Intelligent automobile sale order management method and system based on cross-border trade

Abstract

The invention discloses an intelligent automobile sale order management method and system based on cross-border trade, and the method comprises the steps: obtaining a cross-border trade historical order data set, clustering the historical order data set to construct a multi-dimensional user portrait model, and extracting a user portrait feature vector according to the multi-dimensional user portrait model; performing feature fusion on the user portrait feature vector, external market trend data and a tax policy change factor, inputting the fusion result into a demand prediction model to generate a prediction result, establishing a multi-objective optimization function, and performing optimization solution to generate a warehouse distribution scheduling scheme; and executing a scheduling instruction through a block chain smart contract, collecting actual logistics data and a prediction deviation value in real time, updating a user portrait feature weight through an online learning mechanism, and adjusting hyper-parameters of the demand prediction model by using a Bayesian optimization algorithm to optimize the demand prediction model. According to the invention, the storage resource configuration is scheduled by accurately predicting the user behavior, and the operation efficiency of cross-border automobile sales is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automotive order management, and particularly to an intelligent automotive sales order management method and system based on cross-border trade. Background Art

[0002] In the field of cross-border automotive trade, due to the variety of data types and complex business processes, there are still many challenges in sales order management: on the one hand, traditional demand forecasting methods often rely on single or a small number of dimensions of data, making it difficult to comprehensively capture the diverse user demand characteristics in cross-border trade, resulting in poor accuracy in predicting future order volumes and difficulty in meeting the enterprise's need to accurately grasp market dynamics. On the other hand, in the warehousing allocation and scheduling link, previous solutions have not fully considered various factors such as logistics costs, timeliness deviations, and inventory turnover risks. In the face of warehouse capacity limitations, cross-border transportation timeliness requirements, and customs clearance risks, it is impossible to achieve the optimal allocation of resources, often resulting in cost waste and low operational efficiency. Due to the lack of an effective model dynamic optimization mechanism, it is difficult to adjust in real time according to actual logistics data and prediction deviations, resulting in the gradual disconnection of the order management solution from the actual situation. Summary of the Invention

[0003] To solve at least one of the above-mentioned technical problems, the present invention provides an intelligent automotive sales order management method and system based on cross-border trade.

[0004] In a first aspect, the present invention provides an intelligent automotive sales order management method based on cross-border trade, characterized in that the method includes:

[0005] Obtain a cross-border trade historical order data set, cluster the historical order data set to construct a multi-dimensional user portrait model, and extract user portrait feature vectors according to the multi-dimensional user portrait model;

[0006] Fuse the user portrait feature vectors with external market trend data and tariff policy change factors, and input the fused features into a trained demand forecasting model to generate a prediction result, where the prediction result includes the predicted value of the order volume within a future preset period, the vehicle model demand distribution matrix, and the probability distribution of the transaction time window;

[0007] Establish a multi-objective optimization function based on the prediction result, including establishing an objective function according to logistics costs, timeliness deviations, and inventory turnover risks, and using the maximum capacity of each warehouse, the cross-border transportation timeliness threshold, and the customs clearance risk coefficient as constraint conditions; perform optimization solution on the objective function to generate a warehousing allocation and scheduling plan;

[0008] Based on the warehousing allocation and scheduling plan, execute the scheduling instructions through the blockchain smart contract, collect the actual logistics data and the prediction deviation value in real time, update the weights of the user portrait features through the online learning mechanism, and use the Bayesian optimization algorithm to adjust the hyperparameters of the demand prediction model to optimize the demand prediction model.

[0009] Preferably, before executing the scheduling instructions through the blockchain smart contract, it further includes:

[0010] Input the prediction result output by the demand prediction model into the digital twin simulation system, simulate the load status of the logistics network under different warehousing allocation plans, and obtain the congestion probability of the key nodes of the network load;

[0011] When the congestion probability is greater than the preset threshold, regenerate the scheduling plan including alternative transportation routes and emergency warehouses to generate new scheduling instructions.

[0012] Preferably, the data set includes user attribute data, transaction time series data, automobile model configuration parameters, and cross-border logistics node information.

[0013] Preferably, the output dimensions of the model multi-dimensional user portrait model include user purchasing power grading indicators, brand preference coefficients, vehicle model iteration response cycles, and regional policy sensitivity parameters.

[0014] In a second aspect, the present invention also provides an intelligent automobile sales order management system based on cross-border trade, and the system includes:

[0015] A data collection unit for obtaining a cross-border trade historical order data set, clustering the historical order data set to construct a multi-dimensional user portrait model, and extracting user portrait feature vectors according to the multi-dimensional user portrait model;

[0016] A demand prediction unit for fusing the user portrait feature vectors with external market trend data and tariff policy change factors, and inputting the fused features into the trained demand prediction model to generate prediction results, where the prediction results include the order volume prediction value, vehicle model demand distribution matrix, and transaction time window probability distribution within a preset future period;

[0017] A scheduling and allocation unit for establishing a multi-objective optimization function based on the prediction results, including establishing an objective function according to logistics costs, timeliness deviation, and inventory turnover risk, and using the maximum capacity of each warehouse, cross-border transportation timeliness threshold, and customs clearance risk coefficient as constraints; performing optimization solution on the objective function to generate a warehousing allocation and scheduling plan;

[0018] A model iteration unit, which is used to execute scheduling instructions through a blockchain smart contract based on a warehousing allocation and scheduling plan, collect actual logistics data and prediction deviation values in real time, update the weight of user portrait features through an online learning mechanism, and adjust the hyperparameters of the demand prediction model by using a Bayesian optimization algorithm to optimize the demand prediction model.

[0019] Preferably, the intelligent vehicle sales order management system based on cross-border trade further includes an instruction simulation unit, which is used for:

[0020] Input the prediction result output by the demand prediction model into the digital twin simulation system, simulate the load status of the logistics network under different warehousing allocation plans, and obtain the congestion probability of the key nodes of the network load;

[0021] When the congestion probability is greater than a preset threshold, regenerate a scheduling plan including alternative transportation routes and emergency warehouses to generate new scheduling instructions.

[0022] Preferably, the data set includes user attribute data, transaction time series data, vehicle model configuration parameters, and cross-border logistics node information.

[0023] Preferably, the output dimensions of the model multi-dimensional user portrait model include user purchasing power grading indicators, brand preference coefficients, vehicle model iteration response cycles, and regional policy sensitivity parameters.

[0024] In a third aspect, the present invention also provides an electronic device, including a processor and a memory. The memory is used to store computer program code, and the computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes the method as described in the first aspect and any one of its possible implementation manners above.

[0025] In a fourth aspect, the present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by the processor of an electronic device, the processor is caused to execute the method as described in the first aspect and any one of its possible implementation manners above.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] The intelligent management method for automobile sales orders based on cross-border trade provided by the present invention constructs a multi-dimensional user portrait model by obtaining cross-border trade historical order data sets, and fuses its feature vectors with external market trend data and tariff policy change factors and inputs them into a demand prediction model, which can fully explore market demand information in a complex environment, realize accurate prediction of the order volume, vehicle type demand distribution matrix and transaction time window probability distribution within a preset future period, and provide a solid and reliable data basis for subsequent decision-making. In terms of warehousing allocation and scheduling, a multi-objective optimization function including logistics cost, timeliness deviation and inventory turnover risk is established based on the prediction results, and optimization is solved with warehouse capacity, transportation timeliness and customs clearance risk as constraints. The generated warehousing allocation and scheduling plan can comprehensively balance various factors, realize the optimal allocation of resources, significantly reduce costs and improve operation efficiency. By using blockchain smart contracts to execute scheduling instructions, collecting actual logistics data and prediction deviation values in real time, updating the feature weights of the user portrait through an online learning mechanism and adjusting the hyperparameters of the demand prediction model using the Bayesian optimization algorithm, the entire order management scheme has the ability of dynamic optimization, can closely follow the changes in the cross-border trade environment, and continuously maintain a high level of order management.

[0028] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the present disclosure. Brief Description of the Drawings

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background art, the following will describe the drawings required to be used in the embodiments of the present invention or the background art.

[0030] The drawings herein are incorporated into the specification and constitute a part of this specification. These drawings show embodiments consistent with the present disclosure and, together with the specification, are used to illustrate the technical solutions of the present disclosure.

[0031] Figure 1 It is a schematic flow chart of an intelligent management method for automobile sales orders based on cross-border trade provided by an embodiment of the present invention;

[0032] Figure 2 It is a schematic structural diagram of an intelligent management system for automobile sales orders based on cross-border trade provided by an embodiment of the present invention;

[0033] Figure 3 It is a schematic structural diagram of an intelligent management system for automobile sales orders based on cross-border trade provided by another embodiment of the present invention. Detailed Embodiments

[0034] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0035] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0036] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of an intelligent automobile sales order management method based on cross-border trade provided by an embodiment of the present invention. As Figure 1 shown, the method includes:

[0037] S10. Obtain a cross-border trade historical order data set, perform clustering on the historical order data set to construct a multi-dimensional user portrait model, and extract user portrait feature vectors according to the multi-dimensional user portrait model.

[0038] Extract user attribute data from the internal sales management system of the enterprise, including basic information such as the user's age, gender, occupation, and location, which helps to initially understand the characteristics of the user. Obtain transaction time series data from the transaction record database, record the order placement time, payment time, delivery time, etc. of each order to analyze the user's purchase time pattern. The automobile model configuration parameters can be obtained from the product information library provided by the automobile manufacturer, and details of the engine parameters, interior configuration, exterior color, etc. of each model are recorded, which is crucial for analyzing the user's preference for models. The cross-border logistics node information is obtained from the system of the logistics partner, including the timestamps and status information of the goods at each cross-border logistics node (such as the departure port, transit point, destination port, etc.).

[0039] To improve data quality, first check the integrity of the data. For records with missing values, handle them according to the characteristics of the data. For example, for missing values in user attribute data, if the missing ratio is small, they can be filled by asking the user or based on the characteristics of users of the same type; for missing values in transaction time series data, if the missing time points are critical (such as the order placement time), consider deleting the record to avoid affecting subsequent analysis. Handle duplicate data by comparing all fields, finding and deleting exactly the same records to ensure data uniqueness. Standardize the data. For example, uniformly convert transaction time series data to the same time format, and normalize numerical data (such as engine power) in car model configuration parameters for subsequent analysis and modeling.

[0040] Furthermore, cluster the historical order data set to construct a multi-dimensional user portrait model:

[0041] Preferably, the data set includes user attribute data, transaction time series data, car model configuration parameters, and cross-border logistics node information. Therefore, when performing feature extraction, extract some derived features from the user attribute data, such as calculating the user's potential purchasing power index based on the economic development level and consumption index of the user's location. For transaction time series data, calculate features such as the user's purchase frequency and average purchase interval time. Extract the user's selection tendency features for different configurations (such as high-end configurations, environmental protection configurations, etc.) from the car model configuration parameters. Combine the cross-border logistics node information to analyze the logistics preferences of users in different regions (such as the choice of fast logistics or economic logistics).

[0042] The K-Means clustering algorithm is adopted. This algorithm is simple and efficient and is suitable for large-scale data sets. First, determine the initial number of clusters K based on experience or through multiple trials. For example, you can start from K = 3 and gradually increase the value of K. Evaluate the clustering effect by calculating metrics such as the silhouette coefficient and select the optimal K value. Input the data set after feature engineering into the K-Means algorithm. The algorithm will divide the data into K clusters according to the distance between data points (such as Euclidean distance). For each clustering cluster, calculate the user purchasing power grading index. By analyzing data such as the income level of users within the cluster and the price range of cars purchased, divide the user purchasing power into three levels: high, medium, and low. Calculate the brand preference coefficient. Statistically analyze the frequency of users within the cluster purchasing different car brands to determine the degree of preference of users in this cluster for a specific brand. For example, if the proportion of users within a certain cluster purchasing Tesla cars is significantly higher than other brands, then the brand preference coefficient of users in this cluster for the Tesla brand is higher. Determine the vehicle model iteration response cycle. Observe the purchase time interval of users within the cluster after the release of new car models to evaluate the response speed of users to vehicle model iteration. For example, if the purchase proportion of users in a certain cluster is relatively high within 1 - 3 months after the release of a new model, then the vehicle model iteration response cycle of users in this cluster is shorter. Analyze the regional policy sensitivity parameter. Study the impact of policies in different regions (such as car purchase subsidies, environmental protection policies, etc.) on the purchase behavior of users within the cluster. For example, in some regions, there is a new energy vehicle purchase subsidy policy. If the proportion of users in this region purchasing new energy vehicles increases significantly after the introduction of the subsidy policy, it indicates that the users in this cluster have a high sensitivity to regional policies.

[0043] Finally, extract the user portrait feature vector according to the multi-dimensional user portrait model. For each user, according to the clustering cluster to which it belongs, extract the corresponding user purchasing power grading index, brand preference coefficient, vehicle model iteration response cycle, and regional policy sensitivity parameter. Combine these indicators into a feature vector. For example: [User purchasing power grading (1 represents high, 2 represents medium, 3 represents low), brand preference coefficient (such as the Tesla brand preference coefficient is 0.8), vehicle model iteration response cycle (in months), regional policy sensitivity parameter (ranging from 0 - 1 according to the degree of influence)]. Store the extracted user portrait feature vector in a dedicated database for subsequent feature fusion with external market trend data and tariff policy change factors and input it into the demand prediction model to provide strong support for the management of car sales orders in cross-border trade. Preferably, the output dimensions of the multi-dimensional user portrait model include the user purchasing power grading index, brand preference coefficient, vehicle model iteration response cycle, and regional policy sensitivity parameter.

[0044] S20. Feature - fuse the user portrait feature vector with external market trend data and tariff policy change factors, and input the fused features into the trained demand prediction model to generate a prediction result, where the prediction result includes the predicted order volume, vehicle type demand distribution matrix, and trading time - window probability distribution within a preset future period.

[0045] It should be noted that the external market trend data includes:

[0046] Macroeconomic data: Obtain macro - economic indicators such as the GDP growth rate, inflation rate, and unemployment rate of the target market country or region from authoritative economic data - releasing institutions (such as the International Monetary Fund, the World Bank, etc.). These data reflect the local overall economic situation and have an important impact on the automotive consumer market. For example, when the GDP growth rate is relatively high, consumers' purchasing power may increase, and the demand for automobiles may rise.

[0047] Industry market data: Collect information such as the automotive sales trend, market saturation, and competitor dynamics in the target market through professional automotive industry research reports and data from market research institutions. For example, by understanding the new vehicle release plans of competitors, the potential impact on the demand for its own vehicle models can be analyzed.

[0048] Consumer behavior data: Collect information such as consumers' attention hotspots and preference change trends for automobiles by using social media monitoring tools, online questionnaires, etc. For example, if the discussion heat of electric vehicles on social media continues to rise, it may indicate the future growth of the demand for electric vehicles.

[0049] The tariff policy change factors include:

[0050] Pay attention to the official announcements of each country's government and policy documents issued by the customs department to obtain information on tariff rate adjustments in a timely manner. For example, if a country raises the import tariff on automobiles, this will directly increase the import cost of automobiles and affect the automobile price and demand.

[0051] Analyze the reasons and background of tariff policy changes and evaluate their impact on automobiles of different models and different origins. For example, some countries implement lower tariffs on imported new - energy vehicles to protect their domestic new - energy vehicle industries, while imposing higher tariffs on traditional fuel vehicles.

[0052] Furthermore, clean and preprocess the collected external market trend data and tariff policy change factors. For missing values in macroeconomic data, time series interpolation method can be used for filling; for outliers in industry market data, identify and correct them through statistical analysis methods (such as box plots). Unify the format of all data to make it suitable for fusion with the user portrait feature vector. For example, unify the time series data to the same time interval, and normalize various types of indicator data to make the data in the same dimension for subsequent calculations.

[0053] In one embodiment, adopt the splicing fusion method to horizontally splice the user portrait feature vector with the preprocessed external market trend data and tariff policy change factors. Preferably, the weighted fusion method can also be used. According to the importance of each data for demand prediction, different weights are assigned to the user portrait feature vector, external market trend data, and tariff policy change factors respectively. For example, for some markets, the change of tariff policy has a greater impact on automobile demand, then a higher weight is assigned to the tariff policy change factor, and then the fused feature vector is obtained by weighted summation.

[0054] For training the demand prediction model, models including but not limited to the long short-term memory network (LSTM) model, support vector regression (SVR) model, etc. can be selected. Input the fused feature vector into the trained demand prediction model. Before input, it is necessary to ensure that the input format of the model is consistent with the format of the fused feature vector. For example, if the LSTM model requires the input data to be a three-dimensional tensor (number of samples, time steps, number of features), then the fused feature vector needs to be reshaped according to the corresponding format. The model calculates according to the input feature vector and outputs the predicted value of the order volume within a preset future period.

[0055] For example, the model predicts that the automobile order volumes in a certain region in the next three months are 1,000 vehicles, 1,200 vehicles, and 1,500 vehicles respectively. For the vehicle type demand distribution matrix, the model will predict the demand proportion of different vehicle types within the preset future period according to the input features. For example, the prediction results show that within the next month, the demand proportion of SUV models is 40%, the demand proportion of sedan models is 35%, and the demand proportion of MPV models is 25%. In terms of the probability distribution prediction of the transaction time window, the model will calculate the probability of users placing orders at different time periods. For example, the prediction results show that the probability of users placing orders between 9 am and 11 am on weekdays is 0.2; the probability of users placing orders between 2 pm and 4 pm on weekends is 0.15, etc.

[0056] S30. Establish a multi-objective optimization function based on the prediction results, including establishing an objective function according to logistics costs, timeliness deviations, and inventory turnover risks, and using the maximum capacity of each warehouse, the cross-border transportation timeliness threshold, and the customs clearance risk coefficient as constraints; optimize and solve the objective function to generate a warehousing allocation and scheduling plan.

[0057] Step 1. Assume there are n warehouses and m vehicle types in total. T is the time span of the future preset period, which is divided into t = 1, 2, …, T time stages.

[0058] x ijt : represents the quantity of the j-th vehicle type allocated to the i-th warehouse in the t-th stage, where i = 1, …, n; j = 1, …, m;

[0059] y ijt : represents the quantity of the j-th vehicle type actually shipped from the i-th warehouse in the t-th stage;

[0060] D jt : represents the predicted order volume of the j-th vehicle type in the t-th stage, obtained from the demand prediction model;

[0061] C ij : represents the unit logistics cost of storing the j-th vehicle type in the i-th warehouse;

[0062] S it : represents the inventory turnover risk coefficient of the i-th warehouse in the t-th stage, which can be estimated based on historical inventory data and sales data;

[0063] L it : represents the actual inventory level of the i-th warehouse in the t-th stage;

[0064] represents the maximum capacity of the i-th warehouse;

[0065] T ij : represents the transportation timeliness of shipping the j-th vehicle type from the i-th warehouse to the destination;

[0066] represents the cross-border transportation timeliness threshold;

[0067] R i : represents the customs clearance risk coefficient of the goods in the i-th warehouse, mainly evaluated based on historical customs clearance data and current customs policies.

[0068] Step 2. Establish the objective function:

[0069] 1) Logistics cost objective function Z1:

[0070] The logistics cost is the sum of the products of the unit logistics cost of storing different vehicle types in each warehouse in each stage and the storage quantity:

[0071]

[0072] 2) Aging deviation objective function Z2:

[0073] The aging deviation is measured by the difference between the actual transportation aging and the aging threshold. Let w ij be the aging weight for the j-th vehicle type transported from the i-th warehouse. Then we have:

[0074]

[0075] 3) Inventory turnover risk objective function Z3:

[0076] The inventory turnover risk is related to the inventory levels and inventory turnover risk coefficients of each warehouse at each stage. Let α be the inventory turnover risk weight coefficient:

[0077]

[0078] Taking into account the above three objective functions comprehensively, a multi-objective optimization function Z is constructed. The three objectives are combined by weighted summation. Let δ1, δ2, and δ3 be the weights of logistics cost, aging deviation, and inventory turnover risk respectively, and satisfy δ1 + δ2 + δ3 = 1. Therefore, the multi-objective optimization function Z is:

[0079] Z = δ1Z1 + δ2Z2 + δ3Z3

[0080] Step Three: Determine the constraint conditions:

[0081] Order satisfaction constraint: The sum of the quantities allocated to each vehicle type by each warehouse at each stage should meet the predicted order volume of the corresponding vehicle type at that stage.

[0082]

[0083] Warehouse capacity constraint: The inventory level of each warehouse at each stage cannot exceed its maximum capacity.

[0084]

[0085] Cross-border transportation aging constraint: The aging of transporting vehicle types from each warehouse to the destination cannot exceed the aging threshold.

[0086]

[0087] Customs clearance risk constraint: A general acceptable level of customs clearance risk R max can be set to ensure that the sum of the customs clearance risk coefficients of all warehouses is within the acceptable range.

[0088]

[0089] Step 4. Optimization and solution:

[0090] Use the non-dominated sorting genetic algorithm to solve the above multi-objective optimization problem, specifically including:

[0091] Initialize the population: Randomly generate a set of initial solutions that satisfy all constraint conditions, that is, a series of combinations of x ijt values, as the initial population.

[0092] Calculate the fitness: For each individual in the population (i.e., a warehousing allocation and scheduling plan), calculate its logistics cost Z1, timeliness deviation Z2, and inventory turnover risk Z3 according to the above objective function, and then obtain the comprehensive objective function value Z. The smaller this value is, the better the fitness of the individual.

[0093] Genetic operations: Generate a new population through genetic operations such as selection, crossover, and mutation. In the selection operation, methods such as tournament selection are used to select individuals with better fitness to enter the next generation; the crossover operation generates new individuals by exchanging some genes of two individuals, that is, x ijt values; the mutation operation randomly changes some genes of the individual to increase the diversity of the population.

[0094] Iterative optimization: Repeat the steps of calculating the fitness and genetic operations. After multiple generations of iteration, the population gradually approaches the optimal solution. When certain termination conditions are met (such as the number of iterations reaches the set value, the convergence degree of the population meets the requirements, etc.), stop the iteration and output the optimal solution, that is, obtain the warehousing allocation and scheduling plan that meets the multi-objective optimization requirements.

[0095] S40. Based on the warehousing allocation and scheduling plan, execute the scheduling instructions through the blockchain smart contract, collect the actual logistics data and the predicted deviation value in real time, update the user portrait feature weights through the online learning mechanism, and use the Bayesian optimization algorithm to adjust the hyperparameters of the demand prediction model to optimize the demand prediction model.

[0096] The smart contract code needs to cover the key information in the warehousing allocation and scheduling plan, including detailed instructions on which warehouse to allocate which vehicle type, the quantity to be allocated, and the transportation destination, etc. Compile and test the written smart contract to ensure that the contract logic is correct, without loopholes and errors. Then deploy the smart contract to the blockchain network to make it effective in the distributed nodes of the blockchain. When the conditions for triggering the execution of the scheduling instructions are met (for example, reaching the preset shipping time or receiving a specific order confirmation signal), the system automatically calls the deployed blockchain smart contract. The smart contract sends accurate scheduling instructions to relevant logistics participants (such as the warehouse management system, the transportation company system, etc.) according to the specific arrangements in the warehousing allocation and scheduling plan. These instructions are disseminated through the decentralized network of the blockchain to ensure the immutability and accurate transmission of the instructions. After receiving the instructions, the logistics participants perform corresponding operations according to the requirements of the instructions, such as extracting vehicles from the designated warehouse and arranging transportation, etc. At the same time, the logistics participants feedback the confirmation information of the operation execution to the blockchain, and the smart contract records these feedback information to achieve the transparency and traceability of the entire scheduling process.

[0097] Deploy data collection devices or access the data interfaces of relevant systems at key nodes in logistics transportation (such as warehouse shipping points, important transfer stations during transportation, destination receiving points, etc.). For example, when shipping from the warehouse, automatically collect data such as the vehicle type, quantity, and time of actual shipment through the warehouse management system; during transportation, use Internet of Things devices (such as in-vehicle GPS devices) to collect real-time data such as the vehicle's location and driving speed, which can be used to calculate the actual transportation timeliness. In the order processing link, collect information such as the actual transaction time and customer feedback. For example, record the difference between the actual customer order time and the predicted transaction time window, and the difference between the actual customer demand for vehicle type and the predicted vehicle type demand distribution.

[0098] In the data processing center, compare the real-time collected actual logistics data with the prediction results generated by the previous demand prediction model.

[0099] For the order volume, calculate the difference between the actual order volume and the predicted order volume, that is, the deviation value = actual order volume - predicted order volume, and take the absolute value. For the vehicle type demand distribution, calculate the difference between the actual demand proportion and the predicted demand proportion of each vehicle type. For example, if the actual demand proportion of SUV models is 45% and the predicted proportion is 40%, then the difference is 45% - 40% = 5%. For the trading time window, calculate the time difference between the actual trading time and the predicted trading time window. For example, the actual trading time is 2 hours later than the predicted time. Organize and store these calculated prediction deviation values for subsequent model optimization. When updating the feature weights of the user profile through the online learning mechanism, the Stochastic Gradient Descent (SGD) algorithm can be used as the basic algorithm of the online learning mechanism. The SGD algorithm can update the model parameters each time it receives a new data sample (i.e., the actual logistics data and the prediction deviation value collected in real time), and is suitable for online learning scenarios dealing with large-scale data. Finally, use the Bayesian optimization algorithm to adjust the hyperparameters of the demand prediction model.

[0100] In summary, the intelligent automobile sales order management method based on cross-border trade provided by the present invention can fully exploit market demand information in a complex environment by obtaining a cross-border trade historical order data set to construct a multi-dimensional user profile model and inputting its feature vector, combined with external market trend data and tariff policy change factors, into the demand prediction model, so as to accurately predict the order volume, vehicle type demand distribution matrix, and trading time window probability distribution within a preset future period, providing a solid and reliable data basis for subsequent decision-making. In terms of warehouse allocation and scheduling, a multi-objective optimization function including logistics cost, timeliness deviation, and inventory turnover risk is established based on the prediction results, and optimization is solved with constraints such as warehouse capacity, transportation timeliness, and customs clearance risk. The generated warehouse allocation and scheduling plan can comprehensively balance various factors, realize the optimal allocation of resources, significantly reduce costs, and improve operation efficiency. By using blockchain smart contracts to execute scheduling instructions, collecting actual logistics data and prediction deviation values in real time, updating the feature weights of the user profile through the online learning mechanism, and adjusting the hyperparameters of the demand prediction model using the Bayesian optimization algorithm, the entire order management solution has the ability of dynamic optimization, can closely follow the changes in the cross-border trade environment, and continuously maintain a high level of order management.

[0101] In one embodiment, before executing the scheduling instruction through the blockchain smart contract, it further includes:

[0102] Input the prediction result output by the demand prediction model into the digital twin simulation system to simulate the load status of the logistics network under different warehouse allocation schemes, and obtain the congestion probability of the key nodes of the network load.

[0103] When the congestion probability is greater than a preset threshold, a scheduling plan including alternative transportation routes and emergency warehouses is regenerated to generate a new scheduling instruction.

[0104] Specifically, the predicted order volume, vehicle type demand distribution matrix, trading time window probability distribution, and other prediction results generated by the demand prediction model are accurately imported into the system in the data format required by the digital twin simulation system. In the digital twin simulation system, a virtual model highly similar to the actual logistics network is constructed, covering key elements such as each warehouse, transportation route, and transportation tool. Based on the input prediction results, the system simulates the flow of goods in the logistics network under different warehousing allocation plans, thereby calculating the congestion probability of key nodes in the network load (such as transportation hubs, busy warehouses, etc.).

[0105] Furthermore, a preset threshold is set (for example, the congestion probability reaches 30%), and when the congestion probability simulated by the system is greater than this preset threshold, the system automatically triggers an adjustment mechanism. The algorithm is used to re-plan the transportation route, screen out alternative transportation routes, and determine emergency warehouses at the same time. For example, if a certain conventional transportation route has a too high congestion probability in the simulation, the algorithm can search and recommend other feasible routes according to historical traffic data and real-time road conditions; the emergency warehouse is selected based on factors such as geographical location, warehouse capacity, and distance from key nodes. Based on these new plans, a new scheduling plan including alternative transportation routes and emergency warehouses is generated, and then a new scheduling instruction is generated.

[0106] Therefore, through the above embodiments, by simulating and adjusting in advance, the congestion situation that may occur at key nodes of the logistics network is avoided, ensuring that goods can be transported more smoothly and efficiently, reducing transportation delays, and improving the overall logistics efficiency. When encountering potential congestion risks, alternative transportation routes and emergency warehouses are enabled in a timely manner, enhancing the ability of the logistics system to respond to emergencies and improving the stability and reliability of the entire system. At the same time, it can effectively reduce the additional transportation costs caused by congestion, such as fuel consumption due to increased vehicle waiting time and rising labor costs, and rationally utilize the resources of emergency warehouses to avoid unnecessary resource waste, thereby reducing the operating costs of enterprises.

[0107] See Figure 2 , in a certain embodiment of the present invention, an intelligent car sales order management system based on cross-border trade is further provided. The system includes:

[0108] A data collection unit 100, configured to obtain a cross-border trade historical order data set, cluster the historical order data set to construct a multi-dimensional user portrait model, and extract user portrait feature vectors according to the multi-dimensional user portrait model;

[0109] A demand forecasting unit 200, which is used to perform feature fusion on the user portrait feature vector, external market trend data, and tariff policy change factors, input the fused features into a trained demand forecasting model to generate a forecasting result, where the forecasting result includes the order quantity forecast value, vehicle model demand distribution matrix, and transaction time window probability distribution within a preset future period;

[0110] A scheduling and allocation unit 300, which is used to establish a multi-objective optimization function based on the forecasting result, including establishing an objective function according to logistics cost, timeliness deviation, and inventory turnover risk, and using the maximum capacity of each warehouse, cross-border transportation timeliness threshold, and customs clearance risk coefficient as constraints; perform optimization and solution on the objective function to generate a warehousing allocation and scheduling plan;

[0111] A model iteration unit 400, which is used to execute scheduling instructions through a blockchain smart contract based on the warehousing allocation and scheduling plan, collect actual logistics data and forecast deviation values in real time, update the feature weights of the user portrait through an online learning mechanism, and adjust the hyperparameters of the demand forecasting model using the Bayesian optimization algorithm to optimize the demand forecasting model.

[0112] See Figure 3 , in one embodiment, the intelligent vehicle sales order management system based on cross-border trade further includes an instruction simulation sheet 500, which is used for:

[0113] Input the forecasting result output by the demand forecasting model into a digital twin simulation system, simulate the logistics network load status under different warehousing allocation schemes, and obtain the congestion probability of the key nodes of the network load;

[0114] When the congestion probability is greater than a preset threshold, regenerate a scheduling plan including alternative transportation routes and emergency warehouses to generate a new scheduling instruction.

[0115] Preferably, the data set includes user attribute data, transaction time series data, vehicle model configuration parameters, and cross-border logistics node information.

[0116] Preferably, the output dimensions of the model multi-dimensional user portrait model include user purchasing power grading indicators, brand preference coefficients, vehicle model iteration response cycles, and regional policy sensitivity parameters.

[0117] It can be understood that the functions or modules included in the system provided in this embodiment can be used to execute the methods described in the above method embodiments, and its specific implementation can refer to the description of the above method embodiments. For the sake of brevity, it will not be repeated here.

[0118] The present invention also provides an electronic device, including a processor and a memory. The memory is used for storing computer program code, and the computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes the method in any of the above possible implementation manners.

[0119] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor is caused to execute the method in any of the above possible implementation manners.

[0120] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

Claims

1. An intelligent management method for automobile sales orders based on cross-border trade, characterized in that, The method includes: Obtaining a cross-border trade historical order data set, clustering the historical order data set to construct a multi-dimensional user portrait model, and extracting a user portrait feature vector according to the multi-dimensional user portrait model; Performing feature fusion on the user portrait feature vector with external market trend data and tariff policy change factors, and inputting the fused features into a trained demand prediction model to generate a prediction result, where the prediction result includes an order volume prediction value, a vehicle model demand distribution matrix, and a transaction time window probability distribution within a preset future period; Establishing a multi-objective optimization function based on the prediction result, including establishing an objective function according to logistics cost, timeliness deviation, and inventory turnover risk, and using the maximum capacity of each warehouse, cross-border transportation timeliness threshold, and customs clearance risk coefficient as constraint conditions; performing optimization solution on the objective function to generate a warehousing allocation and scheduling plan; Based on the warehousing allocation and scheduling plan, executing the scheduling instruction through a blockchain smart contract, collecting actual logistics data and prediction deviation values in real time, updating the user portrait feature weights through an online learning mechanism, and using the Bayesian optimization algorithm to adjust the hyperparameters of the demand prediction model to optimize the demand prediction model.

2. The intelligent automobile sales order management method based on cross-border trade according to claim 1, wherein, Before executing the scheduling instruction through the blockchain smart contract, it further includes: Inputting the prediction result output by the demand prediction model into a digital twin simulation system to simulate the logistics network load status under different warehousing allocation plans, and obtaining the congestion probability of key network load nodes; When the congestion probability is greater than a preset threshold, regenerating a scheduling plan including alternative transportation routes and emergency warehouses to generate a new scheduling instruction.

3. The intelligent automobile sales order management method based on cross-border trade according to claim 1, wherein The data set includes user attribute data, transaction time series data, vehicle model configuration parameters, and cross-border logistics node information.

4. The intelligent automobile sales order management method based on cross-border trade according to claim 1, characterized in that The output dimensions of the multi-dimensional user portrait model include user purchasing power grading indicators, brand preference coefficients, vehicle model iteration response cycles, and regional policy sensitivity parameters.

5. An intelligent management system for automobile sales orders based on cross-border trade, characterized in that, The system includes: A data acquisition unit for obtaining a cross-border trade historical order data set, clustering the historical order data set to construct a multi-dimensional user portrait model, and extracting a user portrait feature vector according to the multi-dimensional user portrait model; A demand prediction unit for performing feature fusion on the user portrait feature vector with external market trend data and tariff policy change factors, and inputting the fused features into a trained demand prediction model to generate a prediction result, where the prediction result includes an order volume prediction value, a vehicle model demand distribution matrix, and a transaction time window probability distribution within a preset future period; A scheduling and allocation unit for establishing a multi-objective optimization function based on the prediction result, including establishing an objective function according to logistics cost, timeliness deviation, and inventory turnover risk, and using the maximum capacity of each warehouse, cross-border transportation timeliness threshold, and customs clearance risk coefficient as constraint conditions; performing optimization solution on the objective function to generate a warehousing allocation and scheduling plan; A model iteration unit for, based on the warehousing allocation and scheduling plan, executing the scheduling instruction through a blockchain smart contract, collecting actual logistics data and prediction deviation values in real time, updating the user portrait feature weights through an online learning mechanism, and using the Bayesian optimization algorithm to adjust the hyperparameters of the demand prediction model to optimize the demand prediction model.

6. The intelligent automobile sales order management system based on cross-border trade according to claim 5, wherein It further includes an instruction emulation unit for: inputting the prediction result output by the demand prediction model into the digital twin simulation system, simulating the logistics network load status under different warehousing allocation schemes, and obtaining the congestion probability of the key nodes of the network load; when the congestion probability is greater than a preset threshold, regenerating a scheduling scheme including alternative transportation routes and emergency warehouses to generate new scheduling instructions.

7. The intelligent automobile sales order management system based on cross-border trade according to claim 5, characterized in that The data set includes user attribute data, transaction time series data, automobile model configuration parameters, and cross-border logistics node information.

8. The intelligent automobile sales order management system based on cross-border trade according to claim 5, wherein The output dimensions of the model multi-dimensional user portrait model include user purchasing power grading indicators, brand preference coefficients, vehicle model iteration response cycles, and regional policy sensitivity parameters.

9. An electronic device, characterized in that, It includes: a processor and a memory, the memory is used to store computer program code, the computer program code includes computer instructions, and when the processor executes the computer instructions, the electronic device executes the intelligent automobile sales order management method based on cross-border trade as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, A computer program is stored in the computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by the processor of the electronic device, the processor is made to execute the intelligent automobile sales order management method based on cross-border trade as described in any one of claims 1 to 4.

Images