ERP (Enterprise Resource Planning) management system and method applying information technology

Abstract

The invention discloses an ERP (Enterprise Resource Planning) management system and method applying an information technology, and relates to the technical field of information, and the method comprises the steps: obtaining the detailed information of a sales order; according to the detailed information of the sales order, automatically identifying a required customs declaration document, and extracting related files; according to the detailed information of the sales order, collecting related data, extracting features, and constructing a customs declaration success rate prediction model; according to the prediction result of the customs declaration success rate prediction model, the historical customs declaration data, the declaration cargo information, the customs declaration document and the cargo features, constructing a risk assessment model, and generating a risk level; and displaying the sales order customs declaration success rate, and generating risk early warning and reminding. According to the method, the customs declaration success rate prediction model and the risk assessment model are constructed, the accuracy of customs declaration success rate prediction and risk assessment is improved by using the XGBoost algorithm, and a user is reminded to take proper prevention measures, so that the probability of customs declaration failure is reduced.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an ERP management system and method using information technology. Background Art

[0002] With the continuous development of global trade, cross-border trade activities are increasing day by day. When conducting international trade, it is necessary to comply with the customs regulations and customs declaration procedures of various countries. However, due to the complexity of customs declaration policies, differences in laws and regulations of various countries, and the diversity of logistics and transportation methods, problems such as incomplete customs declaration documents, inaccurate declaration information, and mismatched cargo descriptions are often encountered during the customs declaration process. These problems increase the risk of customs declaration failure.

[0003] The traditional customs declaration management model relies on manual operation and experience judgment, and cannot accurately evaluate the customs declaration success rate and risk of each sales order in real time. This traditional model is particularly inefficient when faced with complex and massive customs declaration data. Especially when faced with goods of different types and countries, the efficiency and accuracy of manual processing are greatly limited. Summary of the invention

[0004] The purpose of the present invention is to provide an ERP management system and method using information technology to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solution: an ERP management method using information technology, the method comprising the steps of: S1. Get detailed information of sales order; S2. Based on the detailed information of the sales order, automatically identify the required customs declaration documents using a machine learning algorithm and extract relevant documents; S3. According to the detailed information of the sales order, collect relevant data, perform data cleaning and feature extraction, and build a customs declaration success rate prediction model; S4. Construct a risk assessment model and generate a risk level based on the prediction results of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and goods characteristics; S5. Display the customs declaration success rate and risk level of the sales order; when the risk level is low risk, medium risk or high risk, issue corresponding risk warnings and reminders; send risk warnings and operation reminders to users through the system interface, and the operation reminders include the customs declaration success rate, influencing factors and specific operation suggestions.

[0006] According to the above scheme, the detailed information includes the HS code, type of goods, quantity, price, origin, packaging and mode of transportation of the goods; the relevant documents include invoice, packing list and certificate of origin.

[0007] According to the above scheme, the relevant data includes historical customs declaration data, declared goods information, customs declaration documents and goods characteristics; The historical customs declaration data includes the type and quantity of goods and the record of successful and failed customs declarations; the declared goods information includes the HS code of the goods, the description of the declared goods, the quantity and the price; the customs declaration documents include invoices, packing lists and certificates of origin; the characteristics of the goods include whether the goods are dangerous or perishable, the mode of transportation and the place of origin; Performing data cleaning on the relevant data; the data cleaning includes using mean filling to process missing values, using outlier detection to identify and process abnormal data, and performing data standardization; The characteristics include historical customs declaration data, accuracy of declared cargo information, completeness of documents and cargo characteristics; The accuracy of the cargo information is to compare the accuracy of the actual description of the cargo with the accuracy of the declared cargo information, and create a binary feature. When the actual description of the cargo is accurate with the declared cargo information, the feature is marked as 1, and when the actual description of the cargo is inaccurate with the declared cargo information, the feature is marked as 0; The document integrity is to analyze whether the customs declaration document is complete and create a binary feature. When the customs declaration document is complete, the feature is marked as 1, and when the customs declaration document is incomplete, the feature is marked as 0; The cargo features are encoded using one-hot encoding to convert the category features into binary features; Construct a feature matrix, which is expressed as follows: ; Among them, X represents the feature matrix; m represents the number of samples; n represents the number of features; Represented as the nth feature of the mth sample.

[0008] According to the above scheme, the customs declaration success rate prediction model uses the XGBoost algorithm to build a model based on the gradient boosting tree, and in each iteration, the decision tree is used to fit the data; The XGBoost algorithm optimizes the parameters of each decision tree by minimizing the loss function. The loss function has the following formula: ; Among them, L(θ) represents the loss function; N represents the number of samples; Represents the loss function for each sample; y i Represented as the true label; Represents the prediction result; i represents the index of the sample number; f k represents the kth decision tree; k represents the index of the decision tree; K represents the total number of decision trees; Represented as a regularization term, it is used to control the complexity of the tree and prevent overfitting; The core of the XGBoost algorithm is to form a strong classifier through multiple decision trees, with the goal of minimizing the overall loss function; The regularization term is formulated as follows: ; Among them, T k Represents the number of leaf nodes of the kth decision tree; w j represents the weight of the jth leaf node of the kth decision tree; j represents the index of the number of leaf nodes; γ and λ represent the regularization parameters for controlling the complexity of the customs declaration success rate prediction model; The XGBoost algorithm implicitly learns the importance of features by splitting nodes of a decision tree; the importance of features includes the information gain brought by each feature, the number of times each feature is split in the decision tree, and the frequency of each feature appearing in the number of splits.

[0009] According to the above scheme, the XGBoost algorithm adopts a step-by-step learning strategy to control the impact of the new decision tree at each iteration by adjusting the learning rate; the prediction update rule for each iteration is as follows: ; in, Represented as the prediction result after the tth iteration; It represents the prediction result after the last iteration; t represents the number of iterations; η represents the learning rate; Represented as the feature vector x of the decision tree added in the tth iteration i Prediction of x i Represented as the feature vector of the i-th sample; Set the maximum depth of the decision tree to control the complexity of the decision tree and prevent overfitting; During the iteration process, the model parameters are adjusted through cross-validation, and the model parameters include the number of decision trees, the learning rate and the maximum depth.

[0010] According to the above scheme, the original output of the customs declaration success rate prediction model is a regression value. The probability of each output sample belonging to a different category is converted into a customs declaration success rate through the sigmoid function. The formula is as follows: ; in, It is expressed as the probability that the sample belongs to the category of successful customs declaration; Represents the original prediction results obtained by the XGBoost algorithm.

[0011] According to the above scheme, the risk assessment model calculates the risk assessment score based on the prediction results of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and goods characteristics. The formula is as follows: ; Among them, S represents the risk assessment score; P represents the prediction result of the customs declaration success rate prediction model; H represents the frequency of successful or failed customs declaration of the same type of goods judged by historical customs declaration data; C represents the binary feature generated for the accuracy of the declared goods information; B represents the binary feature generated for the completeness of the customs declaration documents; T represents the binary feature converted from the goods feature by one-hot encoding; w 1 、w 2 、w 3 、w 4 and w 5 They are respectively represented as the prediction result of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and scoring weights of goods characteristics; the scoring weights are automatically obtained and optimized through the training of the risk assessment model; To avoid excessive influence of a certain feature on the final risk assessment, the features are standardized to ensure that the influence of each feature is balanced. The formula is as follows: ; Among them, Standardized Feature z It represents the eigenvalue after normalization; z represents the index of different features; Feature z Represented as the original value of the zth feature; μ z Expressed as the mean of the z-th feature; σ z Expressed as the standard deviation of the zth feature; The risk assessment model outputs a risk level, which includes no risk, low risk, medium risk and high risk; a success rate greater than 90% is no risk, a success rate between 70% and 90% is low risk, a success rate between 60% and 70% is medium risk, and a success rate between 0% and 60% is high risk; According to the above scheme, when the risk assessment model evaluates that there is no risk, the normal customs declaration process will be carried out; When the risk assessment model evaluates to be low risk, medium risk or high risk, the risk warning and reminder are carried out, and a warning notification is sent to the user through the system visualization interface. The warning notification includes the risk assessment model evaluation result, the prediction result of the customs declaration success rate prediction model, the sales order number and the factors affecting the success rate; the factors affecting the success rate include the historical customs declaration data of the same type of goods, the accuracy of the declaration information, the completeness of the documents and the characteristics of the goods; The reminder generates an operation reminder based on the analysis results of the customs declaration success rate prediction model and the risk assessment model. The operation reminder includes that the historical customs declaration success rate of the same type of goods is low, the invoice, packing list or certificate of origin in the customs declaration documents is incomplete or does not meet the requirements, the description of the declared goods does not match the actual goods, and the declared goods contain dangerous or perishable goods; when there are dangerous or perishable goods in the declared goods, the user is specially reminded to pay attention to the relevant special requirements and customs declaration difficulties; when the historical customs declaration success rate of the same type of goods is low, the user is specially reminded to pay attention and the historical customs declaration success rate of the same type of goods is displayed; According to the above scheme, when a user submits or updates customs declaration materials, a re-analysis is performed based on the newly submitted customs declaration materials, and the customs declaration success rate prediction and risk assessment results are updated. Based on the new customs declaration success rate prediction and risk assessment results, risk warnings and reminders are re-generated to help users adjust and optimize customs declaration materials or customs declaration goods in a timely manner; the customs declaration materials include invoices, packing lists, contracts and certificates of origin; Real-time tracking of the customs declaration status of each sales order, including customs declaration submission, under review, order return, customs declaration success and customs declaration failure; When the customs declaration status of a sales order is under review, the success rate of the sales order is predicted in combination with historical data; when the customs declaration status of a sales order is customs declaration failure, the customs declaration success rate prediction model and risk assessment model are optimized according to the reasons for the failure of the sales order, and the risk assessment is re-performed to generate a new risk assessment, and the user is reminded to take corrective measures; Based on new customs declaration data and customs clearance results, the customs declaration prediction model and risk scoring model are regularly updated and optimized to ensure that the model adapts to environmental changes and regulatory updates; the model is continuously trained and optimized to enhance the predictive capabilities for different cargo types, declaration requirements and customs declaration success rates.

[0012] An ERP management system using information technology, the system comprising: a sales order management module, a customs declaration data identification module, a customs declaration success rate prediction module, a risk assessment module and a visualization module; The sales order management module is used to manage and track sales orders, analyze and extract detailed information of sales orders; the detailed information includes HS code of goods, type of goods, quantity, price, origin, packaging and mode of transportation; and provide real-time update and monitoring of the status of sales orders; The customs declaration data identification module automatically identifies whether the sales order needs customs declaration. When the sales order needs customs declaration, it automatically identifies the required customs declaration documents and extracts relevant documents; the relevant documents include invoices, packing lists and certificates of origin; when the sales order does not need customs declaration, the system directly completes the sales outbound process; The customs declaration success rate prediction module constructs a feature matrix based on historical customs declaration data, accuracy of declared goods information, document integrity and goods characteristics, and uses the XGBoost algorithm to construct a customs declaration success rate prediction model for predicting the customs declaration success rate of the sales order; The risk assessment module constructs a risk assessment model based on the prediction result of the customs declaration success rate prediction model, the declared goods information, the customs declaration documents and the goods characteristics, conducts risk assessment on the sales order, and assesses the customs declaration risk level of the sales order, wherein the risk level includes no risk, low risk, medium risk and high risk; The visualization module displays the customs declaration success rate prediction results and risk levels of the sales order in real time; when the risk assessment model assessment result is low risk, medium risk or high risk, a risk warning and reminder are triggered, and the warning notification includes the risk assessment model assessment result, the prediction result of the customs declaration success rate prediction model, the sales order number and the factors affecting the success rate; the factors affecting the success rate include historical customs declaration data of the same type of goods, accuracy of declaration information, completeness of documents and characteristics of goods; the reminder generates an operation reminder based on the analysis results of the customs declaration success rate prediction model and the risk assessment model, and the operation reminder includes the following: the historical customs declaration success rate of the same type of goods is low, the invoice, packing list or certificate of origin in the customs declaration documents is incomplete or does not meet the requirements, the description of the declared goods does not match the actual goods, and there are dangerous or perishable goods in the declared goods; when there are dangerous or perishable goods in the declared goods, the user is especially reminded to pay attention to the relevant special requirements and customs declaration difficulties.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention constructs a customs declaration success rate prediction model and a risk assessment model to perform customs declaration success rate prediction and risk assessment, complete customs declaration early warning and reminder, and improve the customs declaration pass rate; 2. The present invention uses the XGBoost algorithm to improve the prediction of the success rate of customs declaration and remind users to take appropriate preventive measures, thereby reducing the probability of customs declaration failure; 3. The present invention provides real-time customs declaration status tracking, customs declaration risk warning and reminder through a visual interface, so that users can take immediate measures and intuitively understand the customs declaration status. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flowchart of the steps of an ERP management method using information technology according to the present invention; Figure 2 The present invention is a schematic diagram of the structure of an ERP management system using information technology. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, an ERP management method using information technology; Obtain detailed information of the sales order; the detailed information includes the HS code, type of goods, quantity, price, origin, packaging and transportation method of the goods; Assume that the detailed information is: Sales order number: 1234567; Goods HS code: 12345678; Goods type: tools; Quantity: 500 units; Price: 50,000 yuan; Origin: China; Packaging: carton; Mode of transportation: sea transportation; automatically identifying the required customs declaration documents and extracting relevant documents based on the detailed information of the sales order using a machine learning algorithm; the relevant documents include an invoice, a packing list, and a certificate of origin; The relevant documents are invoice: invoice number: 123456789, amount: 5,000 yuan, description of goods: tools; packing list: contains 500 tools, carton packaging, volume: 10 cubic meters; certificate of origin: issued by the Chinese Chamber of Commerce, certifying that the goods are of Chinese origin; According to the detailed information of the sales order, collect relevant data, perform data cleaning and feature extraction, and build a customs declaration success rate prediction model; The relevant data include historical customs declaration data, declared goods information, customs declaration documents and goods characteristics; Historical customs declaration data: Based on the past 200 order data, the successful customs declaration rate is 85% and the failed customs declaration rate is 15%; Declared goods information: including HS code, description, quantity and price; Customs declaration document information: including the integrity of invoice, packing list and certificate of origin; Cargo characteristics: The cargo is tools, the mode of transportation is sea transportation, and the country of origin is China; Performing data cleaning on the relevant data; the data cleaning includes using mean filling to process missing values, using outlier detection to identify and process abnormal data, and performing data standardization; The characteristics include historical customs declaration data, accuracy of declared cargo information, completeness of documents and cargo characteristics; The accuracy of the cargo information is to compare the accuracy of the actual description of the cargo with the accuracy of the declared cargo information, and create a binary feature. When the actual description of the cargo is accurate with the declared cargo information, the feature is marked as 1, and when the actual description of the cargo is inaccurate with the declared cargo information, the feature is marked as 0; The document integrity is to analyze whether the customs declaration document is complete and create a binary feature. When the customs declaration document is complete, the feature is marked as 1, and when the customs declaration document is incomplete, the feature is marked as 0; The cargo features are encoded using one-hot encoding to convert the category features into binary features; Construct a feature matrix, which is expressed as follows: ; Among them, X represents the feature matrix; m represents the number of samples; n represents the number of features; Represented as the nth feature of the mth sample.

[0017] The customs declaration success rate prediction model uses the XGBoost algorithm to build a model based on the gradient boosting tree. In each iteration, the decision tree is used to fit the data; The XGBoost algorithm optimizes the parameters of each decision tree by minimizing the loss function. The loss function has the following formula: ; Among them, L(θ) represents the loss function; N represents the number of samples; Represents the loss function for each sample; y i Represented as the true label; Represents the prediction result; i represents the index of the sample number; f k represents the kth decision tree; k represents the index of the decision tree; K represents the total number of decision trees; Represented as a regularization term, it is used to control the complexity of the tree and prevent overfitting; The core of the XGBoost algorithm is to form a strong classifier through multiple decision trees, with the goal of minimizing the overall loss function; The regularization term is formulated as follows: ; Among them, T k Represents the number of leaf nodes of the kth decision tree; w j represents the weight of the jth leaf node of the kth decision tree; j represents the index of the number of leaf nodes; γ and λ represent the regularization parameters for controlling the complexity of the customs declaration success rate prediction model; The XGBoost algorithm implicitly learns the importance of features by splitting nodes of a decision tree; the importance of features includes the information gain brought by each feature, the number of times each feature is split in the decision tree, and the frequency of each feature appearing in the number of splits.

[0018] The XGBoost algorithm adopts a step-by-step learning strategy, which controls the impact of the new decision tree at each iteration by adjusting the learning rate; the prediction update rule for each iteration is as follows: ; in, Represented as the prediction result after the tth iteration; It represents the prediction result after the last iteration; t represents the number of iterations; η represents the learning rate; Represented as the feature vector x of the decision tree added in the tth iteration i Prediction of x i Represented as the feature vector of the i-th sample; Set the maximum depth of the decision tree to control the complexity of the decision tree and prevent overfitting; During the iteration process, the model parameters are adjusted through cross-validation, and the model parameters include the number of decision trees, the learning rate and the maximum depth.

[0019] The original output of the customs declaration success rate prediction model is a regression value. The probability of each output sample belonging to a different category is converted into a customs declaration success rate through the sigmoid function. The formula is as follows: ; in, It is expressed as the probability that the sample belongs to the category of successful customs declaration; Represents the original prediction results obtained by the XGBoost algorithm.

[0020] According to the customs clearance success rate prediction model, the customs clearance success rate is predicted to be 85%; Based on the prediction results, a risk assessment model is constructed to calculate the risk score of the sales order and generate a risk level based on the score; The risk assessment model calculates the risk assessment score based on the prediction results of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and goods characteristics. The formula is as follows: ; Among them, S represents the risk assessment score; P represents the prediction result of the customs declaration success rate prediction model; H represents the frequency of successful or failed customs declaration of the same type of goods judged by historical customs declaration data; C represents the binary feature generated for the accuracy of the declared goods information; B represents the binary feature generated for the completeness of the customs declaration documents; T represents the binary feature converted from the goods feature by one-hot encoding; w 1 、w 2 、w 3 、w 4 and w 5 They are respectively represented as the prediction result of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and scoring weights of goods characteristics; the scoring weights are automatically obtained and optimized through the training of the risk assessment model; To avoid excessive influence of a certain feature on the final risk assessment, the features are standardized to ensure that the influence of each feature is balanced. The formula is as follows: ; Among them, Standardized Feature z It represents the eigenvalue after normalization; z represents the index of different features; Feature z Represented as the original value of the zth feature; μ z Expressed as the mean of the z-th feature; σ z Expressed as the standard deviation of the zth feature; The risk assessment model outputs a risk level, which includes no risk, low risk, medium risk and high risk; a success rate greater than 90% is no risk, a success rate between 70% and 90% is low risk, a success rate between 60% and 70% is medium risk, and a success rate between 0% and 60% is high risk; The final risk assessment score was 0.86, and the risk level was low risk; According to the evaluation results, the customs clearance success rate of the sales order is displayed, and corresponding risk warnings and operation reminders are generated; risk warnings and operation reminders are sent to users through the system interface, and the operation reminders include the customs clearance success rate, influencing factors and specific operation suggestions; When the risk assessment model evaluates a low risk, medium risk or high risk, the risk warning and reminder are performed, and a warning notification is sent to the user through the system visualization interface; The early warning notice includes the evaluation results of the risk assessment model, the prediction results of the customs declaration success rate prediction model, the sales order number and the factors affecting the success rate; the factors affecting the success rate include the historical customs declaration data of the same type of goods, the accuracy of the declared information, the completeness of the documents and the characteristics of the goods; The reminder generates an operation reminder based on the analysis results of the customs declaration success rate prediction model and the risk assessment model. The operation reminder includes that the historical customs declaration success rate of the same type of goods is low, the invoice, packing list or certificate of origin in the customs declaration documents is incomplete or does not meet the requirements, the description of the declared goods does not match the actual goods, and the declared goods contain dangerous or perishable goods; when there are dangerous or perishable goods in the declared goods, the user is specially reminded to pay attention to the relevant special requirements and customs declaration difficulties; when the historical customs declaration success rate of the same type of goods is low, the user is specially reminded to pay attention and the historical customs declaration success rate of the same type of goods is displayed; Risk warning notice: Sales order number: 1234567; Risk assessment score: 0.86; Risk assessment: Low risk; Customs declaration success rate prediction: Success rate 85% Influencing factors: Completeness of customs declaration documents; Operation reminder: Please verify as soon as possible whether the invoice, packing list or certificate of origin in the customs declaration documents are complete or meet the requirements.

[0021] The present invention provides another technical solution, an ERP management system using information technology, the system comprising: a sales order management module, a customs declaration data identification module, a customs declaration success rate prediction module, a risk assessment module and a visualization module; The sales order management module is used to manage and track sales orders, analyze and extract detailed information of sales orders; the detailed information includes HS code of goods, type of goods, quantity, price, origin, packaging and mode of transportation; and provide real-time update and monitoring of the status of sales orders; The customs declaration data identification module automatically identifies whether the sales order needs customs declaration. When the sales order needs customs declaration, it automatically identifies the required customs declaration documents and extracts relevant documents; the relevant documents include invoices, packing lists and certificates of origin; when the sales order does not need customs declaration, the system directly completes the sales outbound process; The customs declaration success rate prediction module constructs a feature matrix based on historical customs declaration data, accuracy of declared goods information, document integrity and goods characteristics, and uses the XGBoost algorithm to construct a customs declaration success rate prediction model for predicting the customs declaration success rate of the sales order; The risk assessment module constructs a risk assessment model based on the prediction result of the customs declaration success rate prediction model, the declared goods information, the customs declaration documents and the goods characteristics, conducts risk assessment on the sales order, and assesses the customs declaration risk level of the sales order, wherein the risk level includes no risk, low risk, medium risk and high risk; The visualization module displays the customs declaration success rate prediction results and risk levels of the sales order in real time; when the risk assessment model assessment result is low risk, medium risk or high risk, a risk warning and reminder are triggered, and the warning notification includes the risk assessment model assessment result, the prediction result of the customs declaration success rate prediction model, the sales order number and the factors affecting the success rate; the factors affecting the success rate include historical customs declaration data of the same type of goods, accuracy of declaration information, completeness of documents and characteristics of goods; the reminder generates an operation reminder based on the analysis results of the customs declaration success rate prediction model and the risk assessment model, and the operation reminder includes the following: the historical customs declaration success rate of the same type of goods is low, the invoice, packing list or certificate of origin in the customs declaration documents is incomplete or does not meet the requirements, the description of the declared goods does not match the actual goods, and there are dangerous or perishable goods in the declared goods; when there are dangerous or perishable goods in the declared goods, the user is especially reminded to pay attention to the relevant special requirements and customs declaration difficulties.

[0022] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. An ERP management method using information technology, characterized in that: The method comprises the steps of: S1. Get detailed information of sales order; S2. Automatically identify the required customs declaration documents based on the detailed information of the sales order and extract relevant documents; S3. According to the detailed information of the sales order, collect relevant data, extract features, construct a feature matrix, and use the XGBoost algorithm to build a customs declaration success rate prediction model; S4. Construct a risk assessment model and generate a risk level based on the prediction results of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and goods characteristics; S5. Display the customs declaration success rate and risk level of sales orders; When the risk level is low risk, medium risk or high risk, corresponding risk warnings and reminders are issued.

2. The ERP management method using information technology according to claim 1, characterized in that: The detailed information includes the HS code, type, quantity, price, origin, packaging and mode of transportation of the goods; The relevant documents include invoice, packing list and certificate of origin.

3. The ERP management method using information technology according to claim 1 is characterized in that: The relevant data include historical customs declaration data, declared goods information, customs declaration documents and goods characteristics; The historical customs declaration data includes the type and quantity of goods and the records of successful and failed customs declarations; The declared goods information includes the HS code, description, quantity and price of the goods; The customs declaration documents include invoice, packing list and certificate of origin; The characteristics of the goods include whether the goods are dangerous or perishable, the mode of transportation and the place of origin; Performing data cleaning on the relevant data; the data cleaning includes using mean filling to process missing values, using outlier detection to identify and process abnormal data, and performing data standardization; The characteristics include historical customs declaration data, accuracy of declared cargo information, completeness of documents and cargo characteristics; The accuracy of the declared cargo information, comparing the accuracy of the actual description of the cargo with the accuracy of the declared cargo information, creating a binary feature; The document integrity, analyzing whether the customs declaration documents are complete and creating binary features; The cargo features are encoded using one-hot encoding to convert the category features into binary features; The feature matrix is ​​constructed by combining the binary features of the accuracy of the declared cargo information, the completeness of the documents and the characteristics of the cargo.

4. The ERP management method using information technology according to claim 1, characterized in that: The customs declaration success rate prediction model uses the XGBoost algorithm to build a model based on the gradient boosting tree. In each iteration, the decision tree is used to fit the data; The XGBoost algorithm optimizes the parameters of each decision tree by minimizing the loss function. The loss function has the following formula: ； Among them, L(θ) represents the loss function; N represents the number of samples; Represents the loss function for each sample; y i Represented as the true label; Represents the prediction result; i represents the index of the sample number; f k represents the kth decision tree; k represents the index of the decision tree; K represents the total number of decision trees; Represented as a regularization term, it is used to control the complexity of the tree and prevent overfitting; The regularization term is formulated as follows: ； Among them, T k Represents the number of leaf nodes of the kth decision tree; w j represents the weight of the jth leaf node of the kth decision tree; j represents the index of the number of leaf nodes; γ and λ represent the regularization parameters for controlling the complexity of the customs declaration success rate prediction model; The XGBoost algorithm implicitly learns the importance of features by splitting nodes of a decision tree; the importance of features includes the information gain brought by each feature, the number of times each feature is split in the decision tree, and the frequency with which each feature appears in the split of the number.

5. The ERP management method using information technology according to claim 4 is characterized by: The XGBoost algorithm adopts a step-by-step learning strategy, which controls the impact of the new decision tree at each iteration by adjusting the learning rate; the prediction update rule for each iteration is as follows: ； in, Represented as the prediction result after the tth iteration; It represents the prediction result after the last iteration; t represents the number of iterations; η represents the learning rate; Represented as the feature vector x of the decision tree added in the tth iteration i Prediction of x i Represented as the feature vector of the i-th sample; Set the maximum depth of the decision tree to control the complexity of the decision tree and prevent overfitting; In the iterative process, the parameters of the customs declaration success rate prediction model are adjusted through cross-validation, and the parameters of the customs declaration success rate prediction model include the number of decision trees, the learning rate and the maximum depth.

6. The ERP management method using information technology according to claim 5 is characterized by: The original output of the customs declaration success rate prediction model is a regression value. The probability that each output sample belongs to a different category is converted into a customs declaration success rate through the sigmoid function.

7. The ERP management method using information technology according to claim 1, characterized in that: The risk assessment model calculates a risk assessment score based on the prediction results of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and goods characteristics. The formula is as follows: ； Wherein, S represents the risk assessment score; P represents the prediction result of the customs declaration success rate prediction model; H represents the frequency of successful or failed customs declaration of the same type of goods judged by historical customs declaration data; C represents the binary feature generated for the accuracy of the declared goods information; B represents the binary feature generated for the completeness of the customs declaration documents; T represents the binary feature converted from the goods feature by one-hot encoding; w1, w2, w3, w4 and w5 represent the prediction result of the customs declaration success rate prediction model, historical customs declaration data, declared goods information, customs declaration documents and the scoring weights of goods features respectively; the scoring weights are automatically obtained and optimized through the training of the risk assessment model; To avoid excessive influence of a certain feature on the final risk assessment, the features are standardized to ensure that the influence of each feature is balanced. The formula is as follows: ； Among them, Standardized Feature z It represents the eigenvalue after normalization; z represents the index of different features; Feature z Represented as the original value of the zth feature; μ z Expressed as the mean of the z-th feature; σ z Expressed as the standard deviation of the zth feature; The risk assessment model outputs a risk score and a risk level, wherein the risk level includes no risk, low risk, medium risk and high risk.

8. The ERP management method using information technology according to claim 7 is characterized by: When the risk assessment model evaluates that there is no risk, the normal customs declaration process will be carried out; When the risk assessment model evaluates a low risk, medium risk or high risk, the risk warning and reminder are performed, and a warning notification is sent to the user through the system visualization interface; The early warning notice includes the evaluation results of the risk assessment model, the prediction results of the customs declaration success rate prediction model, the sales order number and the factors affecting the success rate; The factors affecting the success rate include historical customs declaration data of similar goods, accuracy of declaration information, completeness of documents and characteristics of goods; The reminder generates an operation reminder based on the analysis results of the customs declaration success rate prediction model and the risk assessment model; The operation reminders include the low success rate of customs declaration for the same type of goods in the past, the invoice, packing list or certificate of origin in the customs declaration documents being incomplete or not meeting the requirements, the mismatch between the description of the declared goods and the actual goods, and the presence of dangerous or perishable goods in the declared goods; when there are dangerous or perishable goods in the declared goods, users are especially reminded to pay attention to relevant special requirements and difficulties in customs declaration.

9. The ERP management method using information technology according to claim 8, characterized in that: When a user submits or updates customs declaration materials, a re-analysis is performed based on the newly submitted customs declaration materials, and the customs declaration success rate prediction and risk assessment results are updated. Based on the new customs declaration success rate prediction and risk assessment results, risk warnings and reminders are re-generated to help users adjust and optimize customs declaration materials or customs declaration goods in a timely manner; the customs declaration materials include invoices, packing lists, contracts and certificates of origin; Real-time tracking of the customs declaration status of each sales order, including customs declaration submission, under review, order return, customs declaration success and customs declaration failure; When the customs declaration status of a sales order is under review, the success rate of the sales order is predicted based on historical data; When the customs declaration status of a sales order is customs declaration failure, the customs declaration success rate prediction model and risk assessment model are optimized according to the reasons for the failure of the sales order, and the risk assessment is re-performed to issue risk warnings and reminders to remind users to take corrective measures.

10. An ERP management system using information technology, characterized in that: The system includes: sales order management module, customs declaration data identification module, customs declaration success rate prediction module, risk assessment module and visualization module; The sales order management module is used to manage and track sales orders, analyze and extract detailed information of sales orders; the detailed information includes HS code, type of goods, quantity, price, origin, packaging and transportation method of goods; and provide real-time update and monitoring of sales order status; The customs declaration data identification module automatically identifies whether the sales order needs customs declaration. When the sales order needs customs declaration, it automatically identifies the required customs declaration documents and extracts relevant documents; the relevant documents include invoices, packing lists and certificates of origin; when the sales order does not need customs declaration, the system directly completes the sales outbound process; The customs declaration success rate prediction module constructs a feature matrix based on historical customs declaration data, accuracy of declared goods information, document integrity and goods characteristics, and uses the XGBoost algorithm to construct a customs declaration success rate prediction model for predicting the customs declaration success rate of the sales order; The risk assessment module constructs a risk assessment model based on the prediction result of the customs declaration success rate prediction model, the declared goods information, the customs declaration documents and the goods characteristics, conducts risk assessment on the sales order, and assesses the customs declaration risk level of the sales order, wherein the risk level includes no risk, low risk, medium risk and high risk; The visualization module displays the customs declaration success rate prediction results and risk levels of the sales order in real time; when the risk assessment model assessment result is low risk, medium risk or high risk, a risk warning and reminder are triggered, and the warning notification includes the risk assessment model assessment result, the prediction result of the customs declaration success rate prediction model, the sales order number and the factors affecting the success rate; the factors affecting the success rate include historical customs declaration data of the same type of goods, accuracy of declaration information, completeness of documents and characteristics of goods; the reminder generates an operation reminder based on the analysis results of the customs declaration success rate prediction model and the risk assessment model, and the operation reminder includes the following: the historical customs declaration success rate of the same type of goods is low, the invoice, packing list or certificate of origin in the customs declaration documents is incomplete or does not meet the requirements, the description of the declared goods does not match the actual goods, and there are dangerous or perishable goods in the declared goods; when there are dangerous or perishable goods in the declared goods, the user is especially reminded to pay attention to the relevant special requirements and customs declaration difficulties.

Images