Bulk cargo revolution iron intelligent decision-making method and system based on causal inference
Through the principle of causal inference and the causal inference model based on the discount leverage mechanism, the problem of failure to effectively identify the causality of railway transportation costs in the existing technology is solved, and the efficiency and accuracy of the formulation of the "road to rail" transportation strategy is improved.
Patent Information
- Application Number
- CN202510336471.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-21
- Publication Date
- 2025-04-25
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing intelligent decision-making method for bulk goods ‘broad to rail’ has failed to effectively identify the causality of various factors in railway transportation costs, resulting in the lack of theoretical interpretability and objective analysis deviation of model deduction.
The principle of causal inference is adopted, and based on the historical operation data of bulk cargo transportation, a causal inference model based on the discount leverage mechanism is constructed. The most sensitive factors affecting railway transportation costs are tracked through the inverse probability weighting method, and the complexity of the intelligent decision-making problem of "road to rail" is simplified.
It improves the efficiency of the formulation of the ‘road to rail’ transportation strategy, provides a more accurate analysis of railway transportation costs, eliminates the selective bias in the correlation analysis, and ensures the theoretical interpretability of decisions.
Smart Images
Figure CN119887017A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of data processing systems or methods for prediction purposes, and specifically, relates to an intelligent decision-making method and system for "road to rail" of bulk cargo based on causal inference. Background Art
[0002] In order to comprehensively promote the construction of a comprehensive transportation service pattern of "railway where appropriate, road where appropriate", transportation modes such as "road to rail", road-rail combined transport, and river-sea combined transport have become measures for freight structure adjustment and reform in many places.
[0003] At present, the intelligent decision-making of "road to rail" for bulk cargo mostly adopts the prior knowledge of the technical advantages of the transportation mode, uses correlation analysis modeling and analysis methods, analyzes the factors affecting the total freight cost, and then formulates the implementation strategy of "road to rail" in combination with the actual conditions of facilities and resources. The published papers "Research on the passenger flow sharing rate model and its application of high-speed passenger dedicated line" (Journal of the China Railway Society, 2006, 28(3): 18-21), "Research on the calculation method of freight flow sharing rate of high-speed rail express products" (Transportation Systems Engineering and Information, 2019, 19(1): 40-44)", "Research on the countermeasures of road to rail for bulk cargo flow under low-carbon transportation" (China Market, 2022, (03): 151-152), "Research on the comprehensive optimization of bulk cargo flow transfer and flow distribution between road and rail networks under low-carbon transportation" (Chen Lei, 2016), etc., have established quantitative analysis models to calculate the freight flow sharing rate of different modes from the perspective of planning and prediction. , considering the different impacts of factors such as transportation costs and freight time on the total transportation costs of roads and railways, through the sensitivity analysis of various influencing factors, screening sensitive factors and then formulating a "road to rail" strategy. Similar reports include "Research on Strategies to Increase the Market Share of Bulk Freight Transportation by Railways under the Situation of Road to Rail" (Railway Economic Research, 2021, (02): 31-36.). However, the above quantitative analysis model is based on the correlation between various factors and the total freight cost, and only screens and sorts the sensitivity of the impact, without distinguishing the causality between the two. As a result, the railway transportation costs deduced by the model lack theoretical interpretability and have objective research and judgment bias. Summary of the invention
[0004] Different from the above research methods, the present invention adopts the principle of causal inference, and based on the historical operation data of bulk cargo transportation, proposes a bulk cargo "road to rail" intelligent decision-making method based on causal inference from the perspective of railway operators. The method judges the feasibility of rail transportation of goods. If the characteristics of the goods are compatible with rail transportation, the most sensitive factors affecting the railway transportation cost are tracked through the inverse probability weighted method of causal inference; a causal inference model based on a discount leverage mechanism is constructed, and the average treatment effect (ATE) in causal inference is introduced to simplify the complexity of the "road to rail" intelligent decision-making problem from 2n+m to n+m; based on the constructed causal inference model, quotations and discounts for railway transportation are given; after adopting this method, railway companies can complete the formulation of transportation strategies in a shorter time, thereby improving the efficiency of road-to-rail transportation strategy formulation.
[0005] To achieve the above object, the present invention adopts the following technical solutions: An intelligent decision-making method for bulk cargo "road to rail" based on causal inference, the method comprising the following steps: Step A: Determine whether the customer's freight demand is compatible with railway transportation. If so, proceed to step B. Step B. Calculate the average treatment effect of different influencing factors on railway transportation costs in historical waybill data by using the inverse probability weighted method of causal inference, and track the most sensitive factors affecting railway transportation costs; Step C. Construct a causal inference model based on the discount leverage mechanism, introduce the average treatment effect in causal inference, and simplify the complexity of the "road to rail" intelligent decision-making problem from 2n+m to n+m; The causal inference model divides the input data into proxy variables, latent variables, and treatment variables; the cargo type, cargo weight, and mileage in the customer's waybill are proxy variables, the cargo loading method is the treatment variable, and the average treatment effect of each influencing factor on the railway transportation cost obtained based on historical data, the most sensitive factor, and the number of loading tools affected by the cargo loading method are set as latent variables; the causal inference model calculates the current railway transportation cost based on each input data and determines whether it meets the If the conditions are met, the inference results are output; is the current railway transportation cost, which is related to the input data; The discount given by the railway company to realize the "road to rail" transfer of bulk goods; The cost of transporting goods using the road system; Step D: Input the data in the current waybill and provide the railway transportation plan, quotation and discount based on the constructed causal inference model.
[0006] As a preferred embodiment of the present invention, when the step A determines whether the customer's freight demand is compatible with railway transportation, it specifically includes the following steps: Step A1. Enter the customer's freight departure and destination, freight volume, cargo damage requirements, and freight timeliness; Step A2. Check the cargo item number to determine whether there is a requirement for reverse loading and damage. If the sender has a requirement for reverse loading and damage, the freight demand of the cargo is not compatible with railway transportation. If the sender does not have a requirement for reverse loading and damage, proceed to step A3. Step A3. According to the customer's freight departure and destination, determine whether the adjacent sending station and terminal station meet the railway company's loading and unloading conditions. If not, the freight demand of the goods is not compatible with railway transportation; if so, execute step A4; Step A4. According to the customer's freight departure and destination, determine whether there is a dedicated railway freight line between the departure and destination. If there is a dedicated line, it is adapted to railway transportation and step A5 is executed; if there is no dedicated freight line, determine whether there is a direct line between the adjacent sending station and the terminal. If there is no direct line, the freight demand of the goods is not compatible with railway transportation; if there is a direct line between the adjacent sending station and the terminal, it meets the railway transportation conditions and is adapted to railway transportation, and step A5 is executed; Step A5. According to the freight volume of the customer, determine whether it meets the minimum freight volume requirement of the railway company. If not, the freight demand of the goods is not compatible with railway transportation. If it is, execute step A6; Step A6. Determine whether the freight demand of the goods is compatible with rail transportation according to the customer's freight timeliness requirements. For goods that require delivery within 3 days, it is determined that they are not compatible with rail transportation; otherwise, they are compatible with rail transportation.
[0007] As a preferred embodiment of the present invention, the average therapeutic effect calculation formula of different influencing factors on railway transportation costs is: ; in, It is an influencing factor in historical waybills i Average treatment effect on railway transportation costs, influencing factors i Including cargo loading method, number of loading tools, mileage, cargo weight, cargo type; E is the expected symbol; is the propensity score matrix for railway transportation costs.
[0008] As a preferred embodiment of the present invention, the most sensitive factor affecting the railway transportation cost is determined by the following formula: ; In the formula, It is the most sensitive factor affecting railway transportation costs; is the average treatment effect of freight loading mode on rail transport costs; is the average treatment effect of the number of loading tools on rail transport costs; is the average treatment effect of travelled miles on rail transport costs; is the average treatment effect of freight weight on rail transport costs; is the average treatment effect of freight type on rail transport costs; Representative The maximum value in .
[0009] As a preferred embodiment of the present invention, the relationship between the current railway transportation cost and the average treatment effect in the historical waybill, the most sensitive factor and the current waybill data is: ; ; In the formula, is the current waybill railway transportation fee; log is the logarithmic sign; is the current cargo loading method, is the average treatment effect of freight loading mode on rail transport costs; is the number of loading tools for the current waybill, is the average treatment effect of the number of loading vehicles on rail transport costs; is the mileage of the current waybill, is the average treatment effect of travelled miles on rail transport costs; is the current waybill cargo weight, is the average treatment effect of freight weight on rail transport costs; is the type of goods in the current waybill, is the average treatment effect of freight type on rail transport costs; It is the most sensitive factor affecting railway transportation costs in historical waybills. yes The average treatment effect on railway transportation costs; E is the expected sign.
[0010] As preferred embodiment of the present invention, step D comprises the following steps: Step D1. Arrange the data related to the railway transportation costs of bulk goods in the current customer waybill, including: the cargo loading method, the number of loading tools, the mileage, the cargo weight, and the cargo type in the current waybill; Step D2. Based on the constructed causal inference model, provide a railway transportation plan, quotation and discount; Step D2.1. Change the cargo category in the current waybill 、The weight of the goods in the current waybill 、Mileage in the current waybill As a proxy variable, the cargo loading method in the current waybill ( , ) as the treatment variable; the most sensitive factor determined in step B, the average treatment effect of each influencing factor, and the effect of cargo loading mode ( , ) The number of loading tools affected As a latent variable, it is introduced into the causal inference model in step C; Step D2.2. Initialize the railway transport mode, ; Step D2.3. Calculate the current railway transportation cost and determine the difference between railway transportation cost and road transportation cost Is it greater than 0? ; In the formula, The difference between the cost of transporting goods by rail and by road; is the current railway transportation cost; The discount given by the railway company to realize the "road to rail" transfer of bulk goods; The cost of transporting goods using the road system; like , output the inference results, including: the final railway cargo loading method 、Number of loading tools in the current waybill , Railway transportation cost quotation , Discount Rate ; like , adjust the railway transportation mode , Discount Rate , repeat D2.3 until , to realize the formulation of the strategy of “road to rail” for bulk cargo.
[0011] As a further preferred embodiment of the present invention, the calculation formula of the propensity score of railway transportation cost is: ; In the formula, is the propensity score matrix of railway transportation costs, , , , , , All are coefficients; It is the cargo loading method matrix in the historical waybill; Load the tool quantity matrix in the historical waybill; is the mileage matrix in the historical waybills; It is the cargo weight matrix in the historical waybill; It is the cargo type matrix in the historical waybills.
[0012] The present invention also provides a bulk cargo "road to rail" intelligent decision-making system based on causal inference, the system comprising a data input module, an adaptation module, a data processing module, and a causal inference model; The data input module is used to input data, including the freight departure and destination, freight volume, cargo damage requirements, freight timeliness, cargo loading method, number of loading tools, mileage, cargo weight, cargo type, and base price per ton of cargo; The adaptation module is used to determine whether the freight demand is compatible with railway transportation based on the input freight data; The data processing module is used to collect statistics on the railway transportation costs of bulk goods in historical waybills, calculate the propensity score of the railway transportation costs, and calculate the average impact of different influencing factors on the railway transportation costs in the historical waybill data using the inverse probability weighting method based on the propensity score, and determine the most sensitive factor affecting the railway transportation costs; The causal inference model calculates the current railway transportation cost based on the input proxy variables, treatment variables, and latent variable data to determine whether If the conditions are met, the inference results are output; is the current railway transportation cost; The discount given by the railway company to realize the "road to rail" transfer of bulk goods; The cost of transporting goods using the road system; ; ; In the formula, is the current waybill railway transportation fee; log is the logarithmic sign; is the current cargo loading method, is the average treatment effect of freight loading mode on rail transport costs; is the number of loading tools for the current waybill, is the average treatment effect of the number of loading vehicles on rail transport costs; is the mileage of the current waybill, is the average treatment effect of travelled miles on rail transport costs; is the current waybill cargo weight, is the average treatment effect of freight weight on rail transport costs; is the type of goods in the current waybill, is the average treatment effect of freight type on rail transport costs; It is the most sensitive factor affecting railway transportation costs in historical waybills. yes The average treatment effect on railway transportation costs; E is the expected sign.
[0013] Advantages and beneficial effects of the present invention: (1) When comparing the total transportation costs of bulk cargo road and rail transport, identifying the correlation between each influencing factor and the freight costs of each mode is the basis for technical and economic comparison. Operators have not fully identified the causality between the influencing factors and the total transportation costs based on historical data statistical analysis or lean accounting based on operating costs, and simply regard the correlation as causality; if resources are invested in solutions based on incorrect causal relationships, it will lead to ineffective allocation and waste of resources. When analyzing the influencing factors of railway transportation costs, the present invention re-discusses the correlation and causality between each factor and railway transportation costs from the perspective of causal inference, and adopts the inverse probability weighting method of causal inference theory to eliminate the selective bias of each factor and railway transportation costs when using correlation analysis. It can accurately identify the degree of influence of each factor on railway transportation costs, thereby tracking the most sensitive factors affecting the railway transportation costs of bulk cargo, and providing a theoretical basis for the formulation of road-to-rail transportation strategies.
[0014] (2) In the optimization of the "road to rail" transportation strategy, existing methods all need to calculate the base price 1 (loading and unloading cost SUM1) and base price 2 (operating freight SUM2) for different transportation schemes (such as full truck transportation or container transportation). The complexity is 2n+m, where base price 1 is affected by loading and unloading efficiency, equipment investment and site layout. The calculation formula is: SUM1=C1+C2+C3+C4, where C1 is the labor cost, which is proportional to the number of loading and unloading workers and working hours; C2 is the equipment cost, which is related to the frequency of equipment use and maintenance costs; C3 is the site cost, which is related to the site area and usage time; C4 is the time cost, which is inversely proportional to the loading and unloading efficiency. At the same time, base price 2 is related to the transportation distance, energy cost and the total amount of container vehicles, resulting in repeated calculation of multi-variable relationships in strategy formulation, which is inefficient. The method provided by the present invention introduces the average treatment effect (ATE) in causal inference, and by uniformly modeling ATE, the difference between base price 1 and base price 2 in railway transportation cost is attributed to the causal effect of the transportation scheme on δ, thereby avoiding step-by-step calculation; the final result obtained by modeling is δ=SUM(FYJE)×γ-P0; wherein, δ is the difference between railway transportation cost and road system transportation cost; SUM(FYJE) is the railway transportation cost; γ is the discount given by the railway company to realize the "road to rail" of bulk goods; P0 is the road system transportation cost; the change of railway transportation cost caused by the base price 1 and the base price 2 is integrated through a unified modeling framework, the step-by-step calculation and repeated calculation of the traditional method are avoided, and the complexity of the intelligent decision-making problem of "road to rail" of bulk goods is reduced from 2n+m to n+m, so that the model is more concise, and the railway company can complete the formulation of the transportation strategy in a shorter time, thereby improving the formulation efficiency of the road to rail transportation strategy.
[0015] (3) After determining that the characteristics of the goods are compatible with railway transportation, the present invention constructs a causal inference model based on the discount leverage mechanism based on the most sensitive factors that have been tracked to affect the railway transportation costs, and a method to simplify the complexity of the "road to rail" intelligent decision-making problem. The causal inference model divides the input data into proxy variables, latent variables, and treatment variables, fully considering the inherent correlation between the variables, and gives the quotation and discount of railway transportation based on the above variables, ensuring that the final railway freight pricing can fully meet the needs of road to rail.
[0016] (4) The method provided by the present invention can adaptively provide the corresponding optimal transportation plan (provide personalized transportation plan) according to different customer needs, and provide transportation quotes and discounts to achieve the goal of "transferring road to rail" for bulk goods.
[0017] (5) Based on the big data of historical waybills of bulk cargo transportation, the present invention enhances the interpretability of data-driven intelligent decision-making technology through causal inference methods. The proposed intelligent decision-making method for "transferring bulk cargo from road to rail" can more accurately and effectively realize the shift from road transportation to rail transportation in bulk cargo transportation. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] By referring to the following description in conjunction with the accompanying drawings, and with a more comprehensive understanding of the present invention, other objects and results of the present invention will become more clear and easy to understand. In the accompanying drawings: Figure 1 A flow chart of the intelligent decision-making method for “road to rail” of bulk cargo based on causal inference provided by the present invention; Figure 2 A railway transport adaptation decision flow chart to determine the customer's cargo type and freight demand; Figure 3 A flow chart to track the most sensitive factors affecting rail transportation costs; Figure 4 This is a framework diagram of the causal inference model based on the discount lever mechanism; Figure 5 This is a flowchart of intelligent decision-making for the causal inference model based on the discount leverage mechanism; Figure 6 Schematic diagram of the intelligent decision-making system for “road to rail” of bulk cargo based on causal inference provided by the present invention. DETAILED DESCRIPTION
[0019] In order to enable those skilled in the art to better understand the technical solution and advantages of the present invention, the present application is described in detail below in conjunction with the accompanying drawings, but it is not intended to limit the protection scope of the present invention.
[0020] Embodiment 1:
[0021] like Figures 1 to 5 As shown, the intelligent decision-making method for bulk cargo "road to rail" based on causal inference provided in this embodiment includes the following steps: Step A: Determine whether the customer's freight demand is compatible with railway transportation. If so, proceed to step B. Step B. Calculate the average treatment effect (average impact degree) of different influencing factors on railway transportation costs in historical waybill data through the inverse probability weighted method of causal inference, track the most sensitive factors affecting railway transportation costs, and provide a theoretical basis for the formulation of road-to-rail transportation strategies; Step C. Construct a causal inference model based on the discount leverage mechanism, introduce the average treatment effect (ATE) in causal inference, and simplify the complexity of the "public to rail" intelligent decision-making problem from 2n+m to n+m; Step D. Input the data in the current waybill and, based on the constructed causal inference model, provide railway transportation plans, quotations, and discounts to achieve the “road to rail” transition for bulk cargo.
[0022] Specifically, Figure 2 As shown, in this embodiment, when the step A determines whether the customer's freight demand is compatible with railway transportation, it specifically includes the following steps: Step A1. Enter the customer's freight OD (origin, destination), freight volume, cargo damage requirements, and freight timeliness; Step A2. Check the cargo item number to determine whether there is a requirement for reverse loading and damage. If the sender has a requirement for reverse loading and damage, the freight demand of the cargo is not compatible with railway transportation. If the sender does not have a requirement for reverse loading and damage, proceed to step A3. Step A3. According to the customer's freight OD (place of departure, place of receipt), determine whether its adjacent sending station and final destination station (O'D') meet the railway company's loading and unloading conditions. If not, the freight demand of the goods is not compatible with railway transportation; if so, execute step A4; Step A4. According to the customer's freight OD (departure place, destination), determine whether there is a dedicated railway freight line between the departure place and the destination. If there is a dedicated line, it is compatible with railway transportation and executes step A5; if there is no dedicated freight line, determine whether there is a direct line between the adjacent sending station and the terminal (O'D'). If there is no direct line, the freight demand of the goods is not compatible with railway transportation; if there is a direct line between the adjacent (near) sending station and the terminal (in actual transportation, the train will depart from the adjacent departure station and be delivered to the freight station near the destination), it meets the railway transportation conditions and is compatible with railway transportation, and executes step A5; Step A5. According to the freight volume of the customer, determine whether it meets the minimum freight volume requirement of the railway company. If not, the freight demand of the goods is not compatible with railway transportation. If it is, execute step A6; Step A6. Determine whether the freight demand of the goods is compatible with rail transportation based on the customer's freight timeliness requirements. Goods with extremely high timeliness requirements (such as fresh food) and goods with high timeliness requirements (required to be delivered within 3 days) are not suitable for rail transportation; otherwise, they are suitable for rail transportation.
[0023] like Figure 3 As shown, in this embodiment, the specific steps of step B for tracking the most sensitive factors affecting railway transportation costs by the inverse probability weighted method of causal inference are as follows: Step B1. Collect data related to bulk cargo railway transportation costs in historical waybills, including: cargo loading method, number of loading tools, mileage, cargo weight, cargo type, and base price per ton of cargo in historical waybills; Among them, the cargo loading method matrix in the historical waybill is: , divided into two types, For vehicle transportation, For container transport; Matrix of the number of loading tools in historical waybills: ; Mileage matrix in historical waybills: ; Cargo weight matrix in historical waybills: ; Cargo type matrix in historical waybills: ; The matrix of the base price per ton of cargo1 (cost of loading and unloading) in historical waybills: ; The matrix of base price 2 (operating freight) per ton of goods in historical waybills: .
[0024] Step B2. Use the variables such as cargo loading method, number of loading tools, mileage, cargo weight, cargo type, and base price per ton of cargo in the historical waybills to construct the "propensity score" of the transportation mode; Step B2.1. Arrange the railway transportation costs in the historical waybills: ; In the formula, is the railway transportation cost matrix in the historical waybill, is the base price matrix of each ton of cargo in the historical waybill. is the base price per ton of cargo in the historical waybill 2 matrix, is the mileage matrix in the historical waybill, It is the cargo weight matrix in the historical waybills.
[0025] Step B2.2. Use the multiple regression method to perform linear regression on the parameters: ; In the formula, , , , , , are coefficients, It is the cargo loading method matrix in the historical waybill; Load the tool quantity matrix in the historical waybill; is the mileage matrix in the historical waybills; It is the cargo weight matrix in the historical waybill; is the cargo type matrix in the historical waybills; It is the railway transportation cost matrix in the historical waybill.
[0026] Step B2.3. Using the logit regression principle, the above regression equation (step B2.2) is converted into the propensity score of railway transportation costs. The propensity score matrix is denoted by , the "propensity score" regression model is as follows: ; In the formula, log is the logarithmic symbol, is the propensity score matrix of railway transportation costs, It is the railway transportation cost matrix in the historical waybill.
[0027] The propensity score can be obtained by sorting: ; In the formula, is the propensity score matrix of railway transportation costs, It is the cargo loading method matrix in the historical waybill; Load the tool quantity matrix in the historical waybill; is the mileage matrix in the historical waybills; It is the cargo weight matrix in the historical waybill; It is the cargo type matrix in the historical waybills.
[0028] B3. Based on the propensity score, the inverse probability weighted method is used to calculate the average impact of each influencing factor on railway transportation costs in the historical waybill data. The calculation formula is as follows: ; in, It is an influencing factor in historical waybills i Average treatment effect on railway transportation costs, influencing factors i Including cargo loading method, number of loading tools, mileage, cargo weight, cargo type; E is the expectation symbol.
[0029] Specifically, the average impact of cargo loading methods on railway transportation costs is calculated as follows: ; In the formula, is the average treatment effect of the cargo loading matrix on the railway transportation cost in the historical waybill, reflecting the average impact of the cargo loading method on the railway transportation cost; E is the expected symbol; It is the cargo loading method matrix in the historical waybill; is the propensity score matrix for railway transportation costs.
[0030] The average impact of the number of loading tools on railway transportation costs is calculated as follows: ; In the formula, is the average treatment effect of the loading tool quantity matrix in the historical waybill on the railway transportation cost, reflecting the average impact of the cargo loading method on the railway transportation cost; E is the expected symbol; Load the tool quantity matrix in the historical waybill; is the propensity score matrix for railway transportation costs.
[0031] The calculation formula for the average impact of mileage on railway transportation costs is: ; In the formula, is the average treatment effect of the mileage matrix in the historical waybill on the railway transportation cost, reflecting the average impact of the cargo loading method on the railway transportation cost; E is the expected symbol; is the mileage matrix in the historical waybills; is the propensity score matrix for railway transportation costs.
[0032] The average impact of cargo weight on railway transportation costs is calculated as follows: ; In the formula, is the average treatment effect of the cargo weight matrix in the historical waybill on the railway transportation cost, reflecting the average impact of the cargo loading method on the railway transportation cost; E is the expected symbol; It is the cargo weight matrix in the historical waybill; is the propensity score matrix for railway transportation costs.
[0033] The average impact of cargo type on railway transportation costs is calculated as follows: ; In the formula, is the average treatment effect of the cargo type matrix on the railway transportation cost in the historical waybill, reflecting the average impact of the cargo loading method on the railway transportation cost; E is the expected symbol; is the cargo type matrix in the historical waybills; is the propensity score matrix for railway transportation costs.
[0034] Step B4. Determine the most sensitive factor affecting railway transportation costs among the above factors: ; In the formula, It is the most sensitive factor affecting railway transportation costs; is the symbol of the influencing factor corresponding to the maximum value. For example, if Maximum, then cargo loading method It is the most sensitive factor affecting railway transportation costs; is the average treatment effect of freight loading mode on rail transport costs; is the average treatment effect of the number of loading tools on rail transport costs; is the average treatment effect of travelled miles on rail transport costs; is the average treatment effect of freight weight on rail transport costs; is the average treatment effect of freight type on rail transport costs.
[0035] like Figure 4 As shown, in this embodiment, in step C, the causal inference model based on the discount leverage mechanism will input The data is divided into proxy variables, latent variables, and treatment variables; among them, proxy variables refer to variables that are related to the output and cannot be changed. In the causal inference model of this application, the cargo type, cargo weight, and mileage in the customer's waybill are directly related to the final quotation and cannot be changed, so they are set as proxy variables; treatment variables refer to influencing factors that are related to the output and can be adjusted; in this application, the choice of cargo loading method is related to the final output result and is adjustable, so it is set as a treatment variable; latent variables refer to factors that are indirectly obtained from input data and have an impact on the output. In this application, the average treatment effect (ATE) of each influencing factor on the railway transportation cost obtained based on historical data, the most sensitive factor, and the number of loading tools affected by the cargo loading method are set as latent variables; the causal inference model calculates the current railway transportation cost based on each input data and determines whether it meets If the conditions are met, the inference results are output; is the current railway transportation cost; The discount given by the railway company to realize the "road to rail" transfer of bulk goods; The current cost of transporting goods using the road system.
[0036] In this embodiment, based on historical waybill data , establish the relationship between the average treatment effect of each indicator in the historical waybill, the current customer waybill data, and the current freight cost, and provide guidance for formulating the quotation given by the railway company for the current transportation order (it is expected that road to rail can be realized): ; ; In the formula, is the current waybill railway transportation fee; log is the logarithmic sign; is the current cargo loading method, is the average treatment effect of freight loading mode on rail transport costs; is the number of loading tools for the current waybill, is the average treatment effect of the number of loading vehicles on rail transport costs; is the mileage of the current waybill, is the average treatment effect of travelled miles on rail transport costs; is the current waybill cargo weight, is the average treatment effect of freight weight on rail transport costs; is the type of goods in the current waybill, is the average treatment effect of freight type on rail transport costs; It is the most sensitive factor affecting railway transportation costs in historical waybills. yes The average treatment effect on railway transportation costs; E is the expected sign.
[0037] Further, if Figure 4 , Figure 5 As shown, in this embodiment, step D includes the following steps: Step D1. Arrange the data related to the railway transportation costs of bulk goods in the current customer waybill, including: the loading method of the goods in this (current waybill), the number of loading tools, the mileage, the weight of the goods, and the type of goods; define the symbols of each data as follows: The cargo loading method in the current waybill: , divided into two types, For vehicle transportation, For container transport; Number of loading tools in the current waybill: ; Mileage in the current waybill: ; The weight of the goods in the current waybill: ; The type of goods in the current waybill: .
[0038] Step D2. Based on the constructed causal inference model, provide a railway transportation plan, quotation and discount; Step D2.1. Change the cargo category in the current waybill 、The weight of the goods in the current waybill 、Mileage in the current waybill As a proxy variable, the cargo loading method in the current waybill ( , ) as the treatment variable; the most sensitive factor determined in step B, the average treatment effect of each influencing factor, and the effect of cargo loading mode ( , ) The number of loading tools affected As a latent variable, it is introduced into the causal inference model in step C; Step D2.2. Initialize the railway transport mode, ; Step D2.3. Calculate the current railway transportation cost and determine the difference between railway transportation cost and road transportation cost Is it greater than 0?
[0039] In the formula, The difference between the cost of transporting goods by rail and by road; is the current railway transportation cost; The discount given by the railway company to realize the "road to rail" transfer of bulk goods; The cost of transporting goods using the road system; like , output the inference results, including: the final railway cargo loading method 、Number of loading tools in the current waybill , Railway transportation cost quotation , Discount Rate ; like , adjust the railway transportation mode , Discount Rate , repeat D2.3 until , to realize the formulation of the strategy of “road to rail” for bulk cargo.
[0040] Embodiment 2: like Figure 6 As shown, this embodiment also provides a bulk cargo "road to rail" intelligent decision-making system based on causal inference, the system includes a data input module, an adaptation module, a data processing module, and a causal inference model; The data input module is used to input data, including freight OD, freight volume, cargo damage requirements, freight timeliness, cargo loading method, number of loading tools, mileage, cargo weight, cargo type, and base price per ton of cargo; The adaptation module is used to determine whether the freight demand is compatible with railway transportation based on the input freight data; The data processing module is used to collect statistics on the railway transportation costs of bulk goods in historical waybills, calculate the propensity score of the railway transportation costs, and calculate the average impact of different influencing factors on the railway transportation costs in the historical waybill data using the inverse probability weighting method based on the propensity score, and determine the most sensitive factor affecting the railway transportation costs; The causal inference model calculates the current railway transportation cost of the waybill based on the input proxy variables, treatment variables, and latent variable data to determine whether If the conditions are met, the inference results are output; is the current railway transportation cost; The discount given by the railway company to realize the "road to rail" transfer of bulk goods; The cost of transporting goods using the road system; ; ; In the formula, is the current waybill railway transportation fee; log is the logarithmic sign; is the current cargo loading method, is the average treatment effect of freight loading mode on rail transport costs; is the number of loading tools for the current waybill, is the average treatment effect of the number of loading vehicles on rail transport costs; is the mileage of the current waybill, is the average treatment effect of travelled miles on rail transport costs; is the current waybill cargo weight, is the average treatment effect of freight weight on rail transport costs; is the type of goods in the current waybill, is the average treatment effect of freight type on rail transport costs; It is the most sensitive factor affecting railway transportation costs in historical waybills. yes The average treatment effect on railway transportation costs; E is the expected sign.
[0041] The present invention also provides an electronic device, comprising: one or more processors and a memory; wherein the memory is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned intelligent decision-making method for "road to rail" of bulk cargo based on causal inference.
[0042] The present invention also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned intelligent decision-making method for “road to rail” of bulk cargo based on causal inference.
[0043] Those skilled in the art will appreciate that all or part of the functions of the various methods / modules in the above embodiments may be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program may be stored in a computer-readable storage medium, which may include: a read-only memory, a random access memory, a disk, an optical disk, a hard disk, etc. The program is executed by a computer to implement the above functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above functions can be implemented.
[0044] In addition, when all or part of the functions in the above-mentioned embodiments are implemented by means of a computer program, the program can also be stored in a storage medium such as a server, another computer, a disk, an optical disk, a flash drive or a mobile hard disk, and saved to the memory of a local device by downloading or copying, or the system of the local device is updated. When the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.
[0045] The above specific examples are used to illustrate the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art of the present invention, according to the idea of the present invention, some simple deductions, deformations or substitutions can be made. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. An intelligent decision-making method for bulk cargo "transfer from road to rail" based on causal inference, characterized in that: The method comprises the following steps: Step A: Determine whether the customer's freight demand is compatible with railway transportation. If so, proceed to step B. Step B. Calculate the average treatment effect of different influencing factors on railway transportation costs in historical waybill data by using the inverse probability weighted method of causal inference, and track the most sensitive factors affecting railway transportation costs; Step C. Construct a causal inference model based on the discount leverage mechanism, introduce the average treatment effect in causal inference, and simplify the complexity of the "public to rail" intelligent decision-making problem from 2n+m to n+m; The causal inference model divides the input data into proxy variables, latent variables, and treatment variables; the cargo type, cargo weight, and mileage in the customer's waybill are proxy variables, the cargo loading method is the treatment variable, and the average treatment effect of each influencing factor on the railway transportation cost obtained based on historical data, the most sensitive factor, and the number of loading tools affected by the cargo loading method are set as latent variables; the causal inference model calculates the current railway transportation cost based on each input data and determines whether it meets the If the conditions are met, the inference results are output; is the current railway transportation cost, which is related to the input data; The discount given by the railway company to realize the transfer of bulk goods from road to rail; The cost of transporting goods using the road system; Step D: Input the data in the current waybill and provide the railway transportation plan, quotation and discount based on the constructed causal inference model.
2. According to the causal inference-based intelligent decision-making method for bulk cargo "road to rail" according to claim 1, it is characterized by: When determining whether the customer's freight demand is compatible with railway transportation, step A specifically includes the following steps: Step A1. Enter the customer's freight departure and destination, freight volume, cargo damage requirements, and freight timeliness; Step A2. Check the cargo item number to determine whether there is a requirement for reverse loading and damage. If the sender has a requirement for reverse loading and damage, the freight demand of the cargo is not compatible with railway transportation. If the sender does not have a requirement for reverse loading and damage, proceed to step A3. Step A3. According to the customer's freight departure and destination, determine whether the adjacent sending station and terminal station meet the railway company's loading and unloading conditions. If not, the freight demand of the goods is not compatible with railway transportation; if so, execute step A4; Step A4. According to the customer's freight departure and destination, determine whether there is a dedicated railway freight line between the departure and destination. If there is a dedicated line, adapt it to railway transportation and execute step A5; If there is no dedicated freight line, determine whether there is a direct line between the adjacent sending station and the terminal. If there is no direct line, the freight demand of the goods is not compatible with railway transportation. If there is a direct line between the adjacent sending station and the terminal, the transportation conditions of the railway are met and it is compatible with railway transportation, and step A5 is executed; Step A5. According to the freight volume of the customer, determine whether it meets the minimum freight volume requirement of the railway company. If not, the freight demand of the goods is not compatible with railway transportation. If it is, execute step A6; Step A6. Determine whether the freight demand of the goods is compatible with rail transportation according to the customer's freight timeliness requirements. For goods that require delivery within 3 days, it is determined that they are not compatible with rail transportation; otherwise, they are compatible with rail transportation.
3. According to the causal inference-based intelligent decision-making method for bulk cargo "road to rail" in claim 1, it is characterized by: The calculation formula for the average treatment effect of different influencing factors on railway transportation costs is: ； in, It is an influencing factor in historical waybills i Average treatment effect on railway transportation costs, influencing factors i Including cargo loading method, number of loading tools, mileage, cargo weight, cargo type; E is the expected symbol; is the propensity score matrix for railway transportation costs.
4. According to the causal inference-based intelligent decision-making method for bulk cargo "road to rail" in claim 1, it is characterized by: The most sensitive factor affecting railway transportation costs is determined by the following formula: ； In the formula, It is the most sensitive factor affecting railway transportation costs; is the average treatment effect of freight loading mode on rail transport costs; is the average treatment effect of the number of loading tools on rail transport costs; is the average treatment effect of travelled miles on rail transport costs; is the average treatment effect of freight weight on rail transport costs; is the average treatment effect of freight type on rail transport costs; Representative The maximum value in .
5. According to the causal inference-based intelligent decision-making method for bulk cargo "road to rail" in claim 1, it is characterized by: The relationship between the current railway transportation cost and the average treatment effect in the historical waybill, the most sensitive factor and the current waybill data is: ； ； In the formula, The current railway transportation fee of the waybill; log is the logarithmic symbol; is the current cargo loading method, is the average treatment effect of freight loading mode on rail transport costs; is the number of loading tools for the current waybill, is the average treatment effect of the number of loading vehicles on rail transport costs; is the mileage of the current waybill, is the average treatment effect of travelled miles on rail transport costs; is the current waybill cargo weight, is the average treatment effect of freight weight on rail transport costs; is the type of goods in the current waybill, is the average treatment effect of freight type on rail transport costs; It is the most sensitive factor affecting railway transportation costs in historical waybills. yes The average treatment effect on railway transportation costs; E is the expected sign.
6. According to the causal inference-based intelligent decision-making method for bulk cargo "road to rail" in claim 1, it is characterized by: Step D comprises the following steps: Step D1. Arrange the data related to the railway transportation costs of bulk goods in the current customer waybill, including: the cargo loading method, the number of loading tools, the mileage, the cargo weight, and the cargo type in the current waybill; Step D2. Based on the constructed causal inference model, provide a railway transportation plan, quotation and discount; Step D2.
1. Change the cargo category in the current waybill 、The weight of the goods in the current waybill 、Mileage in the current waybill As a proxy variable, the cargo loading method in the current waybill ( , ) as the treatment variable; the most sensitive factor determined in step B, the average treatment effect of each influencing factor, and the number of loading tools affected by the cargo loading method As a latent variable, it is introduced into the causal inference model in step C; Step D2.
2. Initialize the railway transport mode, ; Step D2.
3. Calculate the current railway transportation cost and determine the difference between railway transportation cost and road transportation cost Is it greater than 0? ； In the formula, The difference between the cost of transporting goods by rail and by road; is the current railway transportation cost; The discount given by the railway company to realize the transfer of bulk goods from road to rail; The cost of transporting goods using the road system; like , output the inference results, including: the final railway cargo loading method 、The number of loading tools in the current waybill , Railway transportation cost quotation , Discount Rate ; like , adjust the railway transportation mode , Discount Rate , repeat D2.3 until , to realize the formulation of the strategy of "road to rail" for bulk cargo.
7. According to claim 3, a bulk cargo "road to rail" intelligent decision-making method based on causal inference is characterized in that: The calculation formula for the propensity score of railway transportation costs is: ； In the formula, is the propensity score matrix of railway transportation costs, , , , , , All are coefficients; It is the cargo loading method matrix in the historical waybill; Load the tool quantity matrix in the historical waybill; is the mileage matrix in the historical waybills; It is the cargo weight matrix in the historical waybill; It is the cargo type matrix in the historical waybills.
8. An intelligent decision-making system for bulk cargo "transfer from road to rail" based on causal inference, characterized in that: The system includes a data input module, an adaptation module, a data processing module, and a causal inference model; The data input module is used to input data, including the place of departure and destination of the freight, the freight volume, the cargo damage requirements, the timeliness of the freight, the cargo loading method, the number of loading tools, the mileage, the cargo weight, the cargo type, and the base price per ton of cargo; The adaptation module is used to determine whether the freight demand is compatible with railway transportation based on the input freight data; The data processing module is used to collect statistics on the railway transportation costs of bulk goods in historical waybills, calculate the propensity score of the railway transportation costs, and calculate the average impact of different influencing factors on the railway transportation costs in the historical waybill data using the inverse probability weighting method based on the propensity score, and determine the most sensitive factor affecting the railway transportation costs; The causal inference model calculates the current railway transportation cost based on the input proxy variables, treatment variables, and latent variable data to determine whether If the conditions are met, the inference results are output; is the current railway transportation cost; The discount given by the railway company to realize the transfer of bulk goods from road to rail; The cost of transporting goods using the road system; ； ； In the formula, is the current railway transportation fee of the waybill; log is the logarithmic sign; is the current cargo loading method, is the average treatment effect of freight loading mode on rail transport costs; is the number of loading tools for the current waybill, is the average treatment effect of the number of loading vehicles on rail transport costs; is the mileage of the current waybill, is the average treatment effect of travelled miles on rail transport costs; is the current waybill cargo weight, is the average treatment effect of freight weight on rail transport costs; is the type of goods in the current waybill, is the average treatment effect of freight type on rail transport costs; It is the most sensitive factor affecting railway transportation costs in historical waybills. yes The average treatment effect on railway transportation costs; E is the expected sign.