A method for estimating truck freight activity patterns based on GPS and sampling survey

Abstract

The present invention proposes a method for estimating truck freight activity patterns based on GPS and sampling surveys. The GPS historical data of the truck object under study is obtained, and a sampling survey is conducted on the truck under study to determine the category of goods it transports and the type of activity it performs at the parking point; based on the truck GPS data, the parking time of each truck at the effective parking point and the distance to the parking point are calculated, and then their probability distribution parameters are estimated respectively; and the probability of the truck engaging in a specific activity at the effective parking point is calculated based on this; the probability of the truck engaging in freight-related activities at the parking point is then corrected; and the truck's freight activity itinerary chain is determined based on this, which is the truck freight activity pattern. The method of the present invention comprehensively utilizes multivariate data including GPS and sampling survey data to estimate the truck freight activity pattern, which helps to deepen the understanding of freight demand patterns and corresponding activity locations, and provides an important basis for the optimal design of freight-related infrastructure networks.

Description

Technical Field

[0001] The invention belongs to the field of logistics network planning, and in particular relates to a truck freight activity pattern estimation method based on GPS and sampling survey. Background Art

[0002] The efficient flow of industrial raw materials and goods is crucial to a nation's development. In recent years, due to growing globalization, driven by advances in information technology, changes in consumer shopping habits (such as online shopping), and shifts in industry logistics (such as just-in-time delivery), the movement of raw materials and goods has increased in both quantity and distance, leading to a significant increase in the volume and size of freight transport. Furthermore, the characteristics and behavior of these freight movements have changed significantly over the past few decades. To better adapt to rapidly changing freight demand and serve a nation's economic growth, the optimized design of freight-related infrastructure and logistics supply chain networks is crucial. The successful planning and design of freight infrastructure and supply chain networks relies heavily on understanding and statistically analyzing freight demand patterns and the corresponding locations of freight activity.

[0003] Most previous studies have used sampling surveys of truck drivers to determine their travel patterns and freight activities. However, sampling surveys require significant human resources and costs, making it difficult to obtain large-scale, reliable data. In recent years, some studies have used simple rule-based heuristics or machine learning algorithms (such as random forests and support vector machines) based on truck GPS data to predict activity types. However, these studies have not considered the characteristics of truck activity patterns at specific stops (such as activity duration) or the types of goods carried, making it difficult to identify the statistical patterns of truck freight activity and the driving conditions between different activities. Furthermore, the accuracy of activity pattern estimates based solely on GPS is low. Further sampling surveys are needed to obtain real sample activity pattern data and refine the accuracy of activity pattern estimates based on GPS data, thereby providing effective guidance for infrastructure planning and design. Summary of the Invention

[0004] In order to solve the technical problems mentioned in the above background technology, the present invention proposes a truck freight activity pattern estimation method based on GPS and sampling survey.

[0005] In order to achieve the above technical objectives, the technical solution of the present invention is:

[0006] A method for estimating truck freight activity patterns based on GPS and sampling surveys includes the following steps:

[0007] S1. Obtain GPS data of each truck i through the GPS tracking system, including timestamp, speed v, and latitude and longitude. Define the collected GPS data as an ordered set of timestamp, speed v, and corresponding latitude and longitude;

[0008] S2. Determine the effective parking point s for truck i using the minimum stay time and minimum travel distance, conduct a sampling survey of trucks, and determine the locations where trucks stay and the activities they perform;

[0009] S3. Based on the truck GPS data obtained in step S1, calculate the parking time of each truck i at the valid parking spot s obtained in step S2 and the distance from the nearest main road to the parking spot. The sequential set of each truck's parking location and parking time is called a trip chain. After removing duplicate trip chains of different trucks, all trip chains are set E.

[0010] S4. Assume that the truck's stay time at the valid stop s and the distance from the nearest road to the stop follow lognormal distributions, and estimate the mean and standard deviation of the two lognormal distributions based on the maximum likelihood method.

[0011] S5. Calculate the category and probability of the truck engaging in an activity at the valid parking spot based on the lognormal distribution parameters estimated in step S4 and the relevant location data of the truck at the valid parking spot;

[0012] S6. Based on the sampling survey data and the Bayesian model, further modify the probability of the truck engaging in the specific activity in step S5 at the parking point in step S5;

[0013] S7. Determine the trip chain set of truck i carrying product c based on the probability corrected in step S6. This is the activity pattern related to freight transportation.

[0014] Preferably, step S2 specifically includes the following steps:

[0015] S21. Delete the trucks with incomplete GPS data obtained in step S1, arrange the original GPS data of each truck in the order of its timestamp, define the timestamp when the truck changes from the running phase with a speed v>0 to the stationary phase with a speed v=0 as the start time of the stop at the parking point, and define the timestamp when the truck changes from the stationary phase to the running phase as the end time of the stop at the parking point, and then determine the time difference between any two consecutive GPS data points, and determine the distance difference based on the change in longitude and latitude;

[0016] S22, using the minimum stay time and the minimum driving distance to determine the truck's effective parking point, if the difference between the start time and the end time of truck i at the parking point is greater than the minimum stay time y min, then the parking point is considered to be a valid parking point s, if it is less than the shortest stay time y min , then the parking point is considered not to be a valid parking point. Based on the valid parking point s, the distance traveled by truck i to the next parking point s+1 is collected. If this distance is greater than the minimum driving distance d min , then determine that s+1 is a new potential parking point, otherwise determine that s+1 is the same valid parking point as s. For the new potential parking point, use the shortest stay time to make a judgment, and so on, until the valid parking points of all trucks are obtained;

[0017] S23. Through step S22, a set of valid parking points s can be obtained. For each valid parking point s, the difference between the start time and the end time is taken as the parking time, and the average latitude and longitude during this period is taken as the location of s;

[0018] S24. Arrange the valid parking points of truck i in chronological order, and the resulting sequential set is the trip chain of truck i;

[0019] S25. Conduct a sample survey of the GPS-equipped trucks under study to determine the locations of the effective parking spots and activities performed by the surveyed trucks to improve subsequent activity estimates.

[0020] Preferably, step S3 includes the following steps:

[0021] S31, let the activity of truck i at the effective parking point s be a i,s , the activity set of truck i is A i ,y i,s is the dwell time of truck i at parking point s calculated based on GPS data, x i,s is the distance from the nearest main road to the stop point s when truck i stops at the stop point s;

[0022] S32. Define indicator variables If truck i stops at a valid parking spot s and performs a freight-related activity f, then on the contrary Assume that truck i performs activity f with a vehicle dwell time y f and the distance x from the nearest main road to the potential parking spot f Independent and obey the log-normal distribution, the formula is as follows:

[0023] lny f ~N(μ y,f ,(σ y,f ) 2 ) (1)

[0024] lnx f ~N(μ x,f,(σ x,f ) 2 ) (2)

[0025] Among them, lny f and lnx f y f and x f The logarithm of N represents the normal distribution, μ y,f ,σ y,f is the vehicle dwell time y f The mean and standard deviation of the lognormal distribution, μ x,f ,σ x,f is the distance x from the nearest main road to the potential parking spot f The mean and standard deviation of the lognormal distribution, y f and x f The joint probability density function formula is expressed as follows:

[0026]

[0027] Where Λ=(μ y,f ,μ x,f ,σ y,f ,σ x,f ), M is the number of all trucks in the GPS data, S i is the number of valid parking points that truck i passes through, F s is the set of all feasible activities of the truck at the valid stop s (including freight-related activities and non-freight-related activities);

[0028] y f and x f The probability density function formula is expressed as follows:

[0029]

[0030]

[0031] Among them, h(μ y,f ,σ y,f |y i,s ) is the random variable y f The probability density function, g(μ y,f ,σ x,f |x i,s ) is a random variable x f The probability density function of

[0032] Taking the logarithm of the joint probability density function, the log-likelihood function formula is expressed as follows:

[0033]

[0034] S33: The dwell time y of truck i at parking point s obtained in step S31 i,s and the distance x from the nearest road to the parking spot i,s and activity indicators The parameters of the model can be estimated by maximizing the value of the above log-likelihood function

[0035] Preferably, in step S4, the formula for determining the probability that truck i performs activity f at the valid parking point s is expressed as follows:

[0036]

[0037] where Λ * is the model parameter estimated in right 3, y i,s ,x i,s are the vehicle dwell time and the distance from the nearest main road to the potential parking point for truck i performing activity f obtained from GPS data, respectively;

[0038] Since truck i is at a valid stop s and performs activity f, the variables The value of is 0 or 1, so formula (7) can be further expressed as:

[0039]

[0040] Since the activities f performed by truck i at the valid parking point s are independent of each other, formula (8) can be further expressed as follows according to the probability density function:

[0041]

[0042] in, is the mean and variance of the lognormal distribution of the dwell time of vehicle i at parking point s and the distance from the nearest main road to the potential parking point, estimated in step S33, where activity f is included in the set F defined in step S41 s In, similar, are the mean and variance of the lognormal distribution of the dwell time of vehicle i at parking point s performing activity j and the distance from the nearest main road to the potential parking point, estimated in step S33,

[0043] make Taking the argmax function of formula (9), the formula is expressed as follows, and the estimated activity a performed by truck i when parking at the valid parking point s is obtained: i,s :

[0044]

[0045] Where M is the number of all trucks in the data, S i is the number of valid parking points that truck i passes through.

[0046] Preferably, step S5 specifically includes the following steps:

[0047] When the parking point of truck i is within a certain area l, the probability formula for it performing activity f at parking point s is expressed as follows:

[0048]

[0049] in is the probability that truck i performs activity f at a valid stop s, Pr(i,s,f′) is the probability that truck i performs activity f′ at a valid stop s, Pr(l|f′) is the probability that truck i stays in area l when assuming it performs activity f′, and Pr(l|f) is the probability that truck i stays in area l when assuming it performs activity f. The formula for Pr(l|f) is as follows:

[0050]

[0051] Among them, O f,l is the number of trucks that stop at the investigated location l∈L to perform activity f, O f,l′ is the number of trucks that stop at l′∈L to perform activity f, and L is the set of parking areas with high truck density obtained through a sample survey.

[0052] Preferably, the activities performed by the truck at the effective parking point in step 5 are divided into two categories, wherein freight-related activities include loading and unloading; and non-freight-related activities include stopping for rest, meals and refueling.

[0053] Preferably, step S7 specifically includes the following steps:

[0054] S61. Based on a sampling survey of trucks, obtain the trip chain set Ξ of the surveyed truck i. i , identify all freight-related activities f;

[0055] S62, if i If the trip chain set E is known, the type of goods c carried by the vehicle can be determined. Otherwise, the trip chain set Ξ i The corresponding parking point location matches the corresponding GIS geographic location, and the type of goods it is transporting c is determined based on the geographic location attributes;

[0056] S63: Based on the determined commodity type c and the truck trip chain set Ξ i , update the known trip chain set E, and at the same time, according to the trip chain set of truck i carrying commodity c After that, it is possible to identify, for example, activity patterns related to freight transport.

[0057] The beneficial effects brought about by adopting the above technical solution are:

[0058] The method of the present invention comprehensively utilizes multivariate data including GPS and sampling survey data, which makes the analysis more reliable, helps to deepen the understanding of freight demand patterns and corresponding activity locations, and provides an important basis for the optimized design of freight-related infrastructure networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 is a flow chart of a method for estimating a truck freight activity pattern of the present invention;

[0060] Figure 2 It is a detailed diagram of 3 different types of truck trip chains;

[0061] Figure 3 It is a diagram that explains the meaning of the defined symbols with an example. DETAILED DESCRIPTION

[0062] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0063] like Figure 1 As shown, an embodiment of the present invention discloses a method for estimating truck freight activity patterns based on GPS and sampling survey, comprising the following steps:

[0064] (1) Obtaining GPS historical data of the trucks under study. Specifically, 9,041 trucks were selected and equipped with GPS tracking systems to track the latitude and longitude, speed, travel distance, and timestamp of the trucks’ trajectories. The data from August 2017 was used to estimate their activity patterns according to the proposed method.

[0065] (2) A random survey was conducted on the truck drivers to determine the categories of goods they transported and the types of activities they carried out at the stops. Specifically, 50 trucks were randomly selected from all the trucks equipped with GPS tracking systems and their main stops and transported goods in five provinces in August 2017. The goods transported by the trucks included seven categories, namely beverages, sugar, rice, animal feed, paper, freight containers, and mixed concrete;

[0066] (3) Based on the truck GPS data, the valid parking points and parking times of each truck and the distance from the nearest main road to the parking point are extracted. The sequential set of all parking point locations and parking times becomes a trip chain.

[0067] In this example, incomplete GPS records missing critical data are first deleted. Second, for a single truck's GPS record set, if there is a large time interval between consecutive GPS records, all of them are deleted. The cleaned GPS data for each truck is sorted by its timestamp to determine the time and distance differences between any two consecutive GPS data points. These differences are used to determine valid truck parking locations.

[0068] The minimum stay time and minimum travel distance are used to determine the effective parking point. If the difference between the start time and the end time of truck i at the parking point s is greater than the minimum stay time y min , then the parking point s is considered to be an effective parking point if it is less than the shortest stay time y min , then the parking point s is considered not to be a valid parking point. Based on the valid parking points, the distance traveled by truck i to point s+1 is collected. If this distance is greater than the minimum travel distance d min , then determine that s+1 is a new potential parking point, otherwise determine that s+1 is the same valid parking point as s. For the new potential parking point, use the shortest stay time to make a judgment, and so on.

[0069] In this example, the minimum residence time y min Set to 20 minutes. Since trucks can make multiple stops in large facilities, the minimum driving distance d min At industrial sites or ports and other remaining facilities, the distances are 2 km and 20-100 m respectively, thus determining effective truck parking spots according to the above steps.

[0070] After the above judgment, the set of valid parking points s is obtained. The average latitude and longitude of each valid parking point s in the corresponding time period is taken as the location of s, and the difference between the start time and the end time of parking point s is taken as the parking time. The valid parking points of truck i are sorted in chronological order. The resulting sequential set is the trip chain of truck i. The set of different types of trip chains is the set of activity patterns, such as Figure 2 As shown, there are three different types of activity modes.

[0071] In order to explain the meaning of specific symbols, we further combine Figure 3 For example, the symbolic representation of the relevant activities of Truck 1 is:

[0072] Activity Collection

[0073] A1 = freight-related activities, freight-related activities, meal, freight-related activities, refueling, and vehicle collection.

[0074] Freight-related activity sequence set

[0075] Trip Chain

[0076] According to the sorted GPS data obtained in step S31, the stay time y of truck i at parking point s * and the distance x from the nearest road to the parking spot * and the activity indicator variable z * , the parameters of the model can be estimated through the above log-likelihood function Here are the results:

[0077] Table 1 Parameters of the distribution corresponding to 7 different types of commodities

[0078]

[0079] Then, the probability of a truck performing a specific activity at a valid stop is calculated. The probability of truck i performing activity f at a valid stop s is

[0080]

[0081] where Λ * is the estimated parameter, y i,s ,x i,s are the vehicle dwell time of truck i performing activity f and the distance from the nearest main road to the potential parking point obtained from GPS data; the indicator variable for truck i performing activity f at the valid parking point s is The value of is 0 or 1, so the above probability formula can be further expressed as

[0082]

[0083] Since truck i performs activities f at valid parking points s independently, the probability density function can be further expressed as

[0084]

[0085] in and are all probability density functions, and their parameters are and are the corresponding estimates obtained in Table 1, where are the estimated mean and standard deviation of the distribution of the dwell time of vehicle i at parking point s performing activity f and the distance from the nearest main road to the potential parking point, is the estimated mean and standard deviation of the vehicle i performing activity j at parking point s, where activity j is included in the set F defined in step S41. s Inside.

[0086] make Take the argmax function for the above probability, that is

[0087]

[0088] We can get the activity with the maximum probability value, which is the activity where truck i stops at the valid parking spot s.

[0089] Based on the trip information of the sampled trucks, the probability of trucks engaging in freight-related activities at the parking points is further modified. When truck i stops in area l, the probability of it engaging in activity f at parking point s is:

[0090]

[0091] in is the probability that truck i performs activity f at a valid stop s, Pr(i,s,f′) is the probability that truck i performs activity f′ at a valid stop s, Pr(l|f′) is the probability that truck i stays in area l when assuming it performs activity f′, and Pr(l|f) is the probability that truck i stays in area l when assuming it performs activity f, that is,

[0092]

[0093] Among them, O f,l is the number of trucks that stopped at area l to perform activity f obtained from the survey.

[0094] Here, a total of 50 trucks’ travel information in 5 provinces is investigated, that is, L is the set of all parking points of the 50 trucks in 5 provinces obtained by sampling survey, o f,l The number of trucks at stop l obtained from the survey.

[0095] Using the corrected probabilities, we can identify areas with a high density of truck stops. In this example, the average dwell times for the top five provinces with the most resting activities were 69.8 minutes, 68.5 minutes, 67.0 minutes, 62.3 minutes, 60.2 minutes, and 54.4 minutes, respectively. The average dwell times for the top five provinces with the most loading / unloading activities were 183.7 minutes, 175.0 minutes, 159.7 minutes, 158.0 minutes, 146.6 minutes, and 143.5 minutes, respectively. Resting and loading / unloading activities in these top five provinces accounted for 27.83% and 47.04% of the total on August 3, 2017, respectively. We can see that the characteristics of these activities vary significantly across provinces. This variation may be due to differences in the cargo carried by trucks and the behavior of drivers within these provinces.

[0096] To assess the status of product journey chains or journey segments, we sorted the raw GPS data of each truck in August 2017 by timestamp every three consecutive days. We then predicted the journey chains, freight-related activity patterns, and journey status between two points for seven different types of products:

[0097] 1) Based on a survey of the trip information of 50 trucks in 5 provinces, we obtain the trip chain set E of the investigated trucks and determine all freight-related activities f, such as loading and unloading;

[0098] 2) If the trip chain of truck i is i If the trip chain set E is known, the type of goods c carried by the vehicle can be determined. Otherwise, the trip chain set Ξ i The corresponding parking point location matches the corresponding GIS geographic location, thereby determining the type of goods it is transporting c;

[0099] 3) Based on the determined commodity type c and the truck trip chain set Ξ i , update the previously known itinerary chain set E;

[0100] 4) After obtaining the trip chain of truck i (e.g., rice mill – port – rice mill), the activity pattern related to freight transportation can be determined. If it is assumed that all cargo is unloaded, the trip status (loaded / empty) between two points can be determined.

[0101] The number of different product journey chains varies depending on the type of product. Here, taking the transported product rice as an example, the detailed information of the three different types of truck journey chains is shown in the instruction manual. Figure 3 As shown in the figure, the rice commodity journey chain in Model 1 is always loaded at the rice mill, unloaded at another location (such as a port), and finally returned to the rice mill for parking. In Model 2, in addition to a single unloading location, there are three other models with two unloading locations after a single loading at the rice mill. Model 3 is more special, in which rice is loaded simultaneously at the rice mill and warehouse, and unloaded simultaneously at the factory and distribution center.

[0102] In summary, the method of the present invention comprehensively considers the use of GPS and sampling survey data to determine how to determine parking points, estimate truck activities, etc., providing an important basis for optimizing the design of freight-related infrastructure and supply chain networks, helping to cope with changes in the characteristics and behavior of freight, and promoting the success of freight infrastructure planning and design.

[0103] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal translation scripting language JavaScript, etc.

[0104] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0107] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0108] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for estimating truck freight activity patterns based on GPS and sampling survey, characterized in that: The following steps are involved: S1. Obtain GPS data of each truck i through the GPS tracking system, including timestamp, speed v, and latitude and longitude. Define the collected GPS data as an ordered set of timestamp, speed v, and corresponding latitude and longitude; S2. Determine the effective parking point s for truck i using the minimum stay time and minimum travel distance, conduct a sampling survey of trucks, and determine the locations where trucks stay and the activities they perform; S3. Based on the truck GPS data obtained in step S1, calculate the parking time of each truck i at the valid parking spot s obtained in step S2 and the distance from the nearest main road to the parking spot. The sequential set of each truck's parking location and parking time is called a trip chain. After removing duplicate trip chains of different trucks, all trip chains are set E. S4. Assume that the truck's stay time at the valid stop s and the distance from the nearest road to the stop follow lognormal distributions, and estimate the mean and standard deviation of the two lognormal distributions based on the maximum likelihood method. S5. Calculate the category and probability of the truck engaging in an activity at the valid parking spot based on the lognormal distribution parameters estimated in step S4 and the relevant location data of the truck at the valid parking spot; S6. Based on the sampling survey data and the Bayesian model, further modify the probability of the truck engaging in the specific activity in step S5 at the parking point in step S5; S7. Determine the trip chain set of truck i carrying product c based on the probability corrected in step S6. That is, the activity pattern related to freight transport; The step S2 specifically includes the following steps: S21. Delete the trucks with incomplete GPS data obtained in step S1, arrange the original GPS data of each truck in the order of its timestamp, define the timestamp when the truck changes from the running phase with a speed v>0 to the stationary phase with a speed v=0 as the start time of the stop at the parking point, and define the timestamp when the truck changes from the stationary phase to the running phase as the end time of the stop at the parking point, and then determine the time difference between any two consecutive GPS data points, and determine the distance difference based on the change in longitude and latitude; S22, using the minimum stay time and the minimum driving distance to determine the truck's effective parking point, if the difference between the start time and the end time of truck i at the parking point is greater than the minimum stay time y min , then the parking point is considered to be a valid parking point s, if it is less than the shortest stay time y min , then the parking point is considered not to be a valid parking point. Based on the valid parking point s, the distance traveled by truck i to the next parking point s+1 is collected. If this distance is greater than the minimum driving distance d min , then determine that s+1 is a new potential parking point, otherwise determine that s+1 is the same valid parking point as s. For the new potential parking point, use the shortest stay time to make a judgment, and so on, until the valid parking points of all trucks are obtained; S23. Through step S22, a set of valid parking points s can be obtained. For each valid parking point s, the difference between the start time and the end time is taken as the parking time, and the average latitude and longitude during this period is taken as the location of s; S24. Arrange the valid parking points of truck i in chronological order, and the resulting sequential set is the trip chain of truck i; S25. Conduct a sample survey of the GPS-equipped trucks under study to determine the locations of the effective parking spots and activities performed by the surveyed trucks to improve subsequent activity estimates.

2. The method for estimating truck freight activity patterns based on GPS and sampling survey according to claim 1, characterized in that: Step S3 includes the following steps: S31, let the activity of truck i at the effective parking point s be a i,s , the activity set of truck i is A i ,y i,s is the dwell time of truck i at parking point s calculated based on GPS data, x i,s is the distance from the nearest main road to the stop point s when truck i stops at the stop point s; S32. Define indicator variables If truck i stops at a valid parking spot s and performs a freight-related activity f, then on the contrary Assume that truck i performs activity f with a vehicle dwell time y f and the distance x from the nearest main road to the potential parking spot f Independent and obey the log-normal distribution, the formula is as follows: Linen f ~N(μ y,f ,(σ y,f ) 2 ) (1) lnx f ~N(μ x,f ,(σ x,f ) 2 ) (2) Among them, lny f and lnx f y f and x f The logarithm of N represents the normal distribution, μ y,f ,σ y,f is the vehicle dwell time y f The mean and standard deviation of the lognormal distribution, μ x,f ,σ x,f is the distance x from the nearest main road to the potential parking spot f The mean and standard deviation of the lognormal distribution, y f and x f The joint probability density function formula is expressed as follows: Where Λ=(μ y,f ,μ x,f ,σ y,f ,σ x,f ), M is the number of all trucks in the GPS data, S i is the number of valid parking points that truck i passes through, F s is the set of all feasible activities of the truck at the valid stop s, including freight-related activities and non-freight-related activities; y f and x f The probability density function formula is expressed as follows: Among them, h(μ y,f ,σ y,f |y i,s ) is the random variable y f The probability density function, g(μ y,f ,σ x,f |x i,s ) is a random variable x f The probability density function of Taking the logarithm of the joint probability density function, the log-likelihood function formula is expressed as follows: S33: The dwell time y of truck i at parking point s obtained in step S31 i,s and the distance x from the nearest road to the parking spot i,s and activity indicators The parameters of the model can be estimated by maximizing the value of the above log-likelihood function 3. The method for estimating truck freight activity patterns based on GPS and sampling survey according to claim 2, characterized in that: In step S4, the formula for determining the probability that truck i performs activity f at a valid parking point s is as follows: where Λ * is the model parameter estimated in step S33, y i,s ,x i,s are the vehicle dwell time and the distance from the nearest main road to the potential parking point for truck i performing activity f obtained from GPS data, respectively; Since truck i is at a valid stop s and performs activity f, the variables The value of is 0 or 1, so formula (7) can be further expressed as: Since the activities f performed by truck i at the valid parking point s are independent of each other, formula (8) can be further expressed as follows according to the probability density function: in, is the mean and variance of the lognormal distribution of the dwell time of vehicle i at parking point s and the distance from the nearest main road to the potential parking point, estimated in step S33, where activity f is included in the set F defined in step S41 s Inside, are the mean and variance of the lognormal distribution of the dwell time of vehicle i at parking point s performing activity j and the distance from the nearest main road to the potential parking point, estimated in step S33, make Taking the argmax function of formula (9), the formula is expressed as follows, and the estimated activity a performed by truck i when parking at the valid parking point s is obtained: i,s : Where M is the number of all trucks in the data, S i is the number of valid parking points that truck i passes through.

4. The method for estimating truck freight activity patterns based on GPS and sampling survey according to claim 1, characterized in that: Step S5 specifically includes the following steps: When the parking point of truck i is within a certain area l, the probability formula for it performing activity f at parking point s is expressed as follows: in is the probability that truck i performs activity f at a valid stop s, Pr(i, s, f') is the probability that truck i performs activity f' at a valid stop s, Pr(l|f') is the probability that truck i stays in area l when assuming it performs activity f', and Pr(l|f) is the probability that truck i stays in area l when assuming it performs activity f. The formula for Pr(l|f) is as follows: Among them, O f,l is the number of trucks that stop at the investigated location l∈L to perform activity f, O f,l' is the number of trucks that stop at l'∈L to perform activity f, and L is the set of parking areas with high truck density obtained through a sample survey.

5. The method for estimating truck freight activity patterns based on GPS and sampling survey according to claim 1, characterized in that: In step 5, the activities that trucks perform at valid stops are divided into two categories: freight-related activities include loading and unloading; and non-freight-related activities include stopping for rest, meals, and refueling.

6. The method for estimating truck freight activity patterns based on GPS and sampling survey according to claim 1, characterized in that: Step S7 specifically includes the following steps: S61. Based on a sampling survey of trucks, obtain the trip chain set E of the surveyed truck i i , identify all freight-related activities f; S62, if E i If the trip chain set E is known, the type of goods c carried by the vehicle can be determined. Otherwise, the trip chain set E i The corresponding parking point location matches the corresponding GIS geographic location, and the type of goods it is transporting c is determined based on the geographic location attributes; S63, based on the determined commodity type c and the truck trip chain set E i , for the known itinerary chain set E Update, at the same time, according to the trip chain set of truck i carrying commodity c After that, the activity patterns related to freight transport can be identified.

Images