Vehicle-cargo matching and pricing optimization method and system for logistics scheduling

Abstract

The present invention belongs to the field of logistics technology, and in particular to a vehicle-cargo matching and pricing optimization method and system for logistics scheduling. The method constructs a vehicle transport information database through a natural language processing algorithm, and has a built-in vehicle-cargo matching model to obtain vehicle attribute information, vehicle vacancy information at the current moment, transportation qualification information and historical cargo transportation information. Secondly, an evaluation model is constructed through an evaluation algorithm to obtain a comprehensive transportation difficulty score, and the score is input into the vehicle-cargo matching model to obtain a vehicle-cargo matching score sequence, a matching score threshold is set, and the sufficiency is calculated according to the number of vehicles and cargo demand that meet the threshold at the current moment, and a dynamic pricing model is constructed using reinforcement learning. The transportation price is adjusted according to factors such as vehicle sufficiency, transportation time and transportation difficulty score to ensure that the transaction rate meets the predetermined threshold. The present invention can effectively improve the efficiency and accuracy of logistics scheduling, reduce logistics costs, and improve customer satisfaction.

Description

Technical Field

[0001] The present invention belongs to the field of logistics technology, and in particular relates to a vehicle-cargo matching and pricing optimization method and system for logistics scheduling. Background Art

[0002] With the rapid development of e-commerce and global trade, the logistics industry is facing increasing pressure. How to reduce costs while ensuring service quality has become the focus of enterprises; traditional logistics scheduling methods mostly rely on manual experience or simple calculation models, which is not only inefficient, but also prone to waste of resources. In addition, when matching goods, sometimes some dangerous goods may appear, which may make the dispatched vehicles mismatched, and when pricing, the price may not match the importance and danger of the transported goods.

[0003] For example, the patent with publication number CN116739465A discloses a pricing analysis method and system based on a vehicle-cargo matching platform. This method conducts a comprehensive analysis based on the actual factors affecting freight prices to obtain a reasonable and well-founded pricing strategy. It uses correlation and regression analysis to screen out the main factors that have a greater impact on prices from among the many price influencing factors. At the same time, it uses a particle swarm optimization algorithm to calculate the freight pricing of the platform, and evaluates the rationality of the pricing through a rank sum ratio comprehensive model, and conducts a comprehensive fitting evaluation on the prediction results.

[0004] The above existing technologies have the following problems: when matching, the existing technologies often result in a mismatch between the vehicle and the cargo transportation qualifications, and when determining the transportation price, the particularity of the cargo and the difficulty of transportation are not taken into account, resulting in an unequal cargo-price situation. In order to solve the above problems, the present invention provides a vehicle-cargo matching and pricing optimization method and system for logistics scheduling. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention proposes a vehicle-cargo matching and pricing optimization method and system for logistics scheduling. The present invention constructs a vehicle transportation information database through a natural language processing algorithm, and has a built-in vehicle-cargo matching model to obtain vehicle attribute information, current vehicle vacancy information, transportation qualification information and historical cargo transportation information. Secondly, an evaluation model is constructed through an evaluation algorithm to obtain a comprehensive transportation difficulty score, and the score is input into the vehicle-cargo matching model to obtain a vehicle-cargo matching score sequence, and a matching score threshold is set. The sufficiency is calculated according to the number of vehicles and cargo demand that meet the threshold at the current moment, and a dynamic pricing model is constructed using reinforcement learning. The transportation price is adjusted according to factors such as vehicle sufficiency, transportation time and transportation difficulty score to ensure that the transaction rate meets the predetermined threshold. The present invention can effectively improve the efficiency and accuracy of logistics scheduling, reduce logistics costs, and improve customer satisfaction.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Vehicle-cargo matching and pricing optimization methods for logistics scheduling, including:

[0008] S1: Obtain vehicle attribute information, current vehicle vacancy information, transportation qualification information and historical cargo transportation information, build a vehicle transportation information database through natural language algorithm, and build a vehicle-cargo matching model in the constructed vehicle transportation information database;

[0009] S2: Obtain cargo attribute information and cargo transportation information, build an evaluation model through an evaluation algorithm, input the acquired cargo attribute information and cargo transportation information into the evaluation model for training, obtain a comprehensive transportation difficulty score, input the acquired comprehensive transportation difficulty score and cargo attribute information into the vehicle-cargo matching model built into the vehicle transportation information database, and obtain a vehicle-cargo matching score sequence;

[0010] S3: Set a matching score threshold, obtain the number of vehicles that meet the threshold at the current moment according to the vehicle-cargo matching score and the matching score threshold, and obtain the corresponding vehicle abundance at the current moment according to the number of vehicles that meet the threshold at the current moment and the vehicle demand corresponding to the current cargo;

[0011] S4: Build a dynamic pricing model through reinforcement learning, input the obtained vehicle abundance, transportation time, and transportation difficulty score that meet the threshold conditions into the pricing strategy model, and obtain the optimal transportation price for the current cargo corresponding to the transportation vehicle;

[0012] S5: Set a transaction rate threshold, obtain the vehicle transaction rate in real time according to the current transportation price, and feed the vehicle transaction rate back to the dynamic pricing model to adjust the price at the current moment so that the vehicle transaction rate meets the transaction rate threshold in real time.

[0013] Specifically, the specific steps of constructing the vehicle transport information database include:

[0014] S101, constructing an information extraction model through a Chinese pre-trained Bert model, and inputting the acquired vehicle attribute information, current vehicle vacancy information, transportation qualification information, and historical cargo transportation information into the information extraction model to obtain a vehicle information triplet of the g-th vehicle under the f-th vehicle type;

[0015] S102, extracting the transport qualification and historical cargo transport triplet information from the vehicle information triplet, and inputting the obtained transport qualification and historical cargo transport triplet into the comprehensive evaluation algorithm to obtain the vehicle transport qualification score, and at the same time inputting the vehicle information triplet corresponding to the vehicle attribute information into the comprehensive evaluation algorithm to obtain the vehicle comprehensive status score;

[0016] S103, updating the obtained vehicle transport qualification score and vehicle comprehensive status score into the corresponding vehicle information triplet with the license plate number in the vehicle attribute information as the connection attribute information for storage;

[0017] S104: Save the acquired vehicle information triples into a graph database to obtain a vehicle transport information database.

[0018] Specifically, the specific steps for obtaining the comprehensive transportation difficulty score in S2 include:

[0019] S201. Use cargo attribute information to construct a cargo risk assessment factor corresponding to the current cargo type And evaluation index sequence ; And use the expert experience evaluation method to evaluate each evaluation factor in the cargo risk assessment factor in turn according to the evaluation index sequence, and obtain the fuzzy evaluation matrix R corresponding to the cargo risk assessment factor; Indicates the nth cargo hazard assessment factor, represents the mth evaluation index; N represents the total number of cargo risk assessment factors, and M represents the total number of evaluation indicators;

[0020] S202: According to the obtained fuzzy evaluation matrix R, the weighted coefficient is obtained by the minimum membership weighted average deviation. ,in Indicates the weighting coefficient corresponding to the nth cargo risk assessment factor;

[0021] S203, will be constructed , and Input into the comprehensive fuzzy evaluation algorithm to obtain the cargo hazard level assessment score corresponding to the current cargo type.

[0022] Specifically, the specific steps for obtaining the comprehensive transportation difficulty score in S2 also include:

[0023] S204, constructing a transportation difficulty assessment factor using the cargo transportation information, and inputting the constructed transportation difficulty assessment factor into the PCA algorithm to obtain the contribution of the transportation difficulty assessment factor to the transportation difficulty score;

[0024] S205, setting a contribution threshold, and obtaining a valid transport difficulty assessment factor sequence whose contribution is greater than the contribution threshold through the contribution threshold and the contribution corresponding to the obtained transport difficulty assessment factor;

[0025] S206. Constructing a transport difficulty regression model through a support vector machine, and inputting the obtained effective transport difficulty assessment factor sequence into the transport difficulty regression model to obtain a transport difficulty score;

[0026] S207. Multiply the cargo risk assessment score corresponding to the current cargo type by the transportation difficulty score to obtain a comprehensive transportation difficulty score corresponding to the current cargo type.

[0027] Specifically, the steps of obtaining the vehicle-cargo matching score sequence in S2 include:

[0028] S210: Score the vehicle transport qualification stored in the graph database Compare it with the cargo hazard assessment score corresponding to the current cargo type. If the vehicle transportation qualification score is If the vehicle transport qualification score is greater than the cargo hazard level assessment score corresponding to the current cargo type, then the g-th vehicle under the f-th vehicle type meets the initial vehicle-cargo matching; if the vehicle transport qualification score is less than or equal to the cargo hazard level assessment score corresponding to the current cargo type, then the g-th vehicle under the f-th vehicle type does not meet the initial vehicle-cargo matching;

[0029] S211: When the initial vehicle-cargo matching is met, the vehicle comprehensive status score obtained in S102 and the comprehensive transportation difficulty score obtained in S207 are input into the similarity calculation layer in the vehicle-cargo matching model to calculate the similarity coefficient. At the same time, the cargo attribute information and vehicle attribute information corresponding to the current cargo type are input into the matching layer in the vehicle-cargo matching model to obtain the initial matching score. The specific process of similarity coefficient calculation is as follows: ,in represents the vehicle comprehensive status score corresponding to the g-th vehicle under the f-th vehicle type, represents the comprehensive transportation difficulty score corresponding to the kth cargo type;

[0030] S212, inputting the obtained similarity coefficient and the initial matching score into the fully connected output layer of the vehicle-cargo matching model to obtain the vehicle-cargo matching score of the g-th vehicle under the f-th vehicle type and the k-th cargo type;

[0031] S213. When the initial vehicle-cargo matching is not satisfied, the cargo risk assessment score corresponding to the current cargo type is compared with the vehicle transport qualification score corresponding to the next vehicle in the vehicle transport information database, and S211 and S212 are repeated to obtain a vehicle-cargo matching score sequence corresponding to all vehicles that meet the initial vehicle-cargo matching under F vehicle types.

[0032] Specifically, the step of obtaining the margin of the vehicle corresponding to the current moment in S3 includes:

[0033] S301, comparing the vehicle-cargo matching score sequence obtained in S213 with the matching score threshold, when the vehicle-cargo matching score sequence is greater than the matching score threshold, obtaining the vehicle type that meets the threshold at the current moment and the number of vehicles corresponding to the vehicle type, and obtaining the number of vacant vehicles corresponding to the vehicles that meet the matching score threshold in the f-th vehicle type at the current moment according to the vacancy information of the corresponding type of vehicles obtained at the current moment;

[0034] S302, obtaining the vehicle demand of the fth vehicle type corresponding to the current cargo type according to the load attribute corresponding to the fth vehicle type that meets the matching score threshold and the weight of the current cargo type;

[0035] S303. Obtain the availability of the fth vehicle type according to the number of vacant vehicles corresponding to the vehicles of the fth vehicle type that meet the matching score threshold at the current moment and the vehicle demand of the fth vehicle type corresponding to the current cargo type.

[0036] Specifically, the steps of building a dynamic pricing model in S4 include:

[0037] S401. According to the cargo attribute information and cargo transportation information of the kth cargo type at the current moment, a vehicle pricing optimization function and constraint conditions are constructed at the current moment, so that the transaction rate of vehicles that meet the matching score threshold at the current moment meets the threshold condition and the total revenue is maximized. The specific formula is:

[0038] ,

[0039] ,

[0040] Among them, maxJ represents the vehicle pricing optimization function at the current moment, max represents the maximum value, represents the transaction rate of the f-th vehicle type at the current moment, represents the vehicle pricing of the f-th vehicle type at the current time t, represents the demand for the fth vehicle type required to transport the kth type of cargo, , represent the lower and upper limits of the pricing range for the f-th vehicle type, respectively. represents the transaction rate threshold. When this condition is met, the price of the f-th vehicle type remains in the current state. Indicates the current total budget expenditure for transporting goods; represents the number of vacant vehicles corresponding to the vehicles of the f-th vehicle type that meet the matching score threshold at the current moment, represents the vehicle adequacy of the f-th vehicle type. It means that the vehicles of the fth vehicle type are in sufficient condition. It indicates that the vehicles of the fth vehicle type are in shortage; Indicates the length of transportation time for the kth type of goods.

[0041] Specifically, the steps of constructing the dynamic pricing model in S4 also include:

[0042] S402: construct the current input state according to the acquired vehicle abundance, comprehensive transportation difficulty score, vehicle attribute information and cargo transportation information ;

[0043] S403: Construct the current moment pricing strategy execution action according to the current moment input state ,in Indicates when and When executing action until until, Indicates when and When the execution is reduced action until until, Indicates when Execution promotion action until until;

[0044] S404, Utilization [ p f min ， p f max ] Set price search space , and set the expected value of the vehicle pricing optimization function at the current moment as the reward function;

[0045] S405: Build a dynamic pricing model based on the deep policy gradient algorithm, and use the built current moment input state, current moment pricing strategy execution action and set price search space The reward function is input into the dynamic pricing model for training, and the optimal transportation price of the f-th vehicle type obtained from the price search space at the current moment is output.

[0046] The vehicle-cargo matching and pricing optimization system for logistics scheduling includes: vehicle information database module, matching module, and pricing module;

[0047] The vehicle information database module is used to obtain vehicle attribute information, current vehicle vacancy information and transportation qualification information, build a vehicle transportation information database through natural language algorithms and build a vehicle-cargo matching model;

[0048] The matching module includes an evaluation unit, a vehicle-cargo matching unit and a sufficiency calculation unit; the evaluation unit constructs an evaluation model based on the acquired cargo attribute information and cargo transportation information to obtain a comprehensive transportation difficulty score; the vehicle-cargo matching unit obtains a vehicle-cargo matching score based on the acquired transportation difficulty score and cargo attribute information through the vehicle-cargo matching model built into the vehicle transport information database, and obtains a transport vehicle that meets the threshold through a set matching score threshold; the sufficiency calculation unit obtains the number of vehicles that meet the threshold at the current moment, and obtains the sufficiency of the corresponding vehicle at the current moment based on the number of vehicles that meet the threshold at the current moment and the vehicle demand;

[0049] The pricing module uses reinforcement learning to build a dynamic pricing model, and obtains the current cargo transportation price by obtaining the abundance of vehicles that meet the threshold conditions, transportation time, transportation difficulty score and pricing strategy model.

[0050] A computer-readable storage medium stores computer instructions, which, when executed, execute a vehicle-cargo matching and pricing optimization method for logistics scheduling.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] In view of the deficiencies in the prior art, the present invention constructs a vehicle transportation information database, a vehicle-cargo matching model and obtains a comprehensive transportation difficulty score through an evaluation algorithm, thereby effectively screening out qualified vehicles, solving the problem of mismatch between vehicle and cargo transportation qualifications in the prior art, and improving the accuracy and efficiency of matching. Secondly, by setting a matching score threshold, calculating vehicle redundancy, and combining transportation time and transportation difficulty score, a dynamic pricing model is constructed using reinforcement learning, thereby realizing dynamic optimization of transportation prices, improving logistics scheduling efficiency and maximizing logistics benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of a method for optimizing vehicle-cargo matching and pricing for logistics scheduling according to Embodiment 1 of the present invention;

[0054] Figure 2 This is a flow chart of obtaining a comprehensive transportation difficulty score in Example 1 of the present invention;

[0055] Figure 3 This is a module diagram of a vehicle-cargo matching and pricing optimization system for logistics scheduling implemented in 2 of the present invention. DETAILED DESCRIPTION

[0056] Example 1

[0057] See also Figure 1 , an embodiment provided by the present invention: a method for optimizing vehicle-cargo matching and pricing for logistics scheduling, comprising the following steps:

[0058] S1: Acquire vehicle attribute information, current vehicle vacancy information, transportation qualification information and historical cargo transportation information, construct a vehicle transportation information database through a natural language algorithm, and embed a vehicle-cargo matching model in the constructed vehicle transportation information database; further, the specific steps of constructing the vehicle transportation information database in this embodiment include:

[0059] S101. Build an information extraction model through the Chinese pre-trained Bert model, and input the acquired vehicle attribute information, current vehicle vacancy information, transportation qualification information and historical cargo transportation information into the information extraction model to obtain the vehicle information triplet of the g-th vehicle under the f-th vehicle type; further, in this embodiment, the vehicle attribute information includes vehicle type: van, flatbed truck, refrigerated truck, etc.; vehicle model, vehicle capacity, vehicle size, vehicle status, vehicle location, vehicle age, vehicle insurance information, vehicle performance parameters and vehicle safety rating; transportation qualification information includes dangerous goods transportation permit, refrigerated transportation permit, special cargo transportation permit and insurance coverage;

[0060] S102, extracting the transport qualification and historical cargo transport triplet information from the vehicle information triplet, and inputting the obtained transport qualification and historical cargo transport triplet into the comprehensive evaluation algorithm to obtain the vehicle transport qualification score, and at the same time inputting the vehicle information triplet corresponding to the vehicle attribute information into the comprehensive evaluation algorithm to obtain the vehicle comprehensive status score;

[0061] S103, updating the obtained vehicle transport qualification score and vehicle comprehensive status score into the corresponding vehicle information triplet with the license plate number in the vehicle attribute information as the connection attribute information for storage;

[0062] S104: Save the acquired vehicle information triples into a graph database to obtain a vehicle transport information database.

[0063] S2: Obtain cargo attribute information and cargo transportation information, build an evaluation model through an evaluation algorithm, input the acquired cargo attribute information and cargo transportation information into the evaluation model for training, obtain a comprehensive transportation difficulty score, input the acquired comprehensive transportation difficulty score and cargo attribute information into the vehicle-cargo matching model built into the vehicle transportation information database, and obtain a vehicle-cargo matching score sequence;

[0064] For further information, see Figure 2 In this embodiment, the specific steps of obtaining the comprehensive transportation difficulty score in S2 include:

[0065] S201. Use cargo attribute information to construct a cargo risk assessment factor corresponding to the current cargo type And evaluation index sequence ; And use the expert experience evaluation method to evaluate each evaluation factor in the cargo risk assessment factor in turn according to the evaluation index sequence, and obtain the fuzzy evaluation matrix R corresponding to the cargo risk assessment factor; Indicates the nth cargo hazard assessment factor, represents the mth evaluation index; N represents the total number of cargo risk assessment factors, and M represents the total number of evaluation indicators;

[0066] Further, in this embodiment, the evaluation index sequence and the corresponding scoring interval include: very low, corresponding scoring interval: 1-2; low, corresponding scoring interval: 2-4; medium, corresponding scoring interval: 4-6; high, corresponding scoring interval: 6-8; very high, corresponding scoring interval: 8-10; the evaluation factors of the degree of danger of goods include physical properties, packaging safety, operating conditions and laws and regulations; physical properties include flammability, toxicity, corrosiveness, radioactivity, explosiveness and self-reactivity; packaging safety includes packaging material compatibility, sealing and compressive strength; operating conditions include temperature sensitivity, pressure sensitivity and light sensitivity; laws and regulations include transportation restrictions and licensing requirements;

[0067] S202: According to the obtained fuzzy evaluation matrix R, the weighted coefficient is obtained by the minimum membership weighted average deviation. ,in Indicates the weighting coefficient corresponding to the nth cargo risk assessment factor;

[0068] S203, will be constructed , and Input into the comprehensive fuzzy evaluation algorithm to obtain the cargo danger level assessment score corresponding to the current cargo type;

[0069] S204. Construct a transportation difficulty assessment factor using cargo transportation information, and input the constructed transportation difficulty assessment factor into the PCA algorithm to obtain the contribution of the transportation difficulty assessment factor to the transportation difficulty score; further, in this embodiment, the transportation difficulty assessment factor includes physical properties: cargo weight, cargo volume, special loading requirements; logistics network: route length, traffic conditions, terrain complexity and weather impact; time sensitivity: delivery deadline and spoilage speed;

[0070] S205, setting a contribution threshold, and obtaining a valid transport difficulty assessment factor sequence whose contribution is greater than the contribution threshold through the contribution threshold and the contribution corresponding to the obtained transport difficulty assessment factor; further, in this embodiment, the contribution threshold is 0.8;

[0071] S206. Constructing a transport difficulty regression model through a support vector machine, and inputting the obtained effective transport difficulty assessment factor sequence into the transport difficulty regression model to obtain a transport difficulty score;

[0072] S207. Multiply the cargo risk assessment score corresponding to the current cargo type by the transportation difficulty score to obtain a comprehensive transportation difficulty score corresponding to the current cargo type.

[0073] This process constructed a cargo hazard level assessment factor based on cargo attribute information, and used the expert experience evaluation method to obtain the fuzzy evaluation matrix R, and then calculated the weighting coefficient to ensure the reliability of the hazard level assessment score; secondly, the transportation difficulty regression model constructed by support vector machine was used to realize the quantitative assessment of transportation difficulty, and the cargo hazard level assessment score was combined with the transportation difficulty score to form a comprehensive transportation difficulty score. The comprehensive transportation difficulty score not only takes into account the characteristics of the cargo itself, but also comprehensively considers factors in multiple dimensions such as logistics network and time sensitivity, which greatly improves the matching accuracy of vehicles and cargoes, avoids the waste of scheduling resources and time due to inaccurate matching of vehicles and cargoes, and improves the efficiency of logistics scheduling.

[0074] Furthermore, in this embodiment, the step of obtaining the vehicle-cargo matching score sequence in S2 includes:

[0075] S210: Score the vehicle transport qualification stored in the graph database Compare it with the cargo hazard assessment score corresponding to the current cargo type. If the vehicle transportation qualification score is If the vehicle transport qualification score is greater than the cargo hazard level assessment score corresponding to the current cargo type, then the g-th vehicle under the f-th vehicle type meets the initial vehicle-cargo matching; if the vehicle transport qualification score is less than or equal to the cargo hazard level assessment score corresponding to the current cargo type, then the g-th vehicle under the f-th vehicle type does not meet the initial vehicle-cargo matching;

[0076] S211: When the initial vehicle-cargo matching is met, the vehicle comprehensive status score obtained in S102 and the comprehensive transportation difficulty score obtained in S207 are input into the similarity calculation layer in the vehicle-cargo matching model to calculate the similarity coefficient. At the same time, the cargo attribute information and vehicle attribute information corresponding to the current cargo type are input into the matching layer in the vehicle-cargo matching model to obtain the initial matching score. The specific process of similarity coefficient calculation is as follows: ,in represents the vehicle comprehensive status score corresponding to the g-th vehicle under the f-th vehicle type, represents the comprehensive transportation difficulty score corresponding to the kth cargo type;

[0077] S212, inputting the obtained similarity coefficient and the initial matching score into the fully connected output layer of the vehicle-cargo matching model to obtain the vehicle-cargo matching score of the g-th vehicle under the f-th vehicle type and the k-th cargo type;

[0078] S213. When the initial vehicle-cargo matching is not satisfied, the cargo risk assessment score corresponding to the current cargo type is compared with the vehicle transportation qualification score corresponding to the next vehicle in the vehicle transport information database, and S211 and S212 are repeated to obtain a vehicle-cargo matching score sequence corresponding to all vehicles that satisfy the initial vehicle-cargo matching under F vehicle types;

[0079] The process first ensures through preliminary matching judgment that the vehicle has sufficient transportation qualifications to cope with the dangerousness of the goods, thereby avoiding the participation of inappropriate vehicles in transportation and reducing potential safety risks. Secondly, by calculating the similarity coefficient and the initial matching score, and combining the information on the comprehensive status of the vehicle and the difficulty of cargo transportation, further precise matching is performed to ensure a high degree of match between vehicle performance and cargo demand during transportation, thereby improving the overall quality of transportation. Finally, for vehicles that fail the initial matching, qualified replacement vehicles are automatically searched to ensure that the goods can be transported in a timely and efficient manner, reducing delays and losses caused by improper vehicle selection.

[0080] S3: Set a matching score threshold, obtain the number of vehicles that meet the threshold at the current moment according to the vehicle-cargo matching score and the matching score threshold, and obtain the corresponding vehicle abundance at the current moment according to the number of vehicles that meet the threshold at the current moment and the vehicle demand corresponding to the current cargo;

[0081] Furthermore, in this embodiment, the step of obtaining the margin of the vehicle corresponding to the current moment in S3 includes:

[0082] S301, comparing the vehicle-cargo matching score sequence obtained in S213 with the matching score threshold, when the vehicle-cargo matching score sequence is greater than the matching score threshold, obtaining the vehicle type that meets the threshold at the current moment and the number of vehicles corresponding to the vehicle type, and obtaining the number of vacant vehicles corresponding to the vehicles that meet the matching score threshold in the f-th vehicle type at the current moment according to the vacancy information of the corresponding type of vehicles obtained at the current moment;

[0083] S302, according to the load attribute corresponding to the f-th vehicle type that meets the matching score threshold and the weight of the current cargo type, obtain the vehicle demand of the f-th vehicle type corresponding to the current cargo type; in this embodiment, the weight of the current cargo type is the total weight parameter corresponding to the cargo type that needs to be transported at the current moment;

[0084] S303, obtaining the availability of the fth vehicle type according to the number of vacant vehicles corresponding to the vehicles meeting the matching score threshold among the fth vehicle type at the current moment and the vehicle demand of the fth vehicle type corresponding to the current cargo type;

[0085] By comparing the vehicle-cargo matching score sequence with the matching score threshold, the process can quickly screen out vehicles that meet specific standards, ensuring the safety and efficiency of the transportation process. At the same time, according to the vacancy status of the vehicles and the weight requirements of the cargo, the relationship between the actual demand and the available number of each vehicle type, that is, the vehicle redundancy, is calculated. This process not only helps to improve vehicle utilization and reduce the empty driving rate, but also ensures that the cargo is transported on time and on demand, avoiding transportation delays or cost increases caused by insufficient vehicles.

[0086] S4: A dynamic pricing model is constructed through reinforcement learning, and the obtained vehicle abundance, transportation time, and transportation difficulty scores that meet the threshold conditions are input into the pricing strategy model to obtain the optimal transportation price of the current cargo corresponding to the transportation vehicle; further, the steps of constructing the dynamic pricing model in S4 in this embodiment include:

[0087] S401. According to the cargo attribute information and cargo transportation information of the kth cargo type at the current moment, a vehicle pricing optimization function and constraint conditions are constructed at the current moment, so that the transaction rate of vehicles that meet the matching score threshold at the current moment meets the threshold condition and the total revenue is maximized. The specific formula is:

[0088] ,

[0089] ,

[0090] Among them, maxJ represents the vehicle pricing optimization function at the current moment, max represents the maximum value, represents the vehicle transaction rate of the f-th vehicle type at the current moment, represents the vehicle pricing of the f-th vehicle type at the current time t, represents the demand for the fth vehicle type required to transport the kth type of cargo, , represent the lower and upper limits of the pricing range for the f-th vehicle type, respectively. represents the transaction rate threshold. When this condition is met, the price of the f-th vehicle type remains in the current state. Indicates the current total budget expenditure for transporting goods; Indicates the number of vacant vehicles corresponding to the vehicles of the f-th vehicle type that meet the matching score threshold at the current moment; represents the vehicle adequacy of the f-th vehicle type. It means that the vehicles of the fth vehicle type are in sufficient condition. It indicates that the vehicles of the fth vehicle type are in shortage; represents the length of transportation time for the kth type of cargo;

[0091] Furthermore, the first constraint Constraints on pricing, the second constraint Constrain the transaction rate so that the pricing adjustment is carried out under the condition of satisfying the transaction rate; the third constraint Constrain the current logistics costs, the fourth constraint Constrain the vehicle's abundance; the vehicle pricing optimization function is The inverse of is multiplied by the transaction rate, pricing, demand and abundance of the f-th vehicle type to obtain the current logistics revenue, and the logistics revenue is compared with the total budget cost of transporting goods to obtain the revenue-budget ratio, and the revenue-budget ratio is maximized so that the logistics price of a single type is optimal under the condition of satisfying the transaction rate, so that the logistics revenue is maximized;

[0092] S402: construct the current input state according to the acquired vehicle abundance, comprehensive transportation difficulty score, vehicle attribute information and cargo transportation information ;

[0093] S403: Construct the current moment pricing strategy execution action according to the current moment input state ,in Indicates when and When executing action until until, Indicates when and When the execution is reduced action until until, Indicates when Execution promotion action until until;

[0094] S404, Utilization [ p f min ， p f max ] Set price search space , and set the expected value of the vehicle pricing optimization function at the current moment as the reward function;

[0095] S405: Build a dynamic pricing model based on the deep policy gradient algorithm, and use the built current moment input state, current moment pricing strategy execution action and set price search space The reward function is input into the dynamic pricing model for training, and the optimal transportation price of the f-th vehicle type obtained from the price search space at the current moment is output;

[0096] The process constructs a vehicle pricing optimization function and constraints based on cargo attributes and transportation information to ensure that the total revenue is maximized while meeting the matching score and transaction rate thresholds. Secondly, the vehicle abundance, transportation difficulty score, and cargo and vehicle attribute information are used as input states to construct a pricing strategy execution action, which is used together with the expected value of the pricing optimization function as a reward function to train the dynamic pricing model. Finally, the trained dynamic pricing model is used to output the optimal transportation price. This method can not only automatically adjust prices to adapt to changing market conditions, but also take into account factors such as vehicle abundance, transportation difficulty, and transportation time, thereby improving logistics matching and logistics scheduling speed. In addition, the process can ensure that pricing is within a reasonable range and avoid exceeding the budget or vehicle shortage.

[0097] S5: Set a transaction rate threshold, obtain the vehicle transaction rate in real time according to the current transportation price, and feed the vehicle transaction rate back to the dynamic pricing model to adjust the price at the current moment so that the vehicle transaction rate meets the transaction rate threshold in real time.

[0098] Example 2

[0099] See also Figure 3 , another embodiment provided by the present invention: a vehicle-cargo matching and pricing optimization system for logistics scheduling, comprising a vehicle information base module, a matching module, a pricing module and a feedback optimization module;

[0100] The vehicle information database module is used to obtain vehicle attribute information, current vehicle vacancy information and transportation qualification information, build a vehicle transportation information database through a natural language algorithm and build a vehicle-cargo matching model;

[0101] The matching module is used for the management of cargo information and the real-time matching of cargo and vehicles; the matching module includes an evaluation unit, a vehicle-cargo matching unit and a sufficiency calculation unit; the evaluation unit constructs an evaluation model based on the acquired cargo attribute information and cargo transportation information to obtain a comprehensive transportation difficulty score; the vehicle-cargo matching unit obtains a vehicle-cargo matching score based on the acquired transportation difficulty score and cargo attribute information through a vehicle-cargo matching model built into a vehicle transport information database, and obtains a transport vehicle that meets the threshold through a set matching score threshold; the sufficiency calculation unit obtains the number of vehicles that meet the threshold at the current moment, and obtains the sufficiency of the corresponding vehicle at the current moment based on the number of vehicles that meet the threshold at the current moment and the vehicle demand;

[0102] The pricing module uses reinforcement learning to build a dynamic pricing model, and obtains the current freight transportation price through the abundance of vehicles that meet threshold conditions, transportation time, transportation difficulty score and pricing strategy model;

[0103] The feedback optimization module is used to monitor the transaction rate in real time, and to make feedback adjustments to the current freight transportation price based on the acquired transaction rate and the set threshold.

[0104] Example 3

[0105] A computer-readable storage medium stores computer instructions, which, when executed, execute a vehicle-cargo matching and pricing optimization method for logistics scheduling.

[0106] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements a vehicle-cargo matching and pricing optimization method for logistics scheduling when executing the computer program.

[0107] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the scope of protection of the purpose of the present invention and the claims, and all of these are within the protection of the present invention.

[0108] If the disclosed technical solution involves personal information, the product using the disclosed technical solution has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the disclosed technical solution involves sensitive personal information, the product using the disclosed technical solution has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. A vehicle-cargo matching and pricing optimization method for logistics scheduling, characterized in that: include: S1: Obtain vehicle attribute information, current vehicle vacancy information, transportation qualification information and historical cargo transportation information, build a vehicle transportation information database through natural language algorithm, and build a vehicle-cargo matching model in the constructed vehicle transportation information database; S2: Obtain cargo attribute information and cargo transportation information, build an evaluation model through an evaluation algorithm, input the acquired cargo attribute information and cargo transportation information into the evaluation model for training, obtain a comprehensive transportation difficulty score, input the acquired comprehensive transportation difficulty score and cargo attribute information into the vehicle-cargo matching model built into the vehicle transportation information database, and obtain a vehicle-cargo matching score sequence; S3: Set a matching score threshold, obtain the number of vehicles that meet the threshold at the current moment according to the vehicle-cargo matching score and the matching score threshold, and obtain the corresponding vehicle abundance at the current moment according to the number of vehicles that meet the threshold at the current moment and the vehicle demand corresponding to the current cargo; S4: Build a dynamic pricing model through reinforcement learning, input the obtained vehicle abundance, transportation time, and transportation difficulty score that meet the threshold conditions into the pricing strategy model, and obtain the optimal transportation price for the current cargo corresponding to the transportation vehicle; S5: Set a transaction rate threshold, obtain the vehicle transaction rate in real time according to the current transportation price, and feed the vehicle transaction rate into the dynamic pricing model to adjust the price at the current moment so that the vehicle transaction rate meets the transaction rate threshold in real time; The step of constructing the dynamic pricing model in S4 also includes: S402, constructing a current moment input state according to the acquired vehicle abundance, comprehensive transportation difficulty score, vehicle attribute information and cargo transportation information; S403, constructing a current moment pricing strategy execution action according to the current moment input state; S404, setting a price search space, and setting the expected value of the vehicle pricing optimization function at the current moment as a reward function; S405. Construct a dynamic pricing model based on the deep policy gradient algorithm, and input the constructed current input state, the current pricing policy execution action, and the set price search space and reward function into the dynamic pricing model for training, and output the optimal transportation price of the f-th vehicle type obtained from the price search space at the current moment.

2. The vehicle-cargo matching and pricing optimization method for logistics scheduling according to claim 1, characterized in that: The specific steps of constructing the vehicle transport information database include: S101, constructing an information extraction model through a Chinese pre-trained Bert model, and inputting the acquired vehicle attribute information, current vehicle vacancy information, transportation qualification information and historical cargo transportation information into the information extraction model to obtain a vehicle information triplet; S102, extracting the transport qualification and historical cargo transport triplet information from the vehicle information triplet, and inputting the obtained transport qualification and historical cargo transport triplet into the comprehensive evaluation algorithm to obtain the vehicle transport qualification score, and at the same time inputting the vehicle information triplet corresponding to the vehicle attribute information into the comprehensive evaluation algorithm to obtain the vehicle comprehensive status score; S103, updating the obtained vehicle transport qualification score and vehicle comprehensive status score into the corresponding vehicle information triplet with the license plate number in the vehicle attribute information as the connection attribute information for storage; S104: Save the acquired vehicle information triples into a graph database to obtain a vehicle transport information database.

3. The vehicle-cargo matching and pricing optimization method for logistics scheduling according to claim 2, characterized in that: The specific steps of obtaining the comprehensive transportation difficulty score in S2 include: S201. Use cargo attribute information to construct a cargo risk assessment factor corresponding to the current cargo type And evaluation index sequence , and evaluate each evaluation factor in the cargo risk assessment factor in turn according to the evaluation index sequence to obtain the fuzzy evaluation matrix R corresponding to the cargo risk assessment factor; Indicates the nth cargo hazard assessment factor, represents the mth evaluation index; N represents the total number of cargo risk assessment factors, and M represents the total number of evaluation indicators; S202: According to the obtained fuzzy evaluation matrix R, the weighted coefficient is obtained by the minimum membership weighted average deviation. ,in Indicates the weighting coefficient corresponding to the nth cargo risk assessment factor; S203, will be constructed , and Input into the comprehensive fuzzy evaluation algorithm to obtain the cargo hazard level assessment score corresponding to the current cargo type.

4. The vehicle-cargo matching and pricing optimization method for logistics scheduling according to claim 3, characterized in that: The specific steps of obtaining the comprehensive transportation difficulty score in S2 also include: S204, constructing a transportation difficulty assessment factor using the cargo transportation information, and inputting the constructed transportation difficulty assessment factor into the PCA algorithm to obtain the contribution of the transportation difficulty assessment factor to the transportation difficulty score; S205, setting a contribution threshold, and obtaining a valid transport difficulty assessment factor sequence whose contribution is greater than the contribution threshold through the contribution threshold and the contribution corresponding to the obtained transport difficulty assessment factor; S206. Constructing a transport difficulty regression model through a support vector machine, and inputting the obtained effective transport difficulty assessment factor sequence into the transport difficulty regression model to obtain a transport difficulty score; S207. Multiply the cargo risk assessment score corresponding to the current cargo type by the transportation difficulty score to obtain a comprehensive transportation difficulty score corresponding to the current cargo type.

5. The vehicle-cargo matching and pricing optimization method for logistics scheduling according to claim 4, characterized in that: The step of obtaining the vehicle-cargo matching score sequence in S2 includes: S210, compare the vehicle transport qualification score stored in the graph database with the cargo risk assessment score corresponding to the current cargo type. If the vehicle transport qualification score is greater than the cargo risk assessment score corresponding to the current cargo type, the g-th vehicle under the f-th vehicle type meets the initial vehicle-cargo matching. If the vehicle transport qualification score is less than or equal to the cargo risk assessment score corresponding to the current cargo type, the g-th vehicle under the f-th vehicle type does not meet the initial vehicle-cargo matching. S211: When the initial vehicle-cargo matching is met, the vehicle comprehensive status score obtained in S102 and the comprehensive transportation difficulty score obtained in S207 are input into the similarity calculation layer in the vehicle-cargo matching model to calculate the similarity coefficient. At the same time, the cargo attribute information and vehicle attribute information corresponding to the current cargo type are input into the matching layer in the vehicle-cargo matching model to obtain the initial matching score. The specific process of similarity coefficient calculation is as follows: ,in represents the vehicle comprehensive status score corresponding to the g-th vehicle under the f-th vehicle type, represents the comprehensive transportation difficulty score corresponding to the kth cargo type; S212, inputting the obtained similarity coefficient and the initial matching score into the fully connected output layer of the vehicle-cargo matching model to obtain the vehicle-cargo matching score of the g-th vehicle under the f-th vehicle type and the k-th cargo type; S213. When the initial vehicle-cargo matching is not satisfied, the cargo risk assessment score corresponding to the current cargo type is compared with the vehicle transport qualification score corresponding to the next vehicle in the vehicle transport information database, and S211 and S212 are repeated to obtain a vehicle-cargo matching score sequence corresponding to all vehicles that meet the initial vehicle-cargo matching under F vehicle types.

6. The vehicle-cargo matching and pricing optimization method for logistics scheduling according to claim 5, characterized in that: The step of obtaining the margin of the vehicle at the current moment in S3 includes: S301, comparing the vehicle-cargo matching score sequence obtained in S213 with the matching score threshold, when the vehicle-cargo matching score sequence is greater than the matching score threshold, obtaining the vehicle type that meets the threshold at the current moment and the number of vehicles corresponding to the vehicle type, and obtaining the number of vacant vehicles corresponding to the vehicles that meet the matching score threshold in the f-th vehicle type at the current moment according to the vacancy information of the corresponding type of vehicles obtained at the current moment; S302, obtaining the vehicle demand of the fth vehicle type corresponding to the current cargo type according to the load attribute corresponding to the fth vehicle type that meets the matching score threshold and the weight of the current cargo type; S303. Obtain the availability of the fth vehicle type according to the number of vacant vehicles corresponding to the vehicles of the fth vehicle type that meet the matching score threshold at the current moment and the vehicle demand of the fth vehicle type corresponding to the current cargo type.

7. The vehicle-cargo matching and pricing optimization method for logistics scheduling according to claim 6, characterized in that: The steps of constructing the dynamic pricing model in S4 include: S401. According to the cargo attribute information and cargo transportation information of the kth cargo type at the current moment, a vehicle pricing optimization function and constraint conditions are constructed at the current moment, so that the transaction rate of vehicles that meet the matching score threshold at the current moment meets the threshold condition and the total revenue is maximized. The specific formula is: ， ， Among them, maxJ represents the vehicle pricing optimization function at the current moment, max represents the maximum value, represents the vehicle transaction rate of the f-th vehicle type at the current moment, represents the vehicle pricing of the f-th vehicle type at the current time t, represents the demand for the fth vehicle type required to transport the kth type of cargo, , represent the lower and upper limits of the pricing range for the f-th vehicle type, respectively. represents the transaction rate threshold. When this condition is met, the price of the f-th vehicle type remains in the current state. Indicates the current total budget cost of transporting goods; Indicates the number of vacant vehicles corresponding to the vehicles of the f-th vehicle type that meet the matching score threshold at the current moment; represents the vehicle adequacy of the f-th vehicle type. It means that the vehicles of the fth vehicle type are in sufficient condition. It indicates that the vehicles of the fth vehicle type are in shortage; Indicates the length of transportation time for the kth type of goods.

8. A vehicle-cargo matching and pricing optimization system for logistics scheduling, which is used to implement the vehicle-cargo matching and pricing optimization method for logistics scheduling as described in any one of claims 1 to 7, characterized in that: include: Vehicle information base module, matching module, pricing module; The vehicle information database module is used to obtain vehicle attribute information, current vehicle vacancy information and transportation qualification information, build a vehicle transportation information database through a natural language algorithm and build a vehicle-cargo matching model; The matching module includes an evaluation unit, a vehicle-cargo matching unit and a sufficiency calculation unit; the evaluation unit constructs an evaluation model based on the acquired cargo attribute information and cargo transportation information to obtain a comprehensive transportation difficulty score; The vehicle-cargo matching unit obtains a vehicle-cargo matching score according to the obtained transport difficulty score and cargo attribute information through a vehicle-cargo matching model built into a vehicle transport information database, and obtains a transport vehicle that meets the threshold through a set matching score threshold; The sufficiency calculation unit obtains the number of vehicles that meet the threshold at the current moment, and obtains the sufficiency of the corresponding vehicles at the current moment according to the number of vehicles that meet the threshold at the current moment and the vehicle demand; The pricing module uses reinforcement learning to build a dynamic pricing model, and obtains the current cargo transportation price through the abundance of vehicles that meet threshold conditions, transportation time, transportation difficulty score and pricing strategy model.

9. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the computer instructions are executed, the vehicle-cargo matching and pricing optimization method for logistics scheduling described in any one of claims 1-7 is executed.

Images