[0039] S605. When the load balancing score (LB) is greater than the preset balancing threshold, the task allocation for each resource is adjusted, transferring some tasks from resources with excessively high load rates to resources with excessively low load rates until the load balancing score meets the requirements. The preset balancing threshold ranges from 0.15 to 0.25; for example, it is 0.22. Specifically, excessively high load rate refers to a load rate greater than 0.85, and excessively low load rate refers to a load rate less than 0.3. Tasks with shorter distances and lower timeliness requirements are prioritized for adjustment. S606. Verify whether the adjusted delivery plan meets the timeliness requirements of all tasks and the resource endurance requirements. Perform secondary adjustments on any tasks that do not meet the requirements. S607. Generate a final dynamic allocation plan for logistics resources, including resource IDs, task lists, delivery order, target routes, and estimated arrival times. Distribute this final dynamic allocation plan to the corresponding vehicle-mounted terminal devices of the logistics resources via a 4G / 5G wireless communication network.

[0040] Step S7: Dynamically adjust the steps.

[0041] Specifically, step S7 includes: S701, acquiring state change information during the execution of logistics tasks using the real-time data transmission interface of the logistics resource terminal equipment at a preset sampling frequency f, specifically including the real-time location coordinates of the vehicle, driving speed, number of completed tasks, number of remaining tasks, and temperature and humidity status of the goods. The preset sampling frequency f ranges from 0.5 to 2Hz; for example, f = 1Hz. S702, extracting corresponding real-time feature parameters based on the real-time state change information at the current time t, and constructing a real-time state feature vector of dimension d. The i-th element in the vector represents the value of the i-th real-time feature parameter, where i = 1, 2, ..., d; S703, Obtain the current implicit logistics state vector. The corresponding predicted state feature vector is calculated using the following formula. :

[0042] In the formula, The feature vector matrix with K rows and d columns constructed in step S210; S704, determine the real-time state feature vector according to the following formula. With the predicted state feature vector The degree of relative deviation between :

[0043] In the formula, For L2 norm; S705, the degree of relative deviation Deviation threshold from preset The comparison is performed, and the system also checks for pre-set abnormal events such as vehicle malfunctions, road closures, and customer relocations. When a preset abnormal event is detected, it is determined that the dynamic adjustment conditions are met. When the dynamic adjustment conditions are met, the old data at the corresponding moment in the logistics status feature vector set is replaced with real-time status change information to obtain an updated logistics status feature vector set. The preset deviation threshold ranges from 0.15 to 0.25. For example... S706. Based on the updated set of logistics status feature vectors, re-execute steps S2 to S6 to obtain the updated dynamic allocation scheme for logistics resources, and send it to the terminal devices of the corresponding logistics resources.

[0044] This embodiment collects multi-source heterogeneous logistics data and performs standardized preprocessing to construct a feature vector set that comprehensively reflects the logistics operation status. It employs a combination of spatiotemporal correlation analysis and historical behavior correlation analysis to accurately quantify the causal relationships between feature parameters. Weighted principal component analysis is used to uncover the implicit operational status of the logistics system, enabling early identification and quantitative characterization of potential risks. A logistics risk correlation network is established to simulate the propagation and diffusion process of risks within the system, accurately predicting the scope and intensity of risk impact. Based on the risk propagation results, a collaborative resource pool is constructed, and optimal matching of logistics resources and tasks is achieved through weighted cosine similarity calculation. A dynamic path planning model integrating time, distance, and risk costs is constructed to generate the globally optimal delivery path. Furthermore, a real-time status monitoring and dynamic adjustment mechanism enables dynamic optimization of logistics resource allocation. This invention effectively improves logistics delivery timeliness, reduces vehicle empty load rates, decreases order delay rates, increases response speed to abnormal events, and significantly improves the overall operational efficiency and service quality of the logistics system.

[0045] Example 2

[0046] This embodiment provides an intelligent logistics dynamic allocation system based on big data analysis, used to execute the allocation method as described in Embodiment 1, such as... Figure 2 As shown, the system includes: a data acquisition and preprocessing unit, used to collect and preprocess various types of data in the logistics system to construct a set of logistics status feature vectors; a correlation analysis unit, used to perform correlation analysis on the logistics status feature vectors, generate implicit logistics status vectors, and identify potential risks; a risk propagation determination unit, used to establish a logistics risk correlation network, determine risk propagation weights, and obtain logistics risk propagation results; a resource matching unit, used to construct a collaborative resource pool, determine the dynamic adaptability of resources and tasks, and generate logistics resource matching relationships; a path planning unit, used to construct a dynamic path cost model, determine the dynamic cost value of each candidate path, and determine the target delivery path; an allocation and execution unit, used to generate a dynamic allocation scheme for logistics resources and allocate tasks to corresponding logistics resources; and a dynamic adjustment unit, used to obtain real-time information on changes in task execution status, determine whether dynamic adjustment conditions are met, and update the allocation scheme.

[0047] In this embodiment, the units communicate with each other via Ethernet or a wireless local area network. The data acquisition and preprocessing unit, the correlation analysis unit, the risk propagation determination unit, the resource matching unit, the route planning unit, and the allocation execution unit are deployed on a cloud server. The dynamic adjustment unit is deployed on both the cloud server and the vehicle-mounted terminal. The vehicle-mounted terminal is deployed on the transport vehicle and includes a positioning module, a communication module, a sensor module, and a local storage module.

[0048] Example 3

[0049] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. The memories store computer-readable code, which, when executed by the one or more processors, can perform the above-described intelligent logistics dynamic allocation method based on big data analysis.

[0050] The method according to the embodiments of this application can also be implemented using the architecture of the electronic device shown in this application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. The storage device in the electronic device, such as a ROM or hard disk, may store the intelligent logistics dynamic allocation method based on big data analysis provided in this application. Furthermore, the electronic device may also include a user interface. Of course, the architecture shown in this application is merely exemplary; when implementing different devices, one or more components in the electronic device shown in this application may be omitted according to actual needs.

[0051] Example 4

[0052] One embodiment of this application discloses a computer-readable storage medium. The computer-readable storage medium stores computer-readable instructions. When the computer-readable instructions are executed by a processor, a smart logistics dynamic allocation method based on big data analysis, as described in the above-described embodiments of this application, can be performed. The storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0053] Furthermore, according to embodiments of this application, the processes described in the above-mentioned flowcharts can be implemented as computer software programs. For example, this application provides a non-transitory machine-readable storage medium storing machine-readable instructions that can be executed by a processor to perform instructions corresponding to the method steps provided in this application, such as a smart logistics dynamic allocation method based on big data analysis. When this computer program is executed by a central processing unit (CPU), it performs the functions defined in the method of this application.

[0054] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0055] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0056] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for intelligent logistics dynamic allocation based on big data analysis, characterized in that, Includes the following steps: S1. Collect multi-source data from the logistics system and preprocess it to extract feature parameters that reflect the logistics operation status in order to construct a set of logistics status feature vectors. The multi-source data includes order data, transportation resource data, traffic environment data and historical delivery data. S2. Obtain the changing trend, spatiotemporal correlation, and historical behavior correlation of the feature parameters within a preset time window and perform correlation analysis to determine the correlation strength between each feature parameter in order to generate a corresponding implicit logistics state vector. The implicit logistics state vector is used to characterize the potential operating state of the logistics system and to determine the potential delivery delay risk, potential route congestion risk, and potential capacity imbalance risk based on the changes in correlation strength. S3. Establish a logistics risk association network, and determine the risk propagation weight between each risk node based on the association strength and state propagation relationship between risk nodes, so as to obtain the logistics risk propagation results. S4. Obtain the real-time status of logistics resources and construct a collaborative resource pool based on the results of logistics risk propagation. Determine the dynamic adaptability of each logistics resource to the logistics task based on the correlation between the status of logistics resources and the requirements of logistics tasks, so as to generate logistics resource matching relationships. S5. Obtain the real-time path status and construct a dynamic path cost model by combining the logistics resource matching relationship and the logistics risk propagation result. By predicting the status change trend of candidate paths within a future preset time window, determine the dynamic path cost value corresponding to each candidate path in order to determine the target delivery path. S6. Generate a dynamic allocation plan for logistics resources based on the target delivery route and the matching relationship of logistics resources, and allocate logistics tasks to the corresponding logistics resources.

2. The intelligent logistics dynamic allocation method based on big data analysis according to claim 1, characterized in that, The process of acquiring the changing trends, spatiotemporal correlations, and historical behavioral correlations of feature parameters within a preset time window, and performing correlation analysis to determine the correlation strength between each feature parameter, specifically includes: S201. Based on the set of logistics status feature vectors, determine the window length and sliding step of the preset sliding time window, and generate a continuous time window sequence; S202. Based on the sequence of logistics status feature vectors within each time window, obtain the spatiotemporal correlation coefficient between any two feature parameters i and j; S203. Obtain the historical delivery database to obtain the historical behavior correlation coefficient between any two feature parameters i and j; S204. Determine the comprehensive correlation strength between feature parameters i and j based on the spatiotemporal correlation coefficient and the historical behavior correlation coefficient: In the formula, The comprehensive correlation strength between feature parameter i and feature parameter j within the k-th time window. The dynamic spatiotemporal correlation weight coefficient for the k-th time window. The spatiotemporal correlation coefficient between feature parameter i and feature parameter j within the k-th time window. The correlation coefficient between feature parameter i and feature parameter j is the historical behavior correlation coefficient. The preset time decay factor, This is the window number of the most recent major anomaly event; if no major anomaly event occurred, then... .

3. The intelligent logistics dynamic allocation method based on big data analysis according to claim 2, characterized in that, The generation of the corresponding implicit logistics state vector specifically includes: S205. Based on the comprehensive correlation strength among all feature parameters, construct an asymmetric correlation strength matrix: ； In the formula, d represents the total number of characteristic parameters. To reflect the unidirectional causal relationship between characteristic parameters; S206. Based on the historical abnormal event database, determine the contribution of each characteristic parameter to various logistics risks. Construct the feature weight matrix The elements on the diagonal of the matrix represent the risk contribution of each feature parameter. S207, Correlation Strength Matrix After weighting, the weighted correlation strength matrix is ​​obtained: ； S208. Perform principal component analysis on the weighted correlation strength matrix to obtain the eigenvalues ​​and eigenvectors of the weighted correlation strength matrix; S209, Preset cumulative variance contribution rate threshold and risk explanatory power threshold Determine the number of principal components K, i.e., select those whose cumulative variance contribution rate is greater than or equal to... And the cumulative risk explanatory power is greater than or equal to The minimum value of K; S210. Extract the feature vectors corresponding to the first K principal components and construct a K-row, d-column feature vector matrix, where each row of the matrix corresponds to the feature vector of a principal component. S211. Based on the eigenvector matrix and the average logistics status eigenvector of the current time window, determine the score of each principal component: ； In the formula, Let m be the score of the m-th principal component in the k-th time window, where m = 1, 2, ..., K. Let m be the eigenvector corresponding to the m-th principal component. It is the average value of all logistics status feature vectors within the k-th time window; S212. Based on the scores of all principal components, generate a latent logistics state vector of dimension K. S213. Establish a principal component-risk type mapping table, and map each principal component to one or more specific logistics risk types. The first principal component corresponds to global system risk, the second principal component corresponds to path congestion risk, the third principal component corresponds to capacity imbalance risk, and the fourth principal component corresponds to delivery delay risk. S214. For each principal component m, set a three-level risk threshold: low risk threshold... Medium risk threshold and high risk threshold ,in ; S215. Compare the scores of each principal component in the implicit logistics state vector with the corresponding risk thresholds to determine the risk level corresponding to each principal component. S216. Obtain the prediction accuracy of each principal component for the same type of risk in historical data, and determine the confidence level of the corresponding risk identification. When the score of any principal component exceeds the corresponding medium risk threshold and the confidence level is greater than or equal to the preset confidence threshold, it is determined that there is a potential logistics risk of the corresponding type. S217. Based on the risk level and confidence level of each principal component, generate a logistics risk feature fingerprint containing K elements, where the m-th element is the product of the score of the m-th principal component in the k-th time window and the corresponding risk identification confidence level.

4. The intelligent logistics dynamic allocation method based on big data analysis according to claim 1, characterized in that, Step S3 specifically includes: S301. Based on the various potential risks identified by the implicit logistics state vector, define a set of risk nodes in the logistics risk association network, containing N risk nodes, each risk node... For each specific type of potential risk, p = 1, 2, ..., N; S302. Based on the comprehensive correlation strength between the feature parameters corresponding to each risk node, construct an N-row, N-column risk association network adjacency matrix: ； In the formula, Risk Node and The weight of the associated edges between them, when When it exceeds the preset association threshold, ,otherwise ; S303. Based on the values ​​of each principal component in the implicit logistics state vector, determine the initial risk value for each risk node: ； in, To be with risk nodes The corresponding number The principal component in the first... The score for each time window, Preset risk node weights; S304. Obtain the state propagation relationship between risk nodes and determine the risk propagation direction matrix: ； in, Indicates risk from nodes propagation to nodes , This indicates that there is no transmission relationship; S305. Based on the adjacency matrix and propagation direction matrix of the risk association network, determine the risk nodes. arrive Risk propagation weight: ； in, This is a preset propagation probability coefficient; S306. Construct an iterative formula for risk propagation to calculate the risk diffusion process in the network: ； In the formula, Risk node after the t-th iteration The risk value, Risk nodes after the (t-1)th iteration The risk value, Risk nodes after the (t-1)th iteration The risk value, Risk from nodes propagation to nodes The propagation weight; S307. Iterate the calculation until the change in risk value of all risk nodes is less than the preset convergence threshold to obtain the final logistics risk propagation result.

5. The intelligent logistics dynamic allocation method based on big data analysis according to claim 1, characterized in that, Step S4 specifically includes: S401. Based on the results of logistics risk propagation and real-time logistics resource status data, select logistics resources that are in normal and available status, and build an initial collaborative resource pool. S402. Based on the real-time status data of each available logistics resource, construct a resource feature vector of dimension L, where the l-th element of the vector is the value of the l-th resource feature parameter. S403. Based on the demand data of each logistics task to be assigned, construct a task demand feature vector of dimension L, where the l-th element of the vector is the value of the l-th task demand parameter, and l=1,2,...,L; S404. Based on resource feature vectors and task requirement feature vectors, determine the dynamic adaptability of logistics resources to logistics tasks: ； In the formula, The degree of dynamic adaptation of logistics resource a to logistics task b. Let be the resource feature vector of logistics resource 'a'. Let be the feature vector representing the task requirements of logistics task b. This is a diagonal weight matrix, where the diagonal elements are the weights of each feature parameter. It is the weighted L2 norm; S405. Based on the dynamic adaptability between all resources and tasks, construct an adaptability matrix with A rows and B columns. The element in the a-th row and b-th column of the matrix is ​​the dynamic adaptability of logistics resource a to logistics task b, where A is the total number of available logistics resources and B is the total number of logistics tasks to be assigned. S406. Based on the fitness matrix, the Hungarian algorithm is used to solve the optimal resource matching problem, maximize the total fitness, obtain the initial logistics resource matching relationship, and verify whether the initial matching relationship meets the resource load constraints, volume constraints, and time constraints. Matching relationships that do not meet the constraints are adjusted to obtain the final logistics resource matching relationship.

6. The intelligent logistics dynamic allocation method based on big data analysis according to claim 1, characterized in that, Step S5 specifically includes: S501. Based on logistics resource matching relationships, logistics risk propagation results, and real-time path status data, define a dynamic path cost function: ； In the formula, Value for the path For the cost of time, For the cost of distance, For the cost of risk, , , The preset cost weighting coefficients, and ; S502. Generate a set of candidate delivery routes containing Y candidate routes based on the latitude and longitude coordinates of the starting point and the ending point. S503. Based on historical and real-time path status data, a time series prediction model is used to predict the road traffic speed, congestion index and risk value of each candidate path within a preset time window in the future. S504. Based on the predicted road speed and path length, determine the time cost of the candidate path: ； in, Assign index numbers to candidate paths. Preset the time window length for future path prediction. Candidate paths Total length, Candidate paths The predicted road traffic speed at time t, Candidate paths The predicted congestion index at time t; S505. Determine the distance cost based on the path length and unit distance transportation cost of the candidate paths: ； In the formula, Candidate paths The cost of transportation per unit distance; S506. Based on the results of logistics risk propagation and the predicted risk values ​​of candidate routes, determine the risk cost: ； In the formula, For path At time t, the node is at risk. The extent of the impact, Let be the final risk value of the p-th risk node; S507. Substitute each cost component into the dynamic path cost function to determine each candidate path. Dynamic path value ; S508. Sort the dynamic path cost values ​​of all candidate paths in ascending order and select the candidate path with the lowest cost value as the target delivery path.

7. The intelligent logistics dynamic allocation method based on big data analysis according to claim 1, characterized in that, Step S6 specifically includes: S601. Based on the matching relationship of logistics resources, allocate a set of corresponding delivery tasks to each logistics resource; S602. Based on the target delivery route and the latitude and longitude of the delivery address for each task, determine the task delivery order for each logistics resource and generate an initial delivery sequence; S603. Determine the load rate of each logistics resource based on its load capacity and volume parameters, as well as the total weight and volume of the assigned tasks. S604. Determine the average load factor of all logistics resources. and standard deviation To achieve load balancing: ; S605. When the load balancing degree LB is greater than the preset balancing threshold, the task allocation of each resource is adjusted, and some tasks of the resource with excessive load rate are transferred to the resource with excessive load rate until the load balancing degree meets the requirements. S606. Verify whether the adjusted delivery plan meets the timeliness requirements of all tasks and the resource endurance requirements, and make secondary adjustments to the parts that do not meet the requirements. S607. Generate a final dynamic allocation scheme for logistics resources, including resource ID, task list, delivery order, target route, and estimated arrival time, and send the final dynamic allocation scheme for logistics resources to the terminal devices of the corresponding logistics resources through the wireless communication network.

8. The intelligent logistics dynamic allocation method based on big data analysis according to claim 1, characterized in that, It also includes dynamic adjustment steps: S701. Based on the real-time data transmission interface of the logistics resource terminal equipment, the status change information during the execution of logistics tasks is obtained at a preset sampling frequency f, specifically including the real-time location coordinates of the vehicle, driving speed, number of completed tasks, number of remaining tasks, and temperature and humidity status of the goods. S702. Based on the real-time state change information at the current time t, extract the corresponding real-time feature parameters and construct a real-time state feature vector of dimension d. The i-th element in the vector is the value of the i-th real-time feature parameter, where i = 1, 2, ..., d. S703, Obtain the current implicit logistics state vector The corresponding predicted state feature vector is calculated using the following formula. : ,in, This is the K-row, d-column eigenvector matrix constructed in step S210; S704. Determine the real-time state feature vector according to the following formula. With the predicted state feature vector The degree of relative deviation between : In the formula, It is an L2 norm; S705, Degree of relative deviation Deviation threshold from preset The comparison is performed, and the system also checks for pre-set abnormal events such as vehicle malfunctions, road closures, and customer relocations. when Or, if a preset abnormal event is detected, it is determined that the dynamic adjustment conditions are met; When the dynamic adjustment conditions are met, the old data at the corresponding moment in the logistics status feature vector set is replaced with real-time status change information to obtain the updated logistics status feature vector set. S706. Based on the updated set of logistics status feature vectors, re-execute steps S2 to S6 to obtain the updated dynamic allocation scheme for logistics resources, and send it to the terminal devices of the corresponding logistics resources.

9. A smart logistics dynamic allocation system based on big data analysis, used to execute the allocation method as described in any one of claims 1-8, characterized in that, Specifically, it includes: The data acquisition and preprocessing unit is used to collect various types of data from the logistics system and preprocess them to construct a set of logistics status feature vectors. The correlation analysis unit is used to perform correlation analysis on the logistics status feature vector, generate implicit logistics status vectors, and identify potential risks. The risk propagation determination unit is used to establish a logistics risk association network, determine the risk propagation weights, and obtain the logistics risk propagation results. The resource matching unit is used to build a collaborative resource pool, determine the dynamic adaptability of resources and tasks, and generate logistics resource matching relationships. The route planning unit is used to build a dynamic route cost model, determine the dynamic cost of each candidate route, and determine the target delivery route. The allocation and execution unit is used to generate dynamic allocation schemes for logistics resources and assign tasks to corresponding logistics resources; The dynamic adjustment unit is used to acquire information on changes in task execution status in real time, determine whether the dynamic adjustment conditions are met, and update the allocation scheme.

Images