Cold chain freight vehicle classification method based on travel task complexity

Abstract

The invention belongs to the technical field of cold-chain highway freight, and discloses a cold-chain freight vehicle classification method based on travel task complexity. The method comprises the following steps: collecting cold chain truck track point data based on a Beidou satellite positioning system; preprocessing the collected cold chain truck track point data; an OD pair is constructed based on the preprocessed cold chain truck track point data; constructing a cold-chain freight vehicle trip chain based on the constructed OD pair; extracting trip chain features based on the constructed cold chain freight vehicle trip chain; obtaining a trip characteristic portrait of the cold chain freight vehicle based on the extracted trip chain characteristics; and constructing a vehicle multi-task scheduling optimization model based on the obtained travel characteristic portrait of the cold-chain freight vehicle. The invention aims to provide support for fine management and operation optimization of cold-chain logistics.

Description

Technical Field

[0001] This invention belongs to the field of cold chain road freight technology, specifically relating to a classification method for cold chain freight vehicles based on the complexity of their travel tasks. Background Technology

[0002] In recent years, with the continuous growth of cold chain freight business, cold chain logistics companies have an increasingly urgent need for optimizing transportation routes and rationally arranging vehicle scheduling; government departments have attached increasing importance to strengthening the supervision of cold chain logistics and ensuring the safety of transported goods; the widespread application of my country's Beidou satellite navigation system, with its advantages in precise positioning, short message service, and domestic independent research and development, has also provided reliable data support for the study of the travel characteristics of cold chain vehicles.

[0003] Currently, traditional cold chain freight research mainly focuses on facility site selection or energy consumption, lacking detailed descriptions of the dynamic travel behavior of vehicles. Most studies rely on GPS data, with relatively little analysis and application of BeiDou data. Much research focuses on general freight or specific types (such as container trucks), neglecting the refined and differentiated analysis of the travel chain characteristics, operational models, and influencing factors of cold chain freight, a niche sector with specific requirements for timeliness and temperature control. Cold chain vehicle travel is highly complex, involving not only diverse cargo types and strict temperature control requirements, but also nonlinear route planning, frequent multi-point deliveries, and irregular time rhythms such as morning and evening rush hours and nighttime travel. Furthermore, its operational behavior is influenced by multiple factors such as weather, urban traffic, cargo nature, and customer demand timeframes, resulting in highly dynamic and variable operating trajectories. These characteristics make the cold chain transportation travel chain highly heterogeneous, and traditional methods based on static OD (Origin-Destination) modeling or single parameters such as average speed are no longer sufficient to accurately identify its operational patterns. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention processes BeiDou data to identify and construct key origin-destination (OD) pairs and travel chains. This enables the extraction of multi-dimensional feature indicators reflecting operational status and efficiency from the travel chains, thus creating a profile of cold chain truck travel and providing scientific basis and decision support for the refined management, operational optimization, and network resilience enhancement of cold chain logistics companies.

[0005] To achieve the above objectives, this invention provides a method for classifying cold chain freight vehicles based on the complexity of travel tasks, the steps of which are as follows: Step S1: Collect trajectory point data of cold chain trucks based on the BeiDou satellite positioning system; preprocess the collected trajectory point data of cold chain trucks; Step S2: Construct OD pairs based on the preprocessed cold chain truck trajectory point data; Step S3: Construct the cold chain freight vehicle travel chain based on the constructed OD pair; Step S4: Extract travel chain features based on the constructed cold chain freight vehicle travel chain; Step S5: Based on the extracted travel chain features, obtain a travel characteristic profile of cold chain freight vehicles. Step S6: Construct a vehicle multi-task scheduling optimization model based on the obtained travel characteristic profile of cold chain freight vehicles.

[0006] Furthermore, in step S1, the collected cold chain truck trajectory point data is preprocessed; a spatiotemporal deduplication algorithm is used for preprocessing; the specific steps are as follows: Step S11: Standardize the time format; Step S12: Based on the BeiDou coding, the collected cold chain truck trajectory point data are grouped, and duplicate data are identified using a set dynamic time window of 5 seconds. Step S13: Based on the temperature inside the compartment, exclude missing data that has not been available for more than 30 consecutive minutes; Step S14: Based on the dual verification mechanism of time window and spatial unique key, duplicate data is removed, and only the first record of each group in the dynamic time window is retained. Then, a triplet of "BeiDou number + time + latitude and longitude" is constructed as a unique index to remove duplicate data. Step S15 introduces a spatial fence mechanism, establishing geographical boundary constraints covering the entire land area based on geographical boundary data; intercepting cross-border data, device location drift data, and contaminated data.

[0007] Furthermore, in step S2, OD pairs are constructed based on the preprocessed cold chain truck trajectory point data; the specific steps are as follows: Step S21: Group the preprocessed cold chain truck trajectory point data according to Beidou coding, and sort them in ascending order by time field within each group; Step S22: Convert latitude and longitude to radian values, calculate the spherical distance between adjacent trajectory points using the semi-sine formula, and calculate the speed in combination with the data collection time interval; when the speed exceeds the legal speed limit threshold, it is judged as abnormal data and removed. At the same time, the vehicle's motion status is divided into five categories according to the speed: stationary, low speed, normal driving, high speed driving, and speeding, and marked accordingly. Step S23: Select stationary and low-speed trajectory points with speeds ≤ 5 km / h as candidate stopping points; Step S24: Use a density-based spatial clustering algorithm with noise to process the data; determine a suitable distance radius threshold and minimum number of neighbor points through the silhouette coefficient to achieve clustering of candidate stopping points, remove isolated points and noise points, and extract well-structured and continuous clusters. Step S25: When there are ≥2 clusters, extract the center point of each cluster as the docking point; Step S26: Select stops with a stop time of ≥10 minutes, and extract the trajectory segments between two adjacent stops to construct OD pairs.

[0008] Furthermore, in step S3, a cold chain freight vehicle travel chain is constructed based on the constructed OD; the specific steps are as follows: Step S31: Using the first OD pair of each vehicle as the starting point of the travel chain, initialize the travel chain object, and record the travel chain ID, vehicle number and the start and end time of the current trip to form a travel chain trip list. Step S32: If the time interval between the current trip and the previous trip meets the continuity requirement and the spatial distance meets the proximity condition, the trip chain is extended. If the continuity requirement is not met, a new trip chain is initialized, and the complete trip chain set is finally output. The specific steps are as follows: Step S321: Perform spatiotemporal correlation judgment through time interval constraints, that is, calculate the time interval between the end of the previous trip and the start of the current trip. If the time interval does not exceed 0.5 hours, the trip is considered to be continuous. Step S322: When the time interval is ≤0.5 hours and the spatial distance meets the proximity condition of 0.001 kilometers, it is determined that the current trip belongs to the same trip chain as the last trip in the trip chain being constructed.

[0009] Furthermore, in step S4, travel chain features are extracted based on the constructed cold chain freight vehicle travel chain; the specific steps are as follows: Based on the constructed cold chain freight vehicle travel chain, travel chain features are extracted from four aspects: basic features, efficiency-related features, time pattern features, and travel complexity features. Among them, basic features include the number of trips, total trip duration, and total trip mileage; efficiency features include average speed and number of trips per hour; time pattern features include departure time, morning and evening peak departure signs, and nighttime travel signs; and travel complexity features include temperature control level and temperature control stability.

[0010] Furthermore, in step S5, a travel characteristic profile of cold chain freight vehicles is obtained based on the extracted travel chain features; the specific steps are as follows: Step S51: Robust standardization is used to standardize the travel chain features; Step S52: Principal component analysis is used to reduce the dimensionality of the standardized travel chain features. Principal component analysis is set to retain 90% of the feature variance. Step S53: Based on the dimensionality-reduced travel chain features, a density-based point ranking clustering algorithm is used for clustering. The noise points identified by the density-based point ranking clustering algorithm are treated as independent categories for analysis. The density-based point ranking clustering algorithm classifies the travel chain into three types: cluster 0, cluster 1, and a noise point category labeled -1. -1 is regarded as category one, cluster 0 as category two, and cluster 1 as category three. Among them, category one represents the fragmented and inefficient operation mode with the lowest transportation efficiency; category two represents the operation mode of high-frequency short-haul and multi-point delivery within the city; and category three represents the regional / intercity dedicated line transportation mode with the highest single-point direct transportation efficiency.

[0011] Furthermore, in step S6, a multi-task scheduling optimization model for vehicles is constructed based on the obtained travel characteristic profiles of cold chain freight vehicles; the specific steps are as follows: Step S61: Construct a task complexity feature matrix based on the task complexity characteristics, and construct a vehicle complexity feature matrix based on the vehicle capability characteristics. ; in, This represents the task complexity feature matrix. Let g represent the complexity feature vector of the r-th task, where r = 1, 2, ..., g; g represents the maximum value of the task number. The complexity characteristics of the task include path complexity, cargo temperature control requirements, and timeliness. Path complexity is measured by the task path type in the vehicle travel profile, i.e., multi-point delivery within the city or direct delivery to a single point. Cargo temperature control requirements are measured by the refrigeration temperature requirements of the cargo in the vehicle travel profile. Timeliness is measured by the latest allowed delivery time of the cargo in the vehicle travel profile. ; in, This represents the vehicle complexity feature matrix. Let f represent the complexity feature vector of the c-th vehicle, where c = 1, 2, ..., f; and f represents the maximum value of the vehicle number. Vehicle capability characteristics include temperature control capability, transportation frequency, and task adaptability; among them, temperature control capability is measured by the lowest temperature of the vehicle's temperature control system in the vehicle's trip profile; transportation frequency is measured by the number of trips per hour in the vehicle's trip profile; and task adaptability is measured by the vehicle's trip time pattern in the vehicle's trip profile. Step S62: Calculate the complexity matching degree between the task and the vehicle; the formula is as follows: ; in, Indicates task With vehicles The complexity of matching degree, The feature vector representing the complexity of the task. The feature vector representing the complexity of the vehicle; Step S63: Set the constraints for the objective function, including time window constraints, temperature control matching constraints, capacity and load constraints, and complexity capability boundary constraints; the time window constraint formula is as follows: ; in, Indicates the earliest expected completion time of the task. Indicates the actual time the task was completed. Indicates the penalty time for task timeout; Temperature control matching constraint, formula as follows: ; in, This indicates the lowest temperature inside the refrigerated freight truck compartment. This indicates the required refrigeration temperature for the task; The capacity and load-bearing constraints are expressed by the following formula: ; in, The total number of tasks represents the set of tasks. For the task Volume; Indicates task The weight; Let be the decision variable, representing the task. Whether to assign to a vehicle If the task Assigned to vehicles ,but ,otherwise ; For vehicles Maximum volumetric capacity; Indicates vehicle Maximum load capacity; The complexity capability boundary constraint is given by the following formula: ; in, For the task The complexity; This is the difference in complexity that the vehicle can withstand, that is, the maximum complexity that the vehicle can still bear after performing the current task and the tasks that have been assigned. Step S64: Construct an objective function with complexity matching degree as weight, and use linear programming and heuristic algorithms to solve the task allocation scheme that minimizes the total cost under the premise of satisfying constraints; The objective function is as follows: ; in, This indicates taking the minimum value. Indicates the total number of vehicles. Indicates task The transportation cost allocated to vehicle j; Indicates task With vehicles The complexity of the matching degree; These represent the weights of the matching degree and cost in the objective function, respectively. Step S65: The task allocation scheme is used as the scheduling result after matching vehicles with tasks as input. Combined with real-time traffic data and road condition information, the optimal OD pair sequence route for each vehicle is calculated. Step S66: Output the task list, optimal route, and estimated completion time for each vehicle.

[0012] The beneficial effects of this invention are: 1. This invention integrates high-precision spatiotemporal data from the BeiDou Navigation Satellite System, identifies key stopping points through speed threshold segmentation, and refines OD pairs using the DBSCAN clustering algorithm. This effectively solves the unique timeliness requirements and dynamic stopping behavior issues in cold chain transportation, providing accurate analysis for understanding the specific characteristics of cold chain vehicles.

[0013] 2. This invention utilizes the OPTICS clustering algorithm to analyze the characteristics of the travel chain, which can clearly distinguish different operating modes of cold chain vehicles, transforming complex feature data into intuitive and differentiated vehicle operation profiles, revealing the characteristic differences of various operating modes, and providing a basis for decision-making for logistics companies in vehicle scheduling, route planning, and resource optimization. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the present invention. Detailed Implementation

[0015] This embodiment provides a classification method for cold chain freight vehicles based on the complexity of travel tasks, such as... Figure 1 As shown, it includes the following steps: Step S1: Collect trajectory point data of cold chain trucks based on the BeiDou satellite positioning system; preprocess the collected trajectory point data of cold chain trucks; Step S2: Construct OD pairs based on the preprocessed cold chain truck trajectory point data; Step S3: Construct the cold chain freight vehicle travel chain based on the constructed OD pair; Step S4: Extract travel chain features based on the constructed cold chain freight vehicle travel chain; Step S5: Based on the extracted travel chain features, obtain a travel characteristic profile of cold chain freight vehicles. Step S6: Construct a vehicle multi-task scheduling optimization model based on the obtained travel characteristic profile of cold chain freight vehicles.

[0016] Furthermore, in step S1, the collected cold chain truck trajectory point data is preprocessed; a spatiotemporal deduplication algorithm is used for preprocessing; the specific steps are as follows: Step S11: Standardize the time format; Step S12: Based on the BeiDou coding, the collected cold chain truck trajectory point data are grouped, and duplicate data are identified using a set dynamic time window of 5 seconds. Step S13: Based on the temperature inside the compartment, exclude missing data that has not been available for more than 30 consecutive minutes; Step S14: Based on the dual verification mechanism of time window and spatial unique key, duplicate data is removed, and only the first record of each group in the dynamic time window is retained. Then, a triplet of "BeiDou number + time + latitude and longitude" is constructed as a unique index to remove duplicate data. Step S15 introduces a spatial fence mechanism, establishing geographical boundary constraints covering the entire land area based on geographical boundary data; intercepting cross-border data, device location drift data, and contaminated data.

[0017] Furthermore, in step S2, OD pairs are constructed based on the preprocessed cold chain truck trajectory point data; the specific steps are as follows: Step S21: Group the preprocessed cold chain truck trajectory point data according to Beidou coding, and sort them in ascending order by time field within each group; Step S22: Convert latitude and longitude to radian values, calculate the spherical distance between adjacent trajectory points using the semi-sine formula, and calculate the speed in combination with the data collection time interval; when the speed exceeds the legal speed limit threshold, it is judged as abnormal data and removed. At the same time, the vehicle's motion status is divided into five categories according to the speed: stationary, low speed, normal driving, high speed driving, and speeding, and marked accordingly. Step S23: Select stationary and low-speed trajectory points with speeds ≤ 5 km / h as candidate stopping points; Step S24: Use a density-based spatial clustering algorithm with noise to process the data; determine a suitable distance radius threshold and minimum number of neighbor points through the silhouette coefficient to achieve clustering of candidate stopping points, remove isolated points and noise points, and extract well-structured and continuous clusters. Step S25: When there are ≥2 clusters, extract the center point of each cluster as the docking point; Step S26: Select stops with a stop time of ≥10 minutes, and extract the trajectory segments between two adjacent stops to construct OD pairs.

[0018] Furthermore, in step S3, a cold chain freight vehicle travel chain is constructed based on the constructed OD; the specific steps are as follows: Step S31: Using the first OD pair of each vehicle as the starting point of the travel chain, initialize the travel chain object, and record the travel chain ID, vehicle number and the start and end time of the current trip to form a travel chain trip list. Step S32: If the time interval between the current trip and the previous trip meets the continuity requirement and the spatial distance meets the proximity condition, the trip chain is extended. If the continuity requirement is not met, a new trip chain is initialized, and the complete trip chain set is finally output. The specific steps are as follows: Step S321: Perform spatiotemporal correlation judgment through time interval constraints, that is, calculate the time interval between the end of the previous trip and the start of the current trip. If the time interval does not exceed 0.5 hours, the trip is considered to be continuous. Step S322: When the time interval is ≤0.5 hours and the spatial distance meets the proximity condition of 0.001 kilometers, it is determined that the current trip belongs to the same trip chain as the last trip in the trip chain being constructed.

[0019] Furthermore, in step S4, travel chain features are extracted based on the constructed cold chain freight vehicle travel chain; the specific steps are as follows: Based on the constructed cold chain freight vehicle travel chain, travel chain features are extracted from four aspects: basic features, efficiency-related features, time pattern features, and travel complexity features. Among them, basic features include the number of trips, total trip duration, and total trip mileage; efficiency features include average speed and number of trips per hour; time pattern features include departure time, morning and evening peak departure signs, and nighttime travel signs; and travel complexity features include temperature control level and temperature control stability.

[0020] Furthermore, in step S5, a travel characteristic profile of cold chain freight vehicles is obtained based on the extracted travel chain features; the specific steps are as follows: Step S51: Robust standardization is used to standardize the travel chain features; Step S52: Principal component analysis is used to reduce the dimensionality of the standardized travel chain features. Principal component analysis is set to retain 90% of the feature variance. Step S53: Based on the dimensionality-reduced travel chain features, a density-based point ranking clustering algorithm is used for clustering. The noise points identified by the density-based point ranking clustering algorithm are treated as independent categories for analysis. The density-based point ranking clustering algorithm classifies the travel chain into three types: cluster 0, cluster 1, and a noise point category labeled -1. -1 is regarded as category one, cluster 0 as category two, and cluster 1 as category three. Among them, category one represents the fragmented and inefficient operation mode with the lowest transportation efficiency; category two represents the operation mode of high-frequency short-haul and multi-point delivery within the city; and category three represents the regional / intercity dedicated line transportation mode with the highest single-point direct transportation efficiency.

[0021] Furthermore, in step S6, a multi-task scheduling optimization model for vehicles is constructed based on the obtained travel characteristic profiles of cold chain freight vehicles; the specific steps are as follows: Step S61: Construct a task complexity feature matrix based on the task complexity characteristics, and construct a vehicle complexity feature matrix based on the vehicle capability characteristics. ; in, This represents the task complexity feature matrix. Let g represent the complexity feature vector of the r-th task, where r = 1, 2, ..., g; g represents the maximum value of the task number. The complexity characteristics of the task include path complexity, cargo temperature control requirements, and timeliness. Path complexity is measured by the task path type in the vehicle travel profile, i.e., multi-point delivery within the city or direct delivery to a single point. Cargo temperature control requirements are measured by the refrigeration temperature requirements of the cargo in the vehicle travel profile. Timeliness is measured by the latest allowed delivery time of the cargo in the vehicle travel profile. ; in, This represents the vehicle complexity feature matrix. Let f represent the complexity feature vector of the c-th vehicle, where c = 1, 2, ..., f; and f represents the maximum value of the vehicle number. Vehicle capability characteristics include temperature control capability, transportation frequency, and task adaptability; among them, temperature control capability is measured by the lowest temperature of the vehicle's temperature control system in the vehicle's trip profile; transportation frequency is measured by the number of trips per hour in the vehicle's trip profile; and task adaptability is measured by the vehicle's trip time pattern in the vehicle's trip profile. Step S62: Calculate the complexity matching degree between the task and the vehicle; the formula is as follows: ; in, Indicates task With vehicles The complexity of matching degree, The feature vector representing the complexity of the task. The feature vector representing the complexity of the vehicle; Step S63: Set the constraints for the objective function, including time window constraints, temperature control matching constraints, capacity and load constraints, and complexity capability boundary constraints; the time window constraint formula is as follows: ; in, Indicates the earliest expected completion time of the task. Indicates the actual time the task was completed. Indicates the penalty time for task timeout; Temperature control matching constraint, formula as follows: ; in, This indicates the lowest temperature inside the refrigerated freight truck compartment. This indicates the required refrigeration temperature for the task; The capacity and load-bearing constraints are expressed by the following formula: ; in, The total number of tasks represents the set of tasks. For the task Volume; Indicates task The weight; Let be the decision variable, representing the task. Whether to assign to a vehicle If the task Assigned to vehicles ,but ,otherwise ; For vehicles Maximum volumetric capacity; Indicates vehicle Maximum load capacity; The complexity capability boundary constraint is given by the following formula: ; in, For the task The complexity; This is the difference in complexity that the vehicle can withstand, that is, the maximum complexity that the vehicle can still bear after performing the current task and the tasks that have been assigned. Step S64: Construct an objective function with complexity matching degree as weight, and use linear programming and heuristic algorithms to solve the task allocation scheme that minimizes the total cost under the premise of satisfying constraints; The objective function is as follows: ; in, This indicates taking the minimum value. Indicates the total number of vehicles. Indicates task The transportation cost allocated to vehicle j; Indicates task With vehicles The complexity of the matching degree; These represent the weights of the matching degree and cost in the objective function, respectively. Step S65: The task allocation scheme is used as the scheduling result after matching vehicles with tasks as input. Combined with real-time traffic data and road condition information, the optimal OD pair sequence route for each vehicle is calculated. Step S66: Output the task list, optimal route, and estimated completion time for each vehicle.

[0022] In this process, latitude and longitude are converted into radians, and the spherical distance between adjacent trajectory points is calculated using the semi-versus formula, which is: ; in, Indicates speed, Represents the spherical distance between adjacent trajectory points. Indicates the time interval for data collection; The spherical distance between adjacent trajectory points is calculated using the following formula: ; in, This indicates the difference in latitude and longitude between two points. Indicates latitude, 'c' represents longitude, and 'c' represents spherical distance. Represents the distance between two points on the Earth's surface. Indicates the Earth's radius; Among them, for the travel chain Calculate the profile coefficient The formula is as follows: ; in, Indicates travel chain The cluster it belongs to Indicates travel chain Other clusters that are not present Indicates the first The first travel chain and the first The distance between each travel chain; Indicates the first Travel chain to the same cluster The average distance of other travel chains within the chain; Indicates the first The travel chain to the nearest other cluster The average distance; The formula for calculating the basic features is as follows: ; ; ; in, Let z represent the OD pairs of cold chain freight vehicles; This indicates the end time, i.e., the point in time when the trip is completed; This indicates the start time, i.e., the point in time when the trip begins; Indicates the total number of seconds; This represents the travel distance of a single OD pair; Indicates the total trip duration. Indicates the total travel distance. Indicates the number of trips; The efficiency characteristics are calculated using the following formula: ; ; in, Indicates average speed. Indicates the number of trips per hour; The formula for calculating time characteristics is as follows: ; ; ; ; Indicates the departure time in hours. A binary indicator indicating whether the departure was in the morning. A binary indicator indicating whether the departure is in the evening. A binary indicator representing whether or not nighttime travel occurs; The formula for calculating travel complexity is as follows: ; It indicates the complexity of travel, reflecting the complexity of tasks and the diversity of routes during vehicle transportation; Among them, robust standardization is used to standardize the travel chain features; ; in, This represents the characteristics of the travel chain after standardization. Indicates the characteristics of the travel chain. This represents the median of the input travel chain feature X. It represents the interquartile range, which is the difference between the 75th percentile and the 25th percentile of the data.

[0023] As shown in Table 1, based on the constructed cold chain freight vehicle travel chain, the travel chain features are extracted from four aspects: basic features, efficiency-related features, time pattern features, and travel complexity features. Among them, the basic features include the number of trips, the total trip duration, and the total trip mileage; the efficiency features include the average speed and the number of trips per hour; the time pattern features include the departure time (i.e., the number of hours of departure), the departure indicators for morning and evening peak hours, and the night travel indicators (i.e., whether the departure is in the morning, whether it is in the evening, and whether it is at night); and the travel complexity features include the temperature control level and the temperature control stability.

[0024] Table 1 Some features

[0025] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art 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 appended claims and their equivalents.

Claims

1. A cold chain truck classification method based on trip task complexity, characterized in that, The method comprises: Step S1, collecting BeiDou satellite positioning system-based cold chain truck trajectory point data; and preprocessing the collected cold chain truck trajectory point data; Step S2, constructing an OD pair based on the preprocessed cold chain truck trajectory point data; Step S3, constructing a cold chain truck travel chain based on the constructed OD pair; Step S4, extracting a travel chain feature based on the constructed cold chain truck travel chain; Step S5, obtaining a travel characteristic portrait of the cold chain truck based on the extracted travel chain feature; Step S6, constructing a vehicle multi-task scheduling optimization model based on the obtained travel characteristic portrait of the cold chain truck.

2. The method of claim 1, wherein, In step S1, the collected cold chain truck trajectory point data is preprocessed; a space-time deduplication algorithm is used for preprocessing; and the specific steps are as follows: Step S11, unifying the time format; Step S12, grouping the collected cold chain truck trajectory point data based on BeiDou encoding, and using a dynamic time window of 5 seconds to identify duplicate data; Step S13, excluding missing data without temperature data for more than 30 minutes based on the temperature in the compartment; Step S14, removing duplicate data based on a dual verification mechanism of a time window and a spatial unique key, retaining only the first record of each group within the dynamic time window, and constructing a "BeiDou number + time + longitude and latitude" triplet as a unique index to remove duplicate data; Step S15, introducing a spatial fence mechanism, establishing a geographical boundary constraint condition covering the entire territory based on geographical boundary data, and intercepting data of over-border, device positioning drift and pollution.

3. The method of claim 2, wherein, In step S2, the OD pair is constructed based on the preprocessed cold chain truck trajectory point data; and the specific steps are as follows: Step S21, grouping the preprocessed cold chain truck trajectory point data according to BeiDou encoding, and arranging the data in each group in ascending order according to the time field; Step S22, converting the longitude and latitude into radian values, calculating the spherical distance between adjacent trajectory points by using the formula of the secant of half the chord, and calculating the speed in combination with the data collection time interval; when the speed exceeds the legal speed threshold, the data is determined as abnormal and removed, and the vehicle motion state is divided into five categories of static, low speed, normal driving, high speed driving and overspeed and marked according to the speed; Step S23, taking the trajectory points in the static and low speed states with a speed of ≤5 km / h as candidate stop points; Step S24, processing the candidate stop points by using a density-based spatial clustering algorithm with noise; determining a suitable distance radius threshold and a minimum number of neighbor points by using a contour coefficient, clustering and dividing the candidate stop points, removing isolated points and noise points, and extracting a structured and continuous clustering cluster; Step S25, when the clustering cluster ≥2, extracting the center point of each clustering cluster as a stop point; Step S26, screening the stop points with a stop time ≥10 minutes, and extracting a trajectory segment between adjacent two stop points to construct an OD pair.

4. The method of claim 3, wherein, In step S3, the cold chain truck travel chain is constructed based on the constructed OD pair; and the specific steps are as follows: Step S31, taking the first OD pair of each vehicle as the starting point of the travel chain, initializing the travel chain object, and recording the ID, vehicle number and current travel start and end time of the travel chain to form a travel list of the travel chain; In step S32, the trip chain extension is realized by judging that the time interval between the current trip and the previous trip meets the continuity determination requirement and the spatial distance meets the proximity condition, and the trip chain is initialized if the continuity determination requirement is not met, and finally the complete trip chain set is output; the specific steps are as follows: In step S321, the time interval constraint is used for spatio-temporal correlation judgment, that is, the time interval between the end of the last trip and the start of the current trip is calculated, and when the time interval is not more than 0.5 hours, it is considered that the trips have continuity; In step S322, when the time interval is less than or equal to 0.5 hours and the spatial distance meets the proximity condition of 0.001 kilometers, it is determined that the current trip and the last trip in the trip chain being constructed belong to the same trip chain.

5. The method of claim 4, wherein, In step S4, the trip chain features are extracted based on the constructed cold chain freight vehicle trip chain; The specific steps are as follows: Based on the constructed cold chain freight vehicle trip chain, the trip chain features are extracted from four aspects of basic features, efficiency related features, time pattern features and trip complexity features; wherein the basic features include the number of trips, the total trip duration and the total trip mileage, the efficiency features include the average speed and the number of trips per hour, the time pattern features include the departure time point, the morning and evening peak departure identifier and the night trip identifier, and the trip complexity features include the temperature control level and the temperature control stability.

6. The method of claim 5, wherein, In step S5, the trip characteristics of the cold chain freight vehicle are obtained based on the extracted trip chain features; the specific steps are as follows: In step S51, the robust standardization is used for standardization processing of the trip chain features; In step S52, the principal component analysis is used for dimension reduction of the standardized trip chain features, and the principal component analysis is set to retain 90% of the feature variance; In step S53, the clustering is performed based on the dimension-reduced trip chain features by using the density-based point ordering recognition clustering structure algorithm; the noise points recognized by the density-based point ordering recognition clustering structure algorithm are regarded as an independent category for analysis, the trip chain is classified into three categories by the density-based point ordering recognition clustering structure algorithm: cluster 0 and cluster 1 and a noise point category marked as-1, -1 is regarded as category one, cluster 0 is regarded as category two, and cluster 1 is regarded as category three; wherein, category one represents the most inefficient scattered operation mode; category two represents the intra-city high-frequency short-haul, multi-point distribution operation mode; and category three represents the single-point direct transportation mode with the highest efficiency of regional / intercity special line transportation.

7. The method of claim 6, wherein, In step S6, the vehicle multi-task scheduling optimization model is constructed based on the obtained trip characteristics of the cold chain freight vehicle; the specific steps are as follows: In step S61, a task complexity feature matrix is constructed according to the complexity features of the task, and a vehicle complexity feature matrix is constructed according to the capability features of the vehicle; ； wherein, denotes a task complexity feature matrix, denotes a complexity feature vector of the rth task, r = 1, 2,..., g; g denotes a maximum value of the task number; The complexity features of the task include path complexity, cargo temperature control requirement and timeliness; wherein, the path complexity is measured by the task path type in the vehicle trip image, that is, intra-city multi-point distribution and single-point direct transportation; the cargo temperature control requirement is measured by the refrigeration temperature requirement of the cargo in the vehicle trip image; and the timeliness is measured by the latest delivery time allowed by the cargo in the vehicle trip image; ； wherein, represents a vehicle complexity feature matrix, represents a complexity feature vector of the cth vehicle, c = 1, 2,..., f; f represents a maximum value of the vehicle number; The capability features of the vehicles include temperature control capability, transportation frequency, and task adaptability, wherein the temperature control capability is measured by the minimum temperature of the vehicle temperature control system in the vehicle trip profile, the transportation frequency is measured by the number of trips per hour in the vehicle trip profile, and the task adaptability is measured by the vehicle trip time mode in the vehicle trip profile; In step S62, the complexity matching degree of the task and the vehicle is calculated, and the formula is as follows: ； in, Indicates task With vehicles The complexity of matching degree, The feature vector representing the complexity of the task. The feature vector representing the complexity of the vehicle; In step S63, the constraint conditions of the objective function are set, including time window constraint, temperature control matching constraint, capacity and load constraint, and complexity capability boundary constraint. The time window constraint is as follows: ； wherein, represents the earliest expected completion time of the task, represents the actual completion time of the task, represents the overtime penalty time of the task; The temperature control matching constraint is as follows: ； wherein, represents the minimum temperature inside the cold chain freight vehicle compartment, represents the mission refrigeration requirement temperature; The capacity and load constraint is as follows: ； wherein, is the total number of tasks, representing a set of tasks; is the volume of a task ; is the weight of a task ; is a decision variable, representing whether a task is assigned to a vehicle , if a task is assigned to a vehicle , then , otherwise ; is the maximum volume capacity of a vehicle ; is the maximum load capacity of a vehicle ; The complexity capability boundary constraint is as follows: ； wherein, is the complexity of the task ; and is the complexity difference that the vehicle can take, i.e. the maximum complexity that the vehicle can still carry out after performing the current task and the already assigned tasks. In step S64, the objective function is constructed by taking the complexity matching degree as the weight, and the linear programming and heuristic algorithm are used to solve the task allocation scheme for minimizing the total cost under the premise of meeting the constraints. The objective function is as follows: ; wherein, denotes taking the minimum value, denotes the total number of vehicles, denotes the task transportation cost assigned to vehicle j; denotes the task complexity match with vehicle j; denote the weight of control match and cost in the objective function, respectively; In step S65, the task allocation scheme is taken as the scheduling result after the vehicle and the task are matched, and the optimal OD pair sequence route of each vehicle is calculated in combination with the real-time traffic data and road condition information. In step S66, the task list, optimal route, and estimated completion time of each vehicle are output.

Images