Product production process control method based on Internet of Things

Abstract

The invention discloses a product production process control method based on the Internet of Things, and the method comprises the following steps: 1, configuration setting: setting corresponding rules in a production equipment control system, including a data collection rule, a data cleaning rule, and an alarm rule and an abnormality judgment rule of different severity levels set by a user according to different product types; step 2, data acquisition and cleaning: acquiring data such as appearance, station temperature and humidity, working current, frequency, temperature and power of products and working equipment by using corresponding sensors and monitoring equipment, so as to perform multi-dimensional data acquisition and form different types of data reports; and the data is transmitted to a data warehouse through the data transmission module. The factory equipment interface uniformity, the data safety, the system stability and the function diversity are comprehensively improved, a large number of human resources are saved, and the quality accidents caused by the fact that process control is carried out on a large number of products without real-time production statistical data are avoided.

Description

Technical Field

[0001] The present invention relates to the field of computer software technology, and in particular to a product production process control method based on the Internet of Things. Background Art

[0002] With the rapid advancement of intelligent manufacturing in recent years, tracking product information, process, and quality data during mass production has become extremely important. This has necessitated the emergence of production process control software. This software processes and performs logical analysis on complex process and environmental data generated by machines and instruments during the manufacturing process to ensure compliance with production standards. The process control process utilizes mathematical and statistical methods to analyze and evaluate the production process. Based on this feedback, it promptly identifies factors that may cause production anomalies and implements measures to mitigate these factors, maintaining the process within a predetermined set of factors to achieve quality control. Process control leverages the statistical regularity of process fluctuations to analyze and control the process. It emphasizes maintaining a controlled and capable process, ensuring that products and services consistently meet customer requirements. Process control software first acquires a set of various parameters output by the production process and, incorporating process anomaly detection rules, generates control charts and abnormal process alarms.

[0003] However, the current process control still has the following deficiencies:

[0004] 1. There is no automatic collection, cleaning, or analysis process for test data samples. Production process analysis relies on grabbing original machine operation logs from production equipment, manually screening valid data, and entering them into control chart generation tools;

[0005] 2. The control chart alarm rules are simple and non-configurable, and obtaining valid source data that matches the alarm rules requires a huge amount of manual work;

[0006] 3. There is no complete tracking and resolution process for product quality anomalies reflected by control chart alarms;

[0007] 4. Data analysis can only be generated based on a single condition and cannot control the output of results from a combination of multiple conditions. Summary of the Invention

[0008] The purpose of the present invention is to provide a product production process control method based on the Internet of Things to solve the problems raised in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solution: comprising the following steps:

[0010] Step 1: Configuration settings: Set corresponding rules in the production equipment control system, including data collection rules and data cleaning rules. Users can set alarm rules and abnormality judgment rules with different severity levels according to different product types;

[0011] Step 2: Data collection and cleaning: Use appropriate sensors and monitoring equipment to collect data on the appearance of products and working equipment, workstation temperature and humidity, working current, frequency, temperature, and power, thereby performing multi-dimensional data collection and generating different types of data reports. The data is then transmitted to the data warehouse through the data transmission module, and the collected data is cleaned and a log is generated.

[0012] Step 3: Data Generation and Analysis: Generate control charts from the cleaned data and transmit them to the quality control center in real time. The center then uses SPC to analyze and evaluate the production data. The abnormality rules set in Step 1 are used to identify abnormal data and provide timely feedback.

[0013] Step 4. Abnormal data tracking and processing: Create a task process for abnormal data results, form a problem group, create a problem tracking number, track the problems hierarchically, and designate relevant personnel at each level to track, resolve and review them. At the same time, the alarm configuration enables the production person in charge to promptly investigate and handle abnormal conditions in the production process, repair and close the problems.

[0014] Preferably, the data acquisition in step one is a data acquisition system developed based on the OPC protocol, and the data cleaning module is developed and written in the Python programming language.

[0015] Preferably, the abnormality judgment rules in step 1 can be set from multiple dimensions, such as arbitrarily combining factors such as time range, test point, part number, product group, process, production line and work station to output a control graph.

[0016] Preferably, according to step one, the user can flexibly set process control data extraction and data grouping size based on factors such as product family, product type, product model, test point description, production work order number, etc. to achieve multi-dimensional generation and analysis of process control data.

[0017] Preferably, the specific operation process in step 4 is:

[0018] (1) Identify the problem: Create a problem list based on the abnormal data details and determine the problems that need to be solved;

[0019] (2) Create an issue tracking number: Use an issue management tool or form to create an issue tracking number; include the issue description, scope of impact, priority, status, and person in charge in the number;

[0020] (3) Assigning responsible persons: Based on the urgency and complexity of the problem, assign responsible persons to solve the problem. At the same time, assign a unique identifier to the problem tracking number to facilitate tracking and management of the problem.

[0021] (4) Problem investigation: Investigate the problem, understand the root cause and solution, and record the investigation results and solution in the problem tracking number;

[0022] (5) Problem Solving: Based on the investigation results and solutions, implement appropriate measures to solve the problem and record the solution process and results in the problem tracking number.

[0023] Preferably, the multi-dimensional generation of process control data is achieved by processing a set of related variables, extracting the eigenvalues ​​and eigenvectors of the sample covariance matrix, and re-extracting the sample values ​​under this dimension based on the cumulative contribution rate principle. This variable contains most of the information in the original data analysis dimension, constructs new statistics, and generates data analysis results under a controlled state. Data cleaning is triggered according to the definition of the data analysis dimension, thereby entering and completing step two.

[0024] Preferably, the control chart parameter settings in step three include the number of sampling groups, sample size within the group, standard deviation, range mean, median, upper and lower control limits, and test points to generate statistical results.

[0025] Preferably, the data cleaning in step 3 adopts the optimal intra-group variance method, and different filtering thresholds S are set for different collected samples, and the collected samples within a specific time range are selected;

[0026] The core idea of ​​the optimal within-group variance method is to set the initial variance threshold S and then find the corresponding variance mutation point. Data with a variance lower than S is considered normal data, and vice versa. The variance mutation point of the power data is λ:

[0027]

[0028] In the formula: is the average power from the 1st to the λth point

[0029] The algorithm cleaning process is as follows:

[0030] (1) Divide the interval into intervals based on a specific time period or sample group size and extract data;

[0031] (2) Rearrange the power data of each interval in ascending order of power values;

[0032] (3) In each interval, calculate the power variance values ​​of the first i points in sequence (the variance value of the first point is always 0);

[0033] (4) A threshold S is set in each interval. Power data with a power variance less than S is considered normal power data, and power data with a power variance greater than or equal to S is considered abnormal power data.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. This invention completely replaces data collection work with programs, saving a lot of human resources. The collection service is developed based on the OPC UA architecture, comprehensively improving the uniformity of factory equipment interfaces, data security, system stability and functional diversity. The collected data cleaning function is written in Python, ensuring the timeliness of the data ultimately presented to the process control system.

[0036] 2. The present invention also generates statistical process parameters through visual configuration, so that equipment operation changes and production process anomalies presented by the system can be discovered in time, avoiding quality accidents of large-scale products due to the lack of real-time production statistical data for process control. For production process anomalies that have already occurred, it supports the generation of task management processes for solving anomaly points and assigns relevant technical engineering personnel to track and solve the anomalies; from data collection and cleaning, multi-dimensional production process statistical analysis data generation, to production anomaly point tracking and resolution, a closed-loop management of quality control is formed. DETAILED DESCRIPTION

[0037] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described 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 making creative efforts are within the scope of protection of the present invention.

[0038] The present invention provides a technical solution, a product production process control method based on the Internet of Things:

[0039] Set corresponding rules in the production equipment control system, including data collection rules. When setting data collection, determine the time interval for data collection to meet production needs while avoiding excessive or insufficient data generation. If the PLC data node fails to read, it will retry reading according to the set time length. At the same time, set the maximum number of attempts to re-read the PLC node data. According to different process control basic parameter settings, different process control data generation dimensions are used to configure the frequency of initial data generation and historical refresh. According to the needs of the production process, select the parameters to be collected, such as temperature, pressure, flow, etc., set the conditions for triggering data collection, such as reaching a certain threshold, specific event occurrence or timed collection, etc., data cleaning rules, and verify the legitimacy of the collected data, including range check, format verification and data type verification, etc. , repair or replace abnormal data, such as using interpolation to fill missing values, smooth data or delete obviously erroneous data points, and filter out data that meets the requirements according to specific conditions and rules, such as excluding abnormal data, invalid data or unreliable data, and perform cleaning operations on historical data to ensure the quality and consistency of the entire data set. Users can set alarm rules and judgment rules of different severity levels according to different product types. Judgment rules can be set from multiple dimensions, such as arbitrarily combining factors such as time range, test point, part number, product group, process, production line and work station to output control graphics. Users have flexible control rights. The system enables users to change, operate and configure the control chart alarm rules of the business layer, and can support users to set alarm rules of different severity levels according to different product types;

[0040] in;

[0041] Data cleaning uses the optimal intra-group variance method, sets different filtering thresholds S for different collected samples, and cleans the collected samples within a specific time range;

[0042] The core idea of ​​the optimal within-group variance method is to set the initial variance threshold S and then find the corresponding variance mutation point. Data with a variance lower than S is considered normal data, and vice versa. The variance mutation point of the power data is λ:

[0043]

[0044] In the formula: is the average power from the 1st to the λth point.

[0045] Take the power data sample of the cleaning equipment as an example:

[0046] Extract the power data for each interval and save it in an array or list. This process is implemented in Python.

[0047] Sort the power data for each interval using Python's built-in sort() function, where the key parameter specifies the sorting method (in ascending order of power values).

[0048] The power data of each interval has been sorted in ascending order of power values;

[0049] In each interval, calculate the power variance values ​​of the first i points in turn (the variance value of the first point is always 0);

[0050] A threshold S is set in each interval. Power variance less than S is considered normal power data, and power variance greater than or equal to S is considered abnormal power data.

[0051] Use appropriate sensors and monitoring equipment to collect data on the appearance of products and working equipment, workstation temperature and humidity, working current, frequency, temperature, and power. Real-time data on the physical environment and equipment operation can be collected through sensor equipment. For example, various physical quantities in the production process, such as temperature, humidity, pressure, and vibration, can be collected. Sensors can directly measure physical quantities, or convert other signals (such as current and voltage) into the required signal type through converters. Open API interfaces can also be used to obtain data.

[0052] Duplicate values ​​may affect the accuracy and credibility of data analysis results, so they need to be removed. Use the drop_duplicates() method in software packages such as pandas, or manually write programs to identify and delete duplicates, so as to conduct multi-dimensional data collection and form different types of data reports. The data is then transferred to the data warehouse through the data transmission module, and the collected data is cleaned and logs are generated. At the same time, the data processing process and the operating status of related services can be controlled through the log traceability query process.

[0053] Data generation and analysis: Generate control charts from the cleaned data. When generating control charts, it is first necessary to conduct a systematic analysis of the cleaned data, including calculating indicators such as mean, standard deviation, variance, and range. In Python, use function libraries such as NumPy and Pandas for data statistics. In Python, use function libraries such as Matplotlib and Seaborn to generate control charts. In Python, use methods such as Six Sigma and Lean for quality management. These methods are used to implement quality management, including establishing quality standards, monitoring production processes, evaluating quality levels, and improving quality systems, which can improve production efficiency and customer satisfaction. The data is then transmitted to the quality control center in real time to analyze and evaluate the production data using SPC. The set abnormality judgment rules are used to determine abnormal data and provide timely feedback. The control chart parameter settings include the number of sampling groups, sample size within the group, standard deviation, range mean, median, upper and lower control limits, or test points to generate statistical results.

[0054] Create a task flow for abnormal data results, form a problem list, and create a problem tracking number. Use the task management system to create the problem tracking number. Form a problem group based on the abnormal data details and identify the problems that need to be solved. Include information such as problem description, impact scope, priority, status, and handling personnel in the task. Track the problems in a hierarchical manner, and assign relevant personnel to track, resolve, and review them. At the same time, alarm configuration enables production responsible persons to promptly investigate and handle abnormal production process conditions, repair, and close the problems. Link the task management module and assign responsible personnel based on the urgency and complexity of the problem. At the same time, assign a unique identifier to the problem tracking number to facilitate problem tracking and management. Abnormal data samples are upgraded, tracked, and resolved to achieve closed-loop problem resolution. Abnormal indicators output by process control automatic analysis are categorized, and different email reminders are set for different categories. Handlers and reviewers are set at different levels based on the severity of the abnormality. Investigate the problem to understand the root cause and solution. Record the investigation results and solution in the problem tracking number. Based on the investigation results and solution, implement appropriate measures to resolve the problem. Record the resolution process and results in the problem tracking number.

[0055] Data acquisition is a data acquisition system developed based on the OPC protocol. The data acquisition system developed based on the OPC protocol can communicate with equipment and systems of various suppliers to achieve interoperability between equipment and systems. At the same time, the data cleaning module is developed and written in Python programming language, which has good interoperability with other computer languages. It can be easily connected to other language programs such as Java. In software development, Python is used to build a framework, and modules with high computing performance requirements are written in C or C++, and finally embedded in the Python framework, which can greatly improve development efficiency and the operating performance of software products. At the same time, the data cleaning module is linked to clean the data and extract effective data according to the data sample grouping rules set by the user, avoiding manual screening and entry.

[0056] Users can flexibly set process control data extraction and data grouping size based on factors such as product family, product type, product model, test point description and production work order number to achieve multi-dimensional generation and analysis of process control data. Customers can extract sample data and generate control charts from multiple dimensions, allowing customers to perform data analysis from multiple dimensions and achieve complex production process statistics and control.

[0057] The multi-dimensional generation of process control data is achieved by processing a set of related variables, extracting the eigenvalues ​​and eigenvectors of the sample covariance matrix, and re-extracting the sample values ​​under this dimension based on the cumulative contribution rate principle. This variable contains most of the information in the original data analysis dimension, constructing new statistics and generating data analysis results under a controlled state. Data cleaning is triggered according to the definition of the data analysis dimension, thereby entering and completing data collection and cleaning.

[0058] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0059] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A product production process control method based on the Internet of Things, characterized in that: The following steps are involved: Step 1: Configuration settings: Set corresponding rules in the production equipment control system, including data collection rules and data cleaning rules. Users can set alarm rules and abnormality judgment rules of different severity levels according to different product types; Step 2: Data collection and cleaning: Use corresponding sensors and monitoring equipment to collect data such as the appearance of products and working equipment, workstation temperature and humidity, working current, frequency, temperature and power, so as to collect data in multiple dimensions and form different types of data reports. The data is then transmitted to the data warehouse through the data transmission module, and the collected data is cleaned and a log is generated. Step 3: Data generation and analysis: Generate control charts with the cleaned data and transmit them to the quality control center in real time to analyze and evaluate the production data using SPC. Use the abnormality judgment rules set in step 1 to judge abnormal data and provide timely feedback; Step 4. Abnormal data tracking and processing: Create a task flow for abnormal data results, form a problem list, create a problem tracking number, track the problems in a hierarchical manner, and designate relevant personnel to track, resolve and review them. At the same time, the alarm configuration enables the person in charge of production to promptly investigate and handle abnormal conditions in the production process, and repair and close the problems.

2. The method for controlling a product production process based on the Internet of Things according to claim 1, characterized in that: The data acquisition in step one is a data acquisition system developed based on the OPC protocol, and the data cleaning module is developed and written in the Python programming language.

3. The method for controlling a product production process based on the Internet of Things according to claim 1, characterized in that: The abnormality judgment rules in step 1 can be set from multiple dimensions, such as arbitrarily combining factors such as time range, test point, part number, product group, process, production line and work station to output a control graph.

4. The method for controlling a product production process based on the Internet of Things according to claim 1, characterized in that: According to step 1, the user can flexibly set the process control data extraction and data grouping size according to factors such as product family, product type, product model, test point description, production work order number, etc. to realize multi-dimensional generation and analysis of process control data.

5. The method for controlling a product production process based on the Internet of Things according to claim 1, characterized in that: The specific operation process in step 4 is as follows: (1) Identify the problem: Create a problem list based on the abnormal data details and determine the problem that needs to be solved; (2) Create an issue tracking number: Use an issue management tool or form to create an issue tracking number. Include the issue description, scope of impact, priority, status, and person in charge in the number. (3) Assigning a responsible person: Assign a responsible person to solve the problem based on the urgency and complexity of the problem. At the same time, specify a unique identifier for the problem tracking number to facilitate tracking and management of the problem; (4) Problem investigation: Investigate the problem, understand the root cause and solution, and record the investigation results and solution in the problem tracking number; (5) Problem solving: Based on the investigation results and solutions, implement appropriate measures to solve the problem and record the solution process and results in the problem tracking number.

6. The method for controlling a product production process based on the Internet of Things according to claim 4, characterized in that: The multi-dimensional generation of process control data is achieved by processing a set of related variables, extracting the eigenvalues ​​and eigenvectors of the sample covariance matrix, and re-extracting the sample values ​​under this dimension based on the cumulative contribution rate principle. This variable contains most of the information in the original data analysis dimension, constructs new statistics, and generates data analysis results under a controlled state. Data cleaning is triggered according to the definition of the data analysis dimension, thereby entering and completing step two.

7. The method for controlling a product production process based on the Internet of Things according to claim 1, characterized in that: The control chart parameter settings in step three include the number of sampling groups, sample size within the group, standard deviation, range mean, median, upper and lower control limits, and test points to generate statistical results.

8. The method for controlling a product production process based on the Internet of Things according to claim 1, characterized in that: The data cleaning in step 3 adopts the optimal intra-group variance method, and different collection samples are set with different filtering thresholds S, and the collection samples within a specific time range are selected; The core idea of ​​the optimal within-group variance method is to find the corresponding variance mutation point after setting the initial variance threshold S. Data with variance lower than S is normal data, and vice versa is abnormal. The variance mutation position of power data is: In the formula: is the average power from the 1st to the λth point The algorithm cleaning process is as follows: (1) Divide the interval into intervals based on a specific time period or sample group size and extract data; (2) rearrange the power data of each interval in ascending order of power values; (3) In each interval, calculate the power variance values ​​of the first i points in turn (the variance value of the first point is always 0); (4) A threshold S is set in each interval. Power variance less than S is considered normal power data, and power variance greater than or equal to S is considered abnormal power data.

Images