Intelligent logistics distribution path planning method and system based on big data analysis

Abstract

The invention provides an intelligent logistics distribution path planning method based on big data analysis, relates to the field of logistics distribution, and is used for solving the technical problem that the effect of a logistics distribution path planning scheme is relatively poor in the related technology, and the method comprises the steps: obtaining multi-source heterogeneous data related to a logistics distribution region; based on the multi-source heterogeneous data, obtaining regional state data reflecting each divided region in the logistics distribution region in different time periods; obtaining an expected traffic characteristic index of any path section in the logistics distribution area in any time period, wherein the expected traffic characteristic index is obtained based on the area state data and a state response model; a weighted graph of the path network graph in the logistics distribution area is constructed based on the expected traffic characteristic indexes, and the expected traffic characteristic indexes serve as edge weights of the weighted graph; and on the weighted graph, generating a logistics distribution path with the lowest total edge weight based on the specified starting point and ending point.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of logistics distribution, and in particular to a smart logistics distribution path planning method and system based on big data analysis. BACKGROUND

[0002] Traditional logistics distribution path planning methods usually rely on static road network information and empirical data, such as simple shortest distance or shortest historical average time algorithms. However, the actual logistics distribution environment is extremely complex and dynamic, and factors such as road traffic conditions, weather conditions, unexpected events, and fluctuations in order demand in specific areas can significantly affect the actual traffic characteristics (such as travel time, delay risk, energy consumption) of the path.

[0003] In the prior art, attempts have been made to use real-time traffic data for path planning, such as dynamically adjusting the path according to the current real-time traffic information. Some methods also try to combine historical data for prediction, using statistical models or machine learning models to predict future traffic congestion or travel time. However, these methods often have the following shortcomings:

[0004] Firstly, although multi-source data (such as traffic, weather) are obtained, these heterogeneous data are not effectively fused and structured processed to comprehensively and accurately reflect the comprehensive environmental state of a specific area at a specific time period. The degree of data utilization is not high, and the complex correlations between different types of data are not fully explored.

[0005] Secondly, existing prediction methods or models may lack deterministic output. For the same input conditions, different prediction results may be produced due to the internal mechanism or parameter setting of the model, resulting in a lack of stability in path planning decisions. At the same time, many complex prediction models (such as deep learning models) may have a "black box" problem, with opaque prediction processes that make it difficult to explain the basis for the prediction results and to optimize or correct the model according to actual conditions.

[0006] Furthermore, traditional path planning algorithms (such as Dijkstra) are deterministic, but they rely on the input edge weights. If the generation mechanism of the edge weights is inaccurate or unstable, the final generated path is also difficult to guarantee actual effectiveness. Existing methods still face challenges in converting complex and dynamic environmental factors into quantified indicators required for path planning that are deterministic and can effectively reflect actual traffic characteristics. Especially when facing the superposition of multiple complex factors, it is difficult to accurately predict their comprehensive impact on path traffic characteristics. SUMMARY

[0007] Embodiments of the present application provide a smart logistics distribution path planning method and system based on big data analysis, which is used to improve the technical problem that the actual path passing characteristics cannot be fully reflected by multi-source heterogeneous big data in related technologies, and the effect of logistics distribution path planning scheme is poor.

[0008] To achieve the above object, embodiments of the present application adopt the following technical solutions:

[0009] In a first aspect, the present application provides a smart logistics distribution path planning method based on big data analysis, comprising: acquiring multi-source heterogeneous data related to a logistics distribution area; obtaining regional state data reflecting each divided region in the logistics distribution area at different time periods based on the multi-source heterogeneous data; acquiring an expected passing characteristic index of any path segment in the logistics distribution area at any time period, the expected passing characteristic index being obtained based on the structured processed regional state data and a state response model; constructing a weighted graph of a path network graph in the logistics distribution area based on the expected passing characteristic index, wherein the expected passing characteristic index is used as an edge weight of the weighted graph; and generating a logistics distribution path with the lowest total edge weight based on a specified starting point and a specified ending point on the weighted graph.

[0010] In a possible implementation manner of the first aspect, the multi-source heterogeneous data includes historical traffic data, real-time traffic data, historical weather data, real-time weather data, regional event data, historical order data and geographic information data, and the regional state data includes quantitative data reflecting at least one of a traffic state, a weather state, an event intensity and an order demand intensity of the divided region at the time period.

[0011] In a possible implementation manner of the first aspect, the divided region is a grid unit of a preset size, and the regional state data is obtained by aggregating the multi-source heterogeneous data falling into the same grid unit and at the same time period.

[0012] In a possible implementation manner of the first aspect, the obtaining the expected traffic characteristic indicator of any path segment in the logistics distribution area within any time period comprises: regarding the any path segment as an entity unit in the state response model; regarding factors affecting the expected traffic characteristic of the any path segment as influence sources; determining, based on the structured processed area state data, internal influence parameters in the state response model reflecting influences of inherent attributes of the entity unit on the expected traffic characteristic, and external influence parameters reflecting influences of external environment of the entity unit or states of adjacent entity units on the expected traffic characteristic; and obtaining, based on the state response model, the internal influence parameters and the external influence parameters, an expected intensity of the expected traffic characteristic of the any path segment within the any time period, and taking the expected intensity as the expected traffic characteristic indicator.

[0013] In a possible implementation manner of the first aspect, the expected traffic characteristic indicator represents one of an expected traffic time, an expected delay risk value, an expected energy consumption value or an expected smooth traffic score of the path segment within the time period.

[0014] In a possible implementation manner of the first aspect, the method for generating the logistics distribution path on the weighted graph is one of a Dijkstra algorithm or a Floyd-Warshall algorithm.

[0015] In a possible implementation manner of the first aspect, the method further comprises: after the execution of the logistics distribution path is completed, obtaining actual traffic data of each path segment along the logistics distribution path; based on the actual traffic data and the expected traffic characteristic indicator, obtaining a correction deviation of the expected traffic characteristic indicator of the any path segment within the any time period; and based on the correction deviation, adjusting the internal influence parameter or the external influence parameter affecting the subsequent obtaining of the expected traffic characteristic indicator.

[0016] In a possible implementation manner of the first aspect, the adjusting the internal influence parameter or the external influence parameter affecting the subsequent obtaining of the expected characteristic indicator comprises: based on a type and a size of the correction deviation, obtaining a parameter adjustment amount; and applying the parameter adjustment amount to the internal influence parameter or the external influence parameter to obtain a corrected parameter.

[0017] In a second aspect, the present application provides a smart logistics distribution path planning system based on big data analysis, comprising: a data acquisition module configured to acquire multi-source heterogeneous data related to a logistics distribution area, wherein the multi-source heterogeneous data comprises historical traffic data, real-time traffic data, historical weather data, real-time weather data, regional event data, historical order data and geographic information data; a data processing module configured to clean, standardize and structure the multi-source heterogeneous data to obtain regional state data reflecting the state of each divided region in the logistics distribution area at different time periods; a passing feature index acquisition module configured to acquire an expected passing feature index of any path segment in the logistics distribution area at any time period, wherein the expected passing feature index is a unique numerical value, and the index is acquired based on the structured regional state data and a state response model; a weighted graph construction module configured to construct a weighted graph of a path network graph in the logistics distribution area based on the expected passing feature index, wherein the expected passing feature index is used as an edge weight of the weighted graph; and a path generation module configured to generate a logistics distribution path with the lowest total edge weight based on a specified starting point and a specified ending point on the weighted graph.

[0018] In a possible implementation of the second aspect, the data processing module is configured to aggregate the multi-source heterogeneous data falling into a grid cell of a same preset size and being in a same preset time period to obtain the regional state data, wherein the regional state data comprises quantitative data reflecting at least one of a traffic state, a weather state, an event density and an order demand density of the grid cell in the time period.

[0019] In a possible implementation of the second aspect, the passing feature index acquisition module is configured to: regard the any path segment as an entity unit in the state response model; regard factors affecting the expected passing feature of the any path segment as influence sources; determine, based on the structured regional state data, internal influence parameters reflecting the influence of inherent attributes of the entity unit on the expected passing feature in the state response model, and external influence parameters reflecting the influence of external environment of the entity unit or states of adjacent entity units on the expected passing feature; and acquire, based on the state response model, the internal influence parameters and the external influence parameters, an expected intensity of the expected passing feature of the any path segment in the any time period, and take the expected intensity as the expected passing feature index.

[0020] In a possible implementation of the second aspect, the expected passing feature index represents one of an expected passing time, an expected delay risk value, an expected energy consumption value or an expected smooth passing score of the path segment in the time period.

[0021] In a possible implementation of the second aspect, the system further includes: a feedback module, configured to acquire actual traffic data of each path segment along the logistics distribution path after the execution of the logistics distribution path is completed; a deviation acquisition module, configured to acquire a correction deviation of the expected traffic characteristic index of any path segment in any time period based on the actual traffic data and the expected traffic characteristic index; and a parameter adjustment module, configured to adjust the internal influence parameter or the external influence parameter affecting the process of acquiring the expected traffic characteristic index by the traffic characteristic index acquisition module based on the correction deviation.

[0022] In a possible implementation of the second aspect, the parameter adjustment module is configured to: acquire a parameter adjustment amount based on the type and size of the correction deviation; and apply the parameter adjustment amount to the internal influence parameter or the external influence parameter to obtain a corrected parameter, which is used for subsequent path planning. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 A flowchart of a path planning method is provided for some embodiments of the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application.

[0025] Hereinafter, the terms "first", "second", and the like are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", and the like can explicitly or implicitly include one or more features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0026] In addition, in the present application, the orientation terms such as "up", "down", "left", "right", and the like can include but not limited to the orientation defined by the relative position of the components in the drawings. It should be understood that these directional terms can be relative concepts, which are used for relative description and clarification, and can be changed accordingly according to the change of the position of the components in the drawings.

[0027] In the present application, unless otherwise explicitly specified and limited, the term "connection" should be understood broadly, for example, "connection" can be fixed connection, or detachable connection, or integral; can be directly connected, or indirectly connected through intermediate medium. In addition, the term "electrical connection" can be the mode of electrical connection for realizing signal transmission.

[0028] As used herein, "about," "approximately," or "circa" includes the recited value and reference values within an acceptable range of deviation from the particular value, as determined by one of ordinary skill in the art considering the measurement at issue and the error in measurement associated with the particular quantity being measured (i.e., the limitations of the measurement method).

[0029] Embodiments of the present application provide a smart logistics distribution path planning method and system based on big data analysis, which is used to improve the technical problem that in the related art, the actual passing characteristics of the path cannot be fully reflected by the multi-source heterogeneous big data, and the effect of the logistics distribution path planning scheme is poor.

[0030] As shown in Figure 1 The method comprises the following steps:

[0031] S101, acquiring multi-source heterogeneous data related to a logistics distribution area.

[0032] The multi-source heterogeneous data refers to a collection of data from different channels and with different formats, which reflects various dynamic and static information in the logistics distribution area. The data includes but is not limited to historical traffic data (such as average speed, congestion index, accident record of different road sections and different time periods), real-time traffic data (such as current real-time traffic information, traffic flow), historical weather data (such as rainfall, heavy fog, ice and snow records at different times and locations), real-time weather data (such as current weather conditions, forecasts), regional event data (such as large-scale activities, road construction, temporary regulation, emergency information), historical order data (such as order quantity, delivery time, actual delivery time in a specific area and a specific time period), and geographic information data (such as road network topology, road section length, road grade, and point of interest distribution).

[0033] Exemplarily, the above data can be obtained through various ways. Historical data can be obtained from government transportation departments, meteorological bureaus, operators, map service providers, and business systems of logistics companies. Real-time data can be obtained through sensors deployed on roads, traffic cameras, GPS tracking devices, crowdsourcing information platforms, and real-time weather forecast services.

[0034] S102, obtaining regional state data reflecting the state of each divided region in the logistics distribution area at different time periods based on the multi-source heterogeneous data.

[0035] The original multi-source heterogeneous data is pre-processed to convert it into unified format, unified standard and structured data, so as to facilitate the subsequent state response model. The cleaning process includes removing duplicate data, correcting error data, filling missing data, etc. The standardization process includes unifying the timestamp format, coordinate system, unit, etc. The structured processing is to organize different types of data into a form with a clear structure, and to align and aggregate in space and time.

[0036] For example, the entire logistics distribution area can be divided into grid units of a preset size (for example, 1 km x 1 km), and a day can be divided into preset time periods (for example, every 15 minutes or every 30 minutes). For each grid unit and each time period, all relevant data falling into the grid unit and occurring within the time period are aggregated to form the area state data of the grid unit within the time period.

[0037] For example, the area state data of a grid unit within a specific 15-minute time period can include: the average real-time vehicle speed of all road segments within the grid, the traffic congestion level, whether there is road construction or event, the real-time weather of the area (temperature, humidity, whether it rains), the historical average order volume of the area, the order demand intensity of the time period, etc. These aggregated data can be quantified as a multi-dimensional area state vector or structure, reflecting the comprehensive state of the area at that time.

[0038] S103, obtaining the expected traffic characteristic index of any path segment in the logistics distribution area within any time period. For example, the any path segment is regarded as an entity unit in the state response model. In the model, each specific road path segment (for example, a road segment from point A to point B) is abstracted as an independent entity, which is the object of model analysis and prediction.

[0039] The factors affecting the expected traffic characteristics of the any path segment are regarded as influence sources. These influence factors come from the area state data obtained in S102, which can be factors directly acting on the path segment (such as the real-time traffic condition of the path segment), or indirect influence factors (such as congestion overflow of adjacent areas, events in upstream areas, and vehicle flow increase caused by order peak in specific areas). These factors are regarded as “influence sources” that can affect the “state” (i.e. traffic characteristics) of the entity unit.

[0040] Based on the structured area state data, the internal influence parameters reflecting the influence of the inherent properties of the entity unit on the expected traffic characteristics in the state response model are determined, and the external influence parameters reflecting the influence of the external environment or the state of adjacent entity units on the expected traffic characteristics are determined.

[0041] The internal influence parameter is used to reflect how the inherent characteristics of a path segment affect its traffic feature. These characteristics are relatively stable, such as road grade (expressway, national highway, urban trunk road), speed limit, number of lanes, slope, etc. These parameters can be determined based on geographic information data or historical long-term average data, and they represent the inherent, basic traffic capacity or inherent resistance to external influences of a path segment. Illustratively, the road grade can be mapped to the value of the internal influence parameter, for example, the internal influence parameter of an expressway is a lower value (indicating high basic traffic capacity or low inherent resistance), and the internal influence parameter of an urban auxiliary road is a higher value. The value of the internal influence parameter can be a normalized numerical value, for example, between 0 and 1, where 0 represents the best inherent traffic feature and 1 represents the worst inherent traffic feature.

[0042] The external influence parameter is used to reflect how the environmental state of the surrounding of a path segment or the state of adjacent path segments affects the traffic feature of the path segment through external channels. Understandably, these parameters are dynamic and mainly based on the regional state data obtained in step S102. For example, traffic congestion in a certain region can spread through the road network, affecting adjacent or related path segments; a sudden event or bad weather can affect the traffic capacity of a large area; an order explosion in a specific region can cause a large number of delivery vehicles to flow into related paths. The external influence parameter reflects the "sensitivity" or "susceptibility" of the path segment to these external influences. Illustratively, the geographic location of the path segment, the proximity to major hubs or congestion points can be mapped to the value of the external influence parameter. Path segments close to major commercial areas or traffic hubs can have a higher value of the external influence parameter (indicating susceptibility to external environmental influences), and path segments far from the city center can have a lower value. The external influence parameter can also be a normalized numerical value, for example, between 0 and 1.

[0043] Based on the state response model, the internal influence parameter and the external influence parameter, the expected intensity of the expected traffic feature of the any path segment in the any time period is obtained, and the expected intensity is taken as the expected traffic feature indicator. The state response model is used to describe the traffic feature (state) of a path segment (entity unit) being jointly affected by its internal attributes (internal influence parameter) and external environment (external influence parameter and influence source).

[0044] Illustratively, the state response model can be:

[0045] E_i(t) = C_int * Ip_i + C_ind * Ep_i * S_comp(SD(i), t)

[0046] where Ip_i is the internal influence parameter value for path segment i, Ep_i is the external influence parameter value for path segment i, and SD(i) is the structured area state data for the area (or its influence area) where path segment i is located at time period t.

[0047] S_comp(SD(i),t) (denoted as Sc) is a function that converts the area state data SD(i) into a single, normalized "environmental influence comprehensive score" or "external pressure factor", for example between 0 and 1. Exemplarily, this function can be a weighted sum of certain key components of the area state data vector (such as the normalized congestion index, weather severity score, number of events per unit area, order density per unit area), or a more complex mapping function.

[0048] C_int and C_ind are coefficients or weights of the model, which reflect the relative importance of internal influence and external influence in determining the final expected traffic feature intensity. These coefficients can be trained or fitted from historical data, for example by minimizing the error between the model prediction and the historical actual traffic feature.

[0049] E_i(t) is the calculated expected traffic feature intensity, which is a unique numerical value that will serve as the expected traffic feature indicator for path segment i at time period t.

[0050] Suppose we want to calculate the expected traffic feature indicator for path segment A at Friday evening peak 18:00-18:15 time period, which is set as "expected delay risk score" (0-10 points, the higher the score the greater the risk). Path segment A is a city trunk road, according to its road level, its internal influence parameter Ip_A = 0.3 (lower risk basis) is determined. Path segment A is located at the edge of the city center, near an exit of a large residential area, and its external influence parameter Ep_A = 0.7 (easily affected by surrounding activities) is determined.

[0051] Through historical data training, the model coefficients C_int = 5.0 and C_ind = 6.0 are determined.

[0052] Obtain the area state data and calculate the environmental influence comprehensive score:

[0053] Obtain the area state data of the grid where path segment A is located and its adjacent grids at 18:00-18:15 time period.

[0054] Exemplary state data: real-time congestion index (normalized) = 0.8, weather severity score (such as light rain) = 0.4, number of events per unit area = 1 (with small construction), order density per unit area = 0.6 (order peak).

[0055] Suppose the environmental impact composite score function S_comp is a weighted sum of these normalized indicators: Sc = 0.5 * congestion index + 0.3 * weather severity + 0.1 * event number + 0.1 * order density.

[0056] Then Sc = 0.5 * 0.8 + 0.3 * 0.4 + 0.1 * 1 + 0.1 * 0.6 = 0.4 + 0.12 + 0.1 + 0.06 = 0.68.

[0057] The expected delay risk score is calculated using the state response model:

[0058] E_A(t) = C_int * Ip_A + C_ind * Ep_A * Sc

[0059] E_A(t) = 5.0 * 0.3 + 6.0 * 0.7 * 0.68

[0060] E_A(t) = 1.5 + 4.2 * 0.68

[0061] E_A(t) = 1.5 + 2.856

[0062] E_A(t) = 4.356

[0063] The only numerical value 4.356 calculated is the expected delay risk score of route A at Friday evening peak 18:00-18:15. This score (expected intensity) will be the expected travel characteristic indicator of this route segment at this time period.

[0064] The expected travel characteristic indicator represents one of the expected travel time, expected delay risk value, expected energy consumption value, or expected smooth travel score of the route segment at the time period.

[0065] It is understood that this means the indicator can be one of various specific measures reflecting travel efficiency or impedance. Which indicator to choose depends on specific business needs, for example, pursuing time efficiency can take expected travel time as the indicator, pursuing stability can take expected delay risk value as the indicator. Map the expected intensity (e.g. delay risk score 4.356) calculated here to the final route weight (e.g. expected travel time). This mapping can be another function or lookup table, for example, adjust the base travel time of this route segment (calculated based on length and speed limit) according to the delay risk score to get the final expected travel time as the edge weight.

[0066] S104, construct a weighted graph of the path network graph in the logistics distribution area based on the expected travel characteristic indicator.

[0067] Based on the expected traffic characteristic indicators of each path segment in a specific time period, a weighted graph for path search is constructed, taking the expected traffic characteristic indicators as the edge weights of the graph. The path network graph in the logistics distribution area is a graph structure composed of nodes (such as road intersections, distribution points, service points) and edges (path segments connecting nodes).

[0068] Exemplarily, the road network in the geographic information data can be converted into a graph structure. The nodes of the graph correspond to road intersections or important geographic location points, and the edges of the graph correspond to road segments connecting these nodes. For each edge (representing a path segment) in the graph, the expected traffic characteristic indicator (for example, the expected travel time mapped according to the delay risk score) obtained in a specific time period is assigned as the weight of the edge in the time period. Therefore, the entire path network graph is converted into a weighted graph, in which the weights of the edges dynamically reflect the expected travel efficiency or impedance of the path segment in the time period.

[0069] S105, based on the specified starting point and the ending point, a logistics distribution path with the lowest total edge weight is generated on the weighted graph.

[0070] This step uses a graph search algorithm to find the path with the lowest total edge weight from the specified distribution starting point to the ending point (for example, from a warehouse to a customer address, or from one distribution point to the next distribution point) on the weighted graph constructed in S104. Since the edge weights are generated based on the expected traffic characteristic indicators, the path with the lowest total edge weight is the path with the optimal expected comprehensive travel efficiency or the lowest total impedance.

[0071] Exemplarily, the method can use mature shortest path algorithms such as Dijkstra algorithm or Floyd-Warshall algorithm to generate the path. Dijkstra algorithm is suitable for single-source shortest path problem and can quickly find the shortest path from one distribution point to all other reachable points. Floyd-Warshall algorithm is suitable for all-source shortest path problem and can calculate the shortest path between any two points in the network. These algorithms are deterministic, and for a given weighted graph, they will output a unique path with the lowest total edge weight.

[0072] In some embodiments, the method further comprises:

[0073] After the execution of the logistics distribution path is completed, the actual travel data of each path segment along the logistics distribution path is obtained. The system records or collects the actual travel data of the distribution vehicle when driving along the planned path, such as the actual time consumption of each path segment, the actual encountered congestion level, the actual occurred delay events, the actual energy consumption, etc.

[0074] Based on the actual traffic data and the expected traffic characteristic indicators, a correction bias of the expected traffic characteristic indicators of the any path segment in any time period is obtained. For each path segment on the path, the expected traffic characteristic indicators (e.g., expected travel time or other indicators mapped from delay risk score) predicted at the time of planning are compared with the actual traffic data (e.g., actual travel time) at the time of actual execution, and a bias between the two is calculated. The bias can be the actual value minus the predicted value, or the ratio of the actual value to the predicted value, etc.

[0075] Based on the correction bias, the internal influence parameters or the external influence parameters affecting the subsequent process of obtaining the expected traffic characteristic indicators are adjusted. According to the calculated correction bias, the relevant parameters in the state response model are adjusted.

[0076] Exemplarily, the adjustment process includes:

[0077] Based on the type and size of the correction bias, a parameter adjustment amount is obtained. The type (e.g., whether the actual is greater than or less than the predicted) and size of the bias reflect the degree of inaccuracy of the model's prediction in a particular situation. A set of rules can be preset or a function can be used to map the bias to a specific parameter adjustment amount. For example, if the actual travel time of a path segment in a time period is always significantly slower than expected (positive bias), a positive parameter adjustment amount is calculated. If it is always significantly faster than expected (negative bias), a negative parameter adjustment amount is calculated. The size of the adjustment amount can be positively correlated with the size of the bias.

[0078] The parameter adjustment amount is applied to the internal influence parameters or the external influence parameters to obtain the corrected parameters, which are used for subsequent path planning. The calculated parameter adjustment amount is added to or multiplied by one or more parameters in the state response model. For example, if it is found that the actual delay risk of a path segment in a time period is always higher than the model's prediction, and analysis shows that this is more caused by external environmental changes, the feedback mechanism can increase the value of the external influence parameter (Ep) corresponding to the path segment in the time period. Or, if the model overestimates the traffic capacity of a certain type of road (related to internal attributes), the value of the internal influence parameter (Ip) related to the type of road can be adjusted. For example, the corrected parameter = original parameter + adjustment factor * bias. The corrected parameters can be stored and used by the state response model in future path planning under similar conditions, so that the model's prediction results can be continuously optimized as the actual situation changes.

[0079] The application also provides a smart logistics distribution path planning system based on big data analysis, which is used to realize the above method. The system can be deployed on a server, a cloud computing platform or other computing devices, and interacts with data sources, terminal devices of distribution vehicles, order management systems, etc. through a network.

[0080] The system can include a data acquisition module, a data processing module, a traffic feature index acquisition module, a weighted graph construction module, and a path generation module.

[0081] The data acquisition module is used to acquire multi-source heterogeneous historical data and / or real-time data related to the logistics distribution area. The data acquisition module can be configured with different interfaces and connectors for real-time or periodic data pulling from various data sources such as traffic management systems, weather service platforms, geographic information databases, order and vehicle tracking systems within logistics companies, and third-party data service providers. The data acquisition module ensures that the types and ranges of data acquired meet the needs of subsequent processing.

[0082] Illustratively, the data acquisition module can include a timing capture program, a real-time data stream listener, an API calling client, etc. to adapt to the characteristics of different data sources. For example, real-time congestion indexes of major road segments can be acquired from a traffic information platform every minute, regional weather forecasts can be acquired from a meteorological bureau every ten minutes, and GPS position and speed data from the terminal of a distribution vehicle can be continuously received.

[0083] The data processing module is used to clean, standardize and structure the multi-source heterogeneous data to obtain regional state data reflecting the state of each divided region in the logistics distribution area at different time periods. The module receives raw data from the data acquisition module and performs a series of preprocessing operations.

[0084] Illustratively, the data processing module can include a data cleaning submodule, a data standardization submodule, and a data structuring submodule. The cleaning submodule is responsible for identifying and processing invalid, abnormal or missing data. The standardization submodule is responsible for unifying data formats, timestamps, coordinate systems and units. The structuring submodule is responsible for fusing data of different sources and different formats, and aggregating according to the pre-set spatial division (such as grid) and time division (such as time period).

[0085] The data processing module is configured to aggregate the multi-source heterogeneous data falling into the same grid unit of a preset size and being in the same preset time period to obtain the regional state data, which includes quantified data reflecting at least one of traffic state, weather state, event density, and order demand density of the grid unit in the time period. This means that the module groups data according to grid and time period, and performs statistics or calculation (such as average, summation, counting, etc.) on each group of data to form a regional state data vector corresponding to each grid in each time period. This vector contains the key environmental and business quantified information of the region in the period.

[0086] The passage feature index acquisition module is configured to obtain the expected passage feature index of any path segment in the logistics distribution region in any time period. The module receives the regional state data from the data processing module, and uses the state response model to calculate the expected passage feature index of each path segment. The passage feature index acquisition module is configured to regard the any path segment as an entity unit in the state response model. The module internally maintains a list of all path segments in the logistics network and maps them as objects for model processing.

[0087] The factors affecting the expected passage feature of the any path segment are regarded as influence sources. The module extracts the external environmental factors related to a specific path segment or capable of affecting the path segment from the regional state data, and inputs them as influence sources of the model.

[0088] Based on the structured processed regional state data, the internal influence parameters reflecting the influence of the inherent properties of the entity unit on the expected passage feature, and the external influence parameters reflecting the influence of the external environment or the state of adjacent entity units on the expected passage feature are determined in the state response model. The module loads or calculates the internal influence parameters according to the inherent property information of the path segment (such as road grade, speed limit, etc.), and calculates or looks up the corresponding external influence parameters according to the current or predicted regional state data (such as real-time congestion of surrounding grids, weather conditions, event information, etc.).

[0089] Based on the state response model, the internal influence parameters and the external influence parameters, the expected intensity of the expected passage feature of the any path segment in the any time period is obtained, and the expected intensity is taken as the expected passage feature index.

[0090] Exemplarily, the state response model can be implemented as a piece of program code or a separate computing unit. The model receives path segment identification, time period identification, corresponding internal influence parameter value and external influence parameter value as input. The internal influence parameter can be one or a set of numerical values, for example, normalized numerical values related to road grade, capacity. The external influence parameter can be one or a set of numerical values, for example, values obtained by a preset function or a lookup table according to the average congestion index of the surrounding grid, the event intensity, the order demand intensity, etc. The function describes how the expected state (traffic intensity) of the entity unit (path segment) changes with the changes of internal and external influences. For example, the model can output a numerical value representing the expected travel time according to the input parameters, which is unique and determined with the determination of the input parameters.

[0091] The weighted graph construction module is configured to construct a weighted graph of the path network graph in the logistics distribution area based on the expected traffic characteristic indicators. The module receives the topological structure information of the logistics network and the expected traffic characteristic indicators of each path segment from the traffic characteristic indicator acquisition module.

[0092] Exemplarily, the module represents the logistics network as a graph data structure, in which the nodes represent the intersections and the edges represent the path segments. The module traverses all the path segments and assigns the expected traffic characteristic indicators (such as expected travel time) corresponding to a specific time period (in which the distribution needs to be planned) to the weight of the edge in that time period, thereby constructing a time-dependent, dynamically weighted path network graph.

[0093] The path generation module is configured to generate a logistics distribution path with the lowest total edge weight based on the specified starting point and ending point on the weighted graph. The module receives the weighted graph from the weighted graph construction module and the specified starting point and ending point of the distribution task.

[0094] Exemplarily, the module internally implements a shortest path finding algorithm such as Dijkstra's algorithm or Floyd-Warshall algorithm. According to the specified starting point and ending point, the module performs algorithm calculation on the weighted graph to find the path connecting the two points with the lowest sum of weights of all passing path segments.

[0095] In some embodiments, the system can further include a feedback module, a bias acquisition module and a parameter adjustment module.

[0096] The feedback module is configured to acquire actual traffic data of each path segment along the logistics distribution path after the execution of the logistics distribution path. The module receives actual operation data from the distribution vehicle terminal or other monitoring systems, such as the GPS trajectory of the vehicle, the timestamp of arrival at each intersection, the report of actual traffic conditions encountered, etc.

[0097] The deviation obtaining module is configured to obtain a correction deviation of the expected traffic characteristic indicator of the path segment in the time period based on the actual traffic data and the expected traffic characteristic indicator. The module compares the actual traffic data obtained by the feedback module with the expected indicator generated by the traffic characteristic indicator obtaining module during path planning.

[0098] Exemplarily, for each path segment on the path, if the expected indicator is "expected travel time", the module compares the actual travel time with the expected travel time, and calculates the difference as the correction deviation. For example, deviation = actual travel time - expected travel time.

[0099] The parameter adjusting module is configured to adjust the internal influence parameter or the external influence parameter affecting the process of obtaining the expected traffic characteristic indicator by the traffic characteristic indicator obtaining module based on the correction deviation. The module receives the correction deviation from the deviation obtaining module.

[0100] The module is configured to obtain a parameter adjustment amount based on the type and size of the correction deviation, and apply the parameter adjustment amount to the internal influence parameter or the external influence parameter to obtain a corrected parameter, which is used for subsequent path planning.

[0101] Exemplarily, if the correction deviation of a path segment in a time period is positive and large (the actual time is much longer than the expected time), it indicates that the prediction of the model is low. The module can calculate a positive parameter adjustment amount according to the size of the deviation according to a preset ratio or lookup table. Then, the adjustment amount is added to or multiplied by the external influence parameter corresponding to the path segment in the time period (for example, the external influence parameter can reflect the strength of environmental resistance, and increasing the parameter value will make the model predict greater resistance), to obtain a corrected external influence parameter. These corrected parameters will be stored and used by the traffic characteristic indicator obtaining module when path planning under similar conditions in the future, so that the model can better predict the actual traffic condition. The adjustment process is deterministic, that is, the same deviation input will produce the same parameter adjustment amount and corrected parameter.

[0102] From the above description of the embodiments, those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional modules is exemplified, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

[0103] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the division of the apparatus embodiments is only a logical function division, and there can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another apparatus, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different apparatuses can be indirect couplings or communication connections through some interfaces, apparatuses or units, and can be in electrical, mechanical or other forms.

[0104] The units described as separated components can or can not be physically separated, and the components displayed as units can be one physical unit or multiple physical units, i.e., can be located in one place or distributed to multiple different places. According to the actual needs, some or all of the units can be selected to implement the purposes of the embodiments of the present application.

[0105] In addition, each functional unit in the embodiments of the present application can be integrated in a processing unit, or each unit can exist physically as a separate unit, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware.

[0106] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any changes or replacements within the technical scope disclosed in the present application should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A smart logistics distribution route planning method based on big data analysis, characterized in that: include: Acquire multi-source heterogeneous data related to logistics distribution areas; Based on the multi-source heterogeneous data, regional status data reflecting each divided area in the logistics distribution area in different time periods is obtained; Obtaining an expected traffic characteristic index for any path segment within the logistics distribution area within any time period, wherein the expected traffic characteristic index is obtained based on the area state data and a state response model; constructing a weighted graph of a path network graph within the logistics distribution area based on the expected traffic characteristic index, wherein the expected traffic characteristic index serves as an edge weight of the weighted graph; and On the weighted graph, a logistics distribution path with the lowest total edge weight is generated based on the specified starting point and ending point.

2. The method according to claim 1, characterized in that The multi-source heterogeneous data includes historical traffic data, real-time traffic data, historical weather data, real-time weather data, regional event data, historical order data and geographic information data, and the regional status data includes quantitative data reflecting at least one of the traffic status, weather status, event density and order demand density of the divided area during the time period.

3. The method according to claim 2, characterized in that The divided areas are grid units of a preset size, and the area status data is obtained by aggregating the multi-source heterogeneous data that fall into the same grid unit and are within the same time period.

4. The method according to claim 1, wherein The obtaining of the expected traffic characteristic index of any path segment within the logistics distribution area within any time period includes: Taking any path segment as a physical unit in the state response model; Taking factors that affect the expected traffic characteristics of any path segment as influencing sources; Based on the regional state data, determining, in the state response model, internal influencing parameters reflecting the influence of the inherent properties of the entity unit on the expected traffic characteristics, and external influencing parameters reflecting the influence of the external environment of the entity unit or the state of adjacent entity units on the expected traffic characteristics; and Based on the state response model, the internal influencing parameters and the external influencing parameters, an expected intensity of the expected traffic characteristic of any path segment in any time period is obtained, and the expected intensity is used as the expected traffic characteristic indicator.

5. The method according to claim 4, characterized in that Also includes: After the logistics distribution path is completed, actual traffic data of each path segment along the logistics distribution path is obtained; Based on the actual traffic data and the expected traffic characteristic index, obtaining a corrected deviation of the expected traffic characteristic index of any path segment within any time period; and Based on the corrected deviation, the internal influencing parameter or the external influencing parameter affecting the subsequent process of obtaining the expected traffic characteristic index is adjusted.

6. The method according to claim 5, characterized in that The adjusting of the internal influencing parameter or the external influencing parameter affecting the subsequent process of obtaining the expected traffic characteristic index includes: Obtaining a parameter adjustment amount based on the type and size of the correction deviation; and The parameter adjustment amount is applied to the internal influencing parameter or the external influencing parameter to obtain a corrected parameter.

7. The method according to claim 1, characterized in that The expected traffic characteristic index represents one of the expected travel time, expected delay risk value, expected energy consumption value or expected smooth travel score of the path segment within the time period.

8. A smart logistics distribution route planning system based on big data analysis, characterized by: include: Data acquisition module, used to obtain multi-source heterogeneous data related to the logistics distribution area; A data processing module is used to clean, standardize and structure the multi-source heterogeneous data to obtain regional status data reflecting each divided area in the logistics distribution area in different time periods; a traffic characteristic index acquisition module, configured to acquire an expected traffic characteristic index for any path segment within the logistics distribution area within any time period, wherein the expected traffic characteristic index is acquired based on the structured regional state data and a state response model; a weighted graph construction module, configured to construct a weighted graph of a path network graph within the logistics distribution area based on the expected traffic characteristic index, wherein the expected traffic characteristic index serves as an edge weight of the weighted graph; and The path generation module is used to generate a logistics distribution path with the lowest total edge weight based on the specified starting point and end point on the weighted graph.

9. The system according to claim 8, characterized in that The data processing module is configured to aggregate the multi-source heterogeneous data that fall into grid units of the same preset size and are within the same preset time period to obtain the regional status data, where the regional status data includes quantitative data reflecting at least one of the traffic status, weather status, event density, and order demand density of the grid unit within the time period.

10. The system according to claim 9, characterized in that The traffic characteristic index acquisition module is configured as follows: Taking any path segment as a physical unit in the state response model; Factors that affect the expected traffic characteristics of any path segment are considered as influencing sources; Determining, based on the structured regional state data, internal influencing parameters in the state response model that reflect the influence of the inherent properties of the entity unit on the expected traffic characteristics, and external influencing parameters that reflect the influence of the external environment of the entity unit or the state of adjacent entity units on the expected traffic characteristics; and Based on the state response model, the internal influencing parameters and the external influencing parameters, an expected intensity of the expected traffic characteristic of any path segment in any time period is obtained, and the expected intensity is used as the expected traffic characteristic indicator.

Images