Intelligent factory-side fraud behavior early warning method and system

Abstract

The invention provides an intelligent factory-side fraud behavior early warning method and system, and belongs to the technical field of data early warning. The method comprises the following steps: step 1, acquiring real-time production data of any employee on a production line, and acquiring historical production data of the employee according to an ERP (Enterprise Resource Planning) system; 2, preprocessing real-time production data and historical production data of all employees, and storing the data in a production database after standardization; step 3, screening and obtaining production type data of the production database according to any production parameter type, and identifying and extracting fraud-related key indexes and key features thereof according to the expert database; 4, constructing a basic model by using the key indexes and the key features thereof, taking the standard historical data of any employee as a training set, taking the standard real-time data as a test set, and training the basic model to generate an early warning model; and step 5, establishing a monitoring mechanism based on the early warning model, identifying real-time abnormal behaviors of the employees and performing early warning. The accuracy is improved, and decision support and management transparency are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of data early warning technology, and in particular to an intelligent factory-side fraud early warning method and system. Background Art

[0002] At present, with the continuous development of big data, the trend of intelligence is increasingly expanding to various fields and is also widely used in the industrial field. Usually, factories judge the output of employees' production lines by directly analyzing the production efficiency of employees through the big data of production results. However, this method may lead to employee fraud, resulting in inaccurate judgment of employee production efficiency and low overall production efficiency of the factory.

[0003] Therefore, the present invention proposes an intelligent factory-side fraud behavior early warning method and system. Summary of the invention

[0004] The present invention provides an intelligent factory-side fraud behavior early warning method and system, which is used to obtain real-time production data of employees on the production line, combine the employees' historical production data, build an early warning model, set early warning thresholds and judge whether employees have abnormal fraud behavior.

[0005] On the one hand, the present invention provides an intelligent factory-side fraud behavior early warning method, comprising: Step 1: Obtain the real-time production data of any employee on the production line, and obtain the historical production data of the employee according to the ERP system; Step 2: Preprocess the real-time production data and historical production data of all employees, obtain standard real-time data and standard historical data, and store them in the production database; Step 3: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database; Step 4: Build a basic model with key indicators and their key features, and use the standard historical data of any employee as a training set and the standard real-time data as a test set to train the basic model and generate an early warning model; Step 5: Establish a monitoring mechanism based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings.

[0006] On the other hand, obtain real-time production data of any employee on the production line, including: Configure a preset type of sensor on the target production equipment of the production line, and obtain real-time raw data of the production equipment in real time according to the sensor; The sensor is connected to an API connection device to obtain real-time raw data, and the real-time raw data is filtered according to any employee ID to obtain the real-time production data of the employee.

[0007] On the other hand, the historical production data of the employee is obtained according to the ERP system, including: Connect to the ERP system query interface according to the preset connection information; The employee ID is used as a query condition to generate an interface query statement, and the ERP system query interface is called to obtain the historical production data of the employee.

[0008] On the other hand, pre-process the real-time production data and historical production data of all employees, obtain standard real-time data and standard historical data and store them in the production database, including: Obtaining the real-time production data and historical production data of any production parameter of any employee, and processing the production data and historical production data in standardized units according to preset business rules matching the production parameters to obtain first real-time production data and first historical production data; Processing the first real-time production data and the first historical production data according to a normal distribution model, identifying abnormal values ​​greater than a normal threshold and replacing them with standard normal distribution values, and finally obtaining second real-time production data and second historical production data; Obtain corresponding time periods for the second real-time production data and the second historical production data, and perform standardized time processing on each of them according to preset standard time intervals to obtain standard real-time data and standard historical data with the same time interval; According to the preset database table structure, a production database is constructed, and standard real-time data and standard historical data are stored in the production database using employee ID as the identifier.

[0009] On the other hand, the production type data of the production database is obtained by screening according to any production parameter type, and the key indicators and key features related to fraud are identified and extracted according to the expert database, including: Using any production parameter type as a screening condition, obtaining production type data of the production parameter type in the production database, the production type data including standard historical data and standard real-time data; Based on the production parameter type name, search the expert database for the names of key fraud-related indicators with a similarity greater than a preset threshold; ;in, Indicates the degree of correlation between the production parameter type name and any key indicator name in the expert database, Indicates the ascii value of the i-th character in the production parameter name. The standard ASCII value of the key indicator name character. Indicates the preset weight of the ascii value of the i-th character in the production parameter name, Indicates the standard similarity between the key indicator name and the character origin ascii, The maximum character distance of the key indicator name; If the correlation between the production parameter type name and the key indicator name is greater than a preset threshold, the key indicator corresponding to the key indicator name is the key indicator of the production parameter type; The standard historical data of production parameter types are extracted according to the preset formula of key indicators to obtain the preset indicator data, and the key features of the preset indicator data are extracted using the feature matrix extraction method.

[0010] On the other hand, the key indicators and their key features are used to build a basic model, and the standard historical data of any employee is used as a training set, and the standard real-time data is used as a test set to train the basic model to generate an early warning model, including: The key indicator of any production parameter type is used as the basic model framework node, and the coefficient of the framework node is constructed based on the key characteristics corresponding to the key indicator. A complete basic model is constructed for all production parameter types, key indicators and their key characteristics; Obtain all data in the production database based on the employee ID of any employee, use the standard historical data as a training set, train the basic model, and obtain the first model indicator; Using the standard real-time data as a test set, retraining the first model to obtain the second model index, and adjusting the first model index parameter based on the second model index to obtain the final index; The final indicator is used to configure the first model to obtain the early warning model.

[0011] On the other hand, a monitoring mechanism is established based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings, including: Setting warning thresholds for different types of production parameters based on the pre-output of the warning model, and configuring the warning model into the production line system; If the absolute value deviation between the production parameter type and the standard production parameter type of any employee in the real-time production process of the production line is greater than or equal to the warning threshold, the abnormal behavior of the employee's key indicator matching is identified and fed back to the system; if the absolute value deviation between the production parameter type and the standard production parameter type of the employee in the real-time production process of the production line is less than the warning threshold, it is determined that there is no abnormal behavior.

[0012] In another aspect, the present invention provides an intelligent factory-side fraud early warning system, comprising: Data acquisition module: obtain the real-time production data of any employee's production line, and obtain the employee's historical production data based on the ERP system; Preprocessing module: preprocesses the real-time production data and historical production data of all employees, obtains standard real-time data and standard historical data and stores them in the production database; Feature module: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database; Model module: construct a basic model based on key indicators and their key features, and use the standard historical data of any employee as a training set and the standard real-time data as a test set to train the basic model and generate an early warning model; Monitoring module: Establish a monitoring mechanism based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings.

[0013] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides an intelligent factory-side fraud behavior early warning method and system, which is used to obtain real-time production data of employees on the production line, combine the employees' historical production data, build an early warning model, set early warning thresholds and judge whether employees have abnormal fraud behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 It is a flow chart of an intelligent factory-side fraud behavior early warning method provided by an embodiment of the present invention; Figure 2 It is a structural diagram of an intelligent factory-side fraud behavior early warning system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0017] Embodiment 1: like Figure 1 As shown, an intelligent factory-side fraud behavior early warning method provided by an embodiment of the present invention includes: Step 1: Obtain the real-time production data of any employee on the production line, and obtain the historical production data of the employee according to the ERP system; Step 2: Preprocess the real-time production data and historical production data of all employees, obtain standard real-time data and standard historical data, and store them in the production database; Step 3: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database; Step 4: Build a basic model with key indicators and their key features, and use the standard historical data of any employee as a training set and the standard real-time data as a test set to train the basic model and generate an early warning model; Step 5: Establish a monitoring mechanism based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings.

[0018] In this embodiment, the production line refers to an operating system in which raw materials are processed, assembled, and other processes to obtain finished products in accordance with a certain process flow and layout in industrial production.

[0019] In this embodiment, real-time production data refers to information about employees, equipment, and process parameters that is collected and recorded in real time during the production process, including production speed, output, working hours, equipment status, quality indicators, etc.

[0020] In this embodiment, the ERP system is an integrated software platform for managing the core business processes of an enterprise, such as finance, supply chain, production, sales, and human resources.

[0021] In this embodiment, historical production data refers to production-related information collected over a period of time in the past, including employee output, working hours, equipment usage, production efficiency, quality inspection results, etc.

[0022] In this embodiment, preprocessing refers to the steps of cleaning, converting and organizing the raw data during the data analysis or modeling process.

[0023] In this embodiment, the standard real-time data is real-time production data that has been preprocessed and standardized.

[0024] In this embodiment, the standard historical data is real-time historical data that has been preprocessed and standardized.

[0025] In this embodiment, the production database is a system dedicated to storing and managing production-related data.

[0026] In this embodiment, the production parameter type refers to various indicators used to measure and evaluate the production process, including: output, efficiency, quality indicators, equipment status, etc.

[0027] In this embodiment, production type data refers to various data sets related to specific production activities or processes, including: output, efficiency, quality indicators, equipment status and other data.

[0028] In this embodiment, the expert database is a database containing professional knowledge, experience and best practices, and the content includes: professional knowledge and standard values ​​of production processes, process standards and quality control, etc.

[0029] In this embodiment, fraud-related refers to various factors and indicators that may indicate improper behavior or illegal operations, including: abnormal production data, inconsistency, violation of production process, non-conformity report, etc.

[0030] In this embodiment, the key indicators are specific data types used to measure the performance and efficiency of the production process, such as production efficiency, scrap rate, downtime, compliance, etc.

[0031] In this embodiment, the key feature refers to an important variable that can effectively indicate or affect the production process when analyzing production data, such as production rate, operation abnormality, environmental factors, etc.

[0032] In this embodiment, the basic model refers to a model used to preliminarily evaluate and understand data characteristics during data analysis and machine learning.

[0033] In this embodiment, the training set is a data set used to train a machine learning model.

[0034] In this embodiment, the test set is a data set used to evaluate the performance of the machine learning model.

[0035] In this embodiment, the early warning model is generated by training the basic model, and is a machine model designed to identify abnormal behavior and issue an alarm.

[0036] In this embodiment, the monitoring mechanism is a system established based on the early warning model, which is used to track and analyze the production behavior of employees in real time.

[0037] In this embodiment, real-time abnormal behavior refers to situations in which employee behavior or production data deviates from normal patterns during the production process, including: abnormal production data, inconsistency, violation of production process, non-compliance reporting, etc.

[0038] The working principle and beneficial effects of the above technical solution are: through real-time data collection and processing, key fraud indicators are identified, early warning models are built, employee behavior monitoring is achieved, identification efficiency and accuracy are improved, the risk of corporate losses is reduced, and scientific and transparent production management is ensured.

[0039] Embodiment 2: Based on the above embodiment 1, the real-time production data of any employee on the production line is obtained, including: Configure a preset type of sensor on the target production equipment of the production line, and obtain real-time raw data of the production equipment in real time according to the sensor; The sensor is connected to an API connection device to obtain real-time raw data, and the real-time raw data is filtered according to any employee ID to obtain the real-time production data of the employee.

[0040] In this embodiment, the sensor is a device used to detect and measure specific physical quantities and convert these quantities into signals that can be read or analyzed.

[0041] In this embodiment, the target production equipment refers to a machine specifically used to perform production tasks on a production line.

[0042] In this embodiment, the preset type refers to a specific sensor type configured in the production line, including: temperature, humidity, pressure, flow, etc.

[0043] In this embodiment, the real-time raw data refers to the raw information instantly acquired from the sensor and not processed or analyzed.

[0044] In this embodiment, the API connected device refers to a hardware component that can communicate with other systems or devices through an application program interface (API).

[0045] In this embodiment, the employee ID is a unique identifier used to distinguish and identify each employee on the production line.

[0046] The working principle and beneficial effects of the above technical solution are: by configuring sensors to monitor equipment in real time, screening employee data, realizing production process monitoring, and improving work efficiency and the accuracy of management decisions.

[0047] Embodiment 3: On the basis of the above-mentioned embodiment 2, the historical production data of the employee is obtained according to the ERP system, including: Connect to the ERP system query interface according to the preset connection information; The employee ID is used as a query condition to generate an interface query statement, and the ERP system query interface is called to obtain the historical production data of the employee.

[0048] In this embodiment, the preset connection information refers to basic configuration parameters required when connecting to the ERP system.

[0049] In this embodiment, the interface query statement is a SQL statement or an API request format for extracting data from the ERP system.

[0050] The working principle and beneficial effects of the above technical solution are: querying employees' historical production data through the ERP interface, realizing data integration and efficient analysis, and supporting management decisions and performance evaluation.

[0051] Embodiment 4: On the basis of the above-mentioned embodiment 1, the real-time production data and historical production data of all employees are pre-processed, and standard real-time data and standard historical data are obtained and stored in the production database, including: Obtaining the real-time production data and historical production data of any production parameter of any employee, and processing the production data and historical production data in standardized units according to preset business rules matching the production parameters to obtain first real-time production data and first historical production data; Processing the first real-time production data and the first historical production data according to the normal distribution model, identifying abnormal values ​​greater than a normal threshold and replacing them with standard normal distribution values, and finally obtaining second real-time production data and second historical production data; Obtain corresponding time periods for the second real-time production data and the second historical production data, and perform standardized time processing on each of them according to preset standard time intervals to obtain standard real-time data and standard historical data with the same time interval; According to the preset database table structure, a production database is constructed, and standard real-time data and standard historical data are stored in the production database using employee ID as the identifier.

[0052] In this embodiment, the preset business rules refer to specific standards and conditions based on which production data is processed.

[0053] In this embodiment, the standardized unit processing is to convert the data of different production parameters into a unified unit.

[0054] In this embodiment, the first real-time production data is a value extracted from the real-time production data of a specific employee after being processed by standardized units.

[0055] In this embodiment, the first historical production data is a value extracted from the historical production data of a specific employee after being processed into standardized units.

[0056] In this embodiment, the normal distribution model is a statistical distribution used to describe the central tendency and dispersion of data around a mean value.

[0057] In this embodiment, the normal threshold refers to a critical value in a normal distribution, which is used to determine whether a data point is an outlier.

[0058] In this embodiment, an outlier refers to a data value in a data set that deviates from the overall pattern of the data compared to other data points.

[0059] In this embodiment, the standard normal distribution value refers to a value with a mean of 0 and a standard deviation of 1 in a standard normal distribution.

[0060] In this embodiment, the second real-time production data refers to a data set obtained after the first real-time production data is processed and the abnormal values ​​are identified and replaced by a normal distribution model.

[0061] In this embodiment, the second historical production data refers to a data set obtained after the first historical production data is processed and outliers are identified and replaced through a normal distribution model.

[0062] In this embodiment, the preset database table structure includes data fields such as employee ID, production parameters, standard data, and timestamp.

[0063] The working principle and beneficial effects of the above technical solution are: through standardized processing and outlier identification, the management of employee production data is optimized, data consistency and reliability are ensured, efficient analysis and decision-making are supported, and production quality is improved.

[0064] Embodiment 5: On the basis of the above-mentioned embodiment 1, the production type data of the production database is obtained by screening according to any production parameter type, and the key indicators and key features related to fraud are identified and extracted according to the expert database, including: Using any production parameter type as a screening condition, obtaining production type data of the production parameter type in the production database, the production type data including standard historical data and standard real-time data; Based on the production parameter type name, search the expert database for the names of key fraud-related indicators with a similarity greater than a preset threshold; ;in, Indicates the degree of correlation between the production parameter type name and any key indicator name in the expert database, Indicates the ascii value of the i-th character in the production parameter name. The standard ASCII value of the key indicator name character. Indicates the preset weight of the ascii value of the i-th character in the production parameter name, Indicates the standard similarity between the key indicator name and the character origin ascii, The maximum character distance of the key indicator name; If the correlation between the production parameter type name and the key indicator name is greater than a preset threshold, the key indicator corresponding to the key indicator name is the key indicator of the production parameter type; The standard historical data of production parameter types are extracted according to the preset formula of key indicators to obtain the preset indicator data, and the key features of the preset indicator data are extracted using the feature matrix extraction method.

[0065] In this embodiment, the degree of correlation indicates the degree of matching between the production parameter type name and the key indicator name in the expert database.

[0066] In this embodiment, the ascii value refers to the character code in the American Standard Code for Information Interchange, and each character (including letters, numbers, symbols, etc.) has a corresponding integer value.

[0067] In this embodiment, the maximum character distance refers to the maximum difference between two character strings when comparing them.

[0068] In this embodiment, the preset indicator data refers to a specific data set obtained by calculating and extracting the preset formula based on the standard historical data of the production parameter type and the relevant key indicators.

[0069] In this embodiment, the character origin refers to a preset reference point used to calculate the similarity between characters.

[0070] In this embodiment, the feature matrix extraction method refers to a specific data set obtained after calculation and extraction based on the preset formula of the standard historical data of the production parameter type and the relevant key indicators.

[0071] The working principle and beneficial effects of the above technical solution are: through data screening and similarity analysis, key fraud indicators are identified and features are extracted to improve production data monitoring and decision-making efficiency.

[0072] Embodiment 6: On the basis of the above-mentioned embodiment 1, a basic model is constructed with key indicators and their key features, and the standard historical data of any employee is used as a training set and the standard real-time data is used as a test set to train the basic model to generate an early warning model, including: The key indicator of any production parameter type is used as the basic model framework node, and the coefficient of the framework node is constructed based on the key characteristics corresponding to the key indicator. A complete basic model is constructed for all production parameter types, key indicators and their key characteristics; Obtain all data in the production database based on the employee ID of any employee, use the standard historical data as a training set, train the basic model, and obtain the first model indicator; Using the standard real-time data as a test set, retraining the first model to obtain the second model index, and adjusting the first model index parameter based on the second model index to obtain the final index; The final indicator is used to configure the first model to obtain the early warning model.

[0073] In this embodiment, the basic model framework node refers to a key indicator that serves as a core component in the production parameter analysis.

[0074] In this embodiment, the coefficient refers to a numerical value used to quantify the relationship between the key indicator and its key feature.

[0075] In this embodiment, the first model indicator refers to the preliminary result obtained based on standard historical data after training the basic model, including: accuracy, precision, recall rate, etc.

[0076] In this embodiment, the second model indicator refers to an evaluation indicator obtained based on a standard real-time data test set after retraining the basic model.

[0077] In this embodiment, the final indicator is the adjusted model evaluation result, including: comprehensive accuracy, mean square error, model stability, etc.

[0078] The working principle and beneficial effects of the above technical solution are: through dynamic model training and adjustment, an early warning model is constructed to achieve real-time monitoring and accurate risk prediction, thereby improving production management efficiency.

[0079] Embodiment 7: Based on the above-mentioned Example 6, a monitoring mechanism is established based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings, including: Setting warning thresholds for different types of production parameters based on the pre-output of the warning model, and configuring the warning model into the production line system; If the absolute value deviation between the production parameter type and the standard production parameter type of any employee in the real-time production process of the production line is greater than or equal to the warning threshold, the abnormal behavior of the employee's key indicator matching is identified and fed back to the system; if the absolute value deviation between the production parameter type and the standard production parameter type of the employee in the real-time production process of the production line is less than the warning threshold, it is determined that there is no abnormal behavior.

[0080] In this embodiment, the production line system refers to a comprehensive information system for monitoring, managing and optimizing the production process.

[0081] The working principle and beneficial effects of the above technical solution are: by setting early warning thresholds, real-time monitoring of production parameters, timely identification of abnormal behaviors, improving production quality and safety, and optimizing management decisions.

[0082] Embodiment 8: like Figure 2 As shown, an intelligent factory-side fraud behavior early warning system provided by an embodiment of the present invention includes: Data acquisition module: obtain the real-time production data of any employee's production line, and obtain the employee's historical production data based on the ERP system; Preprocessing module: preprocesses the real-time production data and historical production data of all employees, obtains standard real-time data and standard historical data and stores them in the production database; Feature module: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database; Model module: construct a basic model based on key indicators and their key features, and use the standard historical data of any employee as a training set and the standard real-time data as a test set to train the basic model and generate an early warning model; Monitoring module: Establish a monitoring mechanism based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings.

[0083] The working principle and beneficial effects of the above technical solution are: through real-time data collection and processing, key fraud indicators are identified, early warning models are built, employee behavior monitoring is achieved, identification efficiency and accuracy are improved, the risk of corporate losses is reduced, and scientific and transparent production management is ensured.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent factory-side fraud warning method, characterized in that: include: Step 1: Obtain the real-time production data of any employee on the production line, and obtain the historical production data of the employee according to the ERP system; Step 2: Preprocess the real-time production data and historical production data of all employees, obtain standard real-time data and standard historical data, and store them in the production database; Step 3: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database; Step 4: Build a basic model with key indicators and their key features, and use the standard historical data of any employee as a training set and the standard real-time data as a test set to train the basic model and generate an early warning model; Step 5: Establish a monitoring mechanism based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings.

2. According to claim 1, an intelligent factory-side fraud warning method is characterized in that: Get real-time production data of any employee on the production line, including: Configure a preset type of sensor on the target production equipment of the production line, and obtain real-time raw data of the production equipment in real time according to the sensor; The sensor is connected to an API connection device to obtain real-time raw data, and the real-time raw data is filtered according to any employee ID to obtain the real-time production data of the employee.

3. The intelligent factory-side fraud warning method according to claim 2 is characterized in that: And obtain the employee's historical production data based on the ERP system, including: Connect to the ERP system query interface according to the preset connection information; The employee ID is used as a query condition to generate an interface query statement, and the ERP system query interface is called to obtain the historical production data of the employee.

4. The intelligent factory-side fraud warning method according to claim 1 is characterized in that: Pre-process the real-time production data and historical production data of all employees, obtain standard real-time data and standard historical data and store them in the production database, including: Obtaining the real-time production data and historical production data of any production parameter of any employee, and processing the production data and historical production data in standardized units according to preset business rules matching the production parameters to obtain first real-time production data and first historical production data; Processing the first real-time production data and the first historical production data according to a normal distribution model, identifying abnormal values ​​greater than a normal threshold and replacing them with standard normal distribution values, and finally obtaining second real-time production data and second historical production data; Obtain corresponding time periods for the second real-time production data and the second historical production data, and perform standardized time processing on each of them according to preset standard time intervals to obtain standard real-time data and standard historical data with the same time interval; According to the preset database table structure, a production database is constructed, and standard real-time data and standard historical data are stored in the production database using employee ID as the identifier.

5. The intelligent factory-side fraud warning method according to claim 1 is characterized in that: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database, including: Using any production parameter type as a screening condition, obtaining production type data of the production parameter type in the production database, the production type data including standard historical data and standard real-time data; Based on the production parameter type name, search the expert database for the names of key fraud-related indicators with a similarity greater than a preset threshold; ;in, Indicates the degree of correlation between the production parameter type name and any key indicator name in the expert database, Indicates the ascii value of the i-th character in the production parameter name. The standard ASCII value of the key indicator name character. Indicates the preset weight of the ascii value of the i-th character in the production parameter name, Indicates the standard similarity between the key indicator name and the character origin ascii, The maximum character distance of the key indicator name; If the correlation between the production parameter type name and the key indicator name is greater than a preset threshold, the key indicator corresponding to the key indicator name is the key indicator of the production parameter type; The standard historical data of production parameter types are extracted according to the preset formula of key indicators to obtain the preset indicator data, and the key features of the preset indicator data are extracted using the feature matrix extraction method.

6. The intelligent factory-side fraud warning method according to claim 1 is characterized in that: The key indicators and their key features are used to build a basic model, and the standard historical data of any employee is used as a training set, and the standard real-time data is used as a test set to train the basic model to generate an early warning model, including: The key indicator of any production parameter type is used as the basic model framework node, and the coefficient of the framework node is constructed based on the key characteristics corresponding to the key indicator. A complete basic model is constructed for all production parameter types, key indicators and their key characteristics; Obtain all data in the production database based on the employee ID of any employee, use the standard historical data as a training set, train the basic model, and obtain the first model indicator; Using the standard real-time data as a test set, retraining the first model to obtain the second model index, and adjusting the first model index parameter based on the second model index to obtain the final index; The final indicator is used to configure the first model to obtain the early warning model.

7. The intelligent factory-side fraud warning method according to claim 6 is characterized in that: Establish a monitoring mechanism based on the early warning model to identify abnormal employee behavior in real time and issue early warnings, including: Setting warning thresholds for different types of production parameters based on the pre-output of the warning model, and configuring the warning model into the production line system; If the absolute value deviation between the production parameter type and the standard production parameter type of any employee in the real-time production process of the production line is greater than or equal to the warning threshold, the abnormal behavior of the employee's key indicator matching is identified and fed back to the system; if the absolute value deviation between the production parameter type and the standard production parameter type of the employee in the real-time production process of the production line is less than the warning threshold, it is determined that there is no abnormal behavior.

8. An intelligent factory-side fraud warning system, characterized in that: include: Data acquisition module: obtain the real-time production data of any employee's production line, and obtain the employee's historical production data based on the ERP system; Preprocessing module: preprocesses the real-time production data and historical production data of all employees, obtains standard real-time data and standard historical data and stores them in the production database; Feature module: Filter and obtain production type data from the production database according to any production parameter type, and identify and extract fraud-related key indicators and key features based on the expert database; Model module: construct a basic model based on key indicators and their key features, and use the standard historical data of any employee as a training set and the standard real-time data as a test set to train the basic model and generate an early warning model; Monitoring module: Establish a monitoring mechanism based on the early warning model to identify abnormal behaviors of employees in real time and issue early warnings.

Images