Smart factory management method and system based on AI big data
By identifying and preprocessing data sources in real time, distributed storage is carried out according to importance and frequency of use, and using intelligent analysis models to generate early warning information, the problem of mass production data processing is solved, production efficiency and product quality are improved, and operational costs are reduced.
Patent Information
- Application Number
- CN202411209498.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology is difficult to efficiently identify and process massive production data, resulting in waste of resources and performance bottlenecks. The traditional factory management methods are inefficient and difficult to cope with complex and changeable production data.
Using a smart factory management method based on AI big data, we use real-time identification and preprocess data sources, and distributed storage according to the importance and frequency of use, and use intelligent analysis models to generate early warning information to achieve efficient storage and real-time analysis of data.
Improves production efficiency and product quality, reduces downtime and operating costs, and provides rich data visualization tools and reporting functions to help managers fully understand the operating status of the factory and promptly detect and respond to potential safety risks and equipment failures.
Smart Images

Figure CN119087937A_ABST
Abstract
Description
Technical Field
[0001] The present invention proposes a smart factory management method and system based on AI big data, belonging to the technical field of smart factories. Background Art
[0002] In the field of modern industrial manufacturing, with the continuous development of information and intelligent technologies, smart factories have become an important direction for the transformation and upgrading of the industry. By integrating advanced information technology, Internet of Things technology, artificial intelligence technology and big data analysis technology, smart factories have realized the automation, digitization and intelligence of the production process, significantly improved production efficiency and product quality, and reduced operating costs. However, in the actual operation of smart factories, there are still some technical challenges and management difficulties. Among them, how to efficiently identify and process the massive data sources in the factory, realize real-time preprocessing and distributed storage of data, and how to perform intelligent analysis based on these data to generate valuable early warning information are issues that need to be solved urgently.
[0003] Traditional factory management methods often rely on manual data collection and analysis, which is not only inefficient but also difficult to deal with large-scale, complex and changeable production data. At the same time, due to the differences in the importance and frequency of use of data sources, if a unified data storage and processing method is adopted, it will often lead to resource waste and performance bottlenecks. Summary of the invention
[0004] The present invention provides a smart factory management method and system based on AI big data to solve the above problems of traditional factory management and smart factory management in the prior art. The technical solutions adopted are as follows:
[0005] A smart factory management method based on AI big data, the smart factory management method based on AI big data comprising:
[0006] Identify the data source in the factory in real time, and perform data preprocessing on the data information obtained from the data source to obtain the preprocessed data information;
[0007] Distributed storage of the processed data information is performed according to the importance and usage frequency of the data source;
[0008] Perform real-time intelligent analysis on the stored data information, obtain intelligent analysis results, and generate early warning data information for factory production based on the intelligent analysis results.
[0009] Furthermore, the data source in the factory is identified in real time, and data information obtained from the data source is preprocessed, and the preprocessed data information obtained includes:
[0010] Identify data sources within the factory in real time, including equipment sensors and production line status monitoring data;
[0011] Collect data information for each data source in real time to obtain the data information of each data source corresponding to a unit of time;
[0012] Data preprocessing is performed on the data information obtained from the data source to obtain the processed data information; wherein the data preprocessing includes timestamp alignment processing, data format unification processing and noise removal processing.
[0013] Furthermore, the processed data information is stored in a distributed manner according to the importance and usage frequency of the data source, including:
[0014] Extract the first-level weight coefficient corresponding to each data source;
[0015] Extract the secondary weight coefficient and data usage frequency corresponding to the data information obtained by each data source;
[0016] Using the first-level weight coefficient, the second-level weight coefficient and the data usage frequency, obtain a comprehensive weight coefficient corresponding to each data source;
[0017] Compare the comprehensive weight coefficient with a pre-examination first weight coefficient threshold and a second weight coefficient threshold to obtain a comparison result; wherein the value corresponding to the first weight coefficient threshold is greater than the value corresponding to the second weight coefficient threshold;
[0018] According to the comparison results, the data sources are divided into importance categories to obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources;
[0019] Data is distributedly stored for primary, secondary and tertiary importance data sources.
[0020] Furthermore, the data sources are divided into importance categories according to the comparison results to obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources, including:
[0021] comparing the comprehensive weight coefficient with a pre-screened first weight coefficient threshold and a second weight coefficient threshold;
[0022] When the comprehensive weight coefficient exceeds the preset first weight coefficient threshold, the data source corresponding to the comprehensive weight coefficient exceeding the preset first weight coefficient threshold is determined to be a first-level importance data source;
[0023] When the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold is a secondary importance data source;
[0024] When the comprehensive weight coefficient does not exceed the preset second weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient not exceeding the preset second weight coefficient threshold is a third-level importance data source.
[0025] Furthermore, real-time intelligent analysis is performed on the stored data information to obtain intelligent analysis results, and early warning data information of factory production is generated according to the intelligent analysis results, including:
[0026] Retrieving a trained fault identification model from a database; wherein the fault identification model corresponds to a data source;
[0027] The stored data information corresponding to the data source is input into the fault identification model for fault identification, and when it is determined that the stored data information corresponding to the data source has abnormal operation data, an abnormal operation alarm of the data source is issued.
[0028] A smart factory management system based on AI big data, the smart factory management system based on AI big data comprising:
[0029] A real-time identification module is used to identify the data source in the factory in real time, and perform data preprocessing on the data information obtained from the data source to obtain the preprocessed data information;
[0030] A distributed storage module, used for performing distributed storage of the processed data information according to the importance and usage frequency of the data source;
[0031] The fault analysis module is used to perform real-time intelligent analysis on the stored data information, obtain the intelligent analysis results, and generate early warning data information of the factory production according to the intelligent analysis results.
[0032] Furthermore, the real-time recognition module includes:
[0033] A data source identification module, used to identify data sources within the factory in real time, wherein the data sources include equipment sensors and production line status monitoring data;
[0034] The data collection module is used to collect data information from each data source in real time and obtain the data information of each data source corresponding to a unit time;
[0035] The data preprocessing module is used to perform data preprocessing on the data information obtained by the data source to obtain the processed data information; wherein the data preprocessing includes timestamp alignment processing, data format unification processing and noise removal processing.
[0036] Furthermore, the distributed storage module includes:
[0037] A first-level weight coefficient extraction module is used to extract the first-level weight coefficient corresponding to each data source;
[0038] A secondary weight coefficient extraction module is used to extract the secondary weight coefficient and data usage frequency corresponding to the data information obtained by each data source;
[0039] A comprehensive weight coefficient acquisition module is used to obtain a comprehensive weight coefficient corresponding to each data source by using the first-level weight coefficient, the second-level weight coefficient and the data usage frequency;
[0040] A comparison module, configured to compare the comprehensive weight coefficient with a first weight coefficient threshold and a second weight coefficient threshold of a preliminary examination to obtain a comparison result; wherein a value corresponding to the first weight coefficient threshold is greater than a value corresponding to the second weight coefficient threshold;
[0041] An importance level classification module is used to classify the importance of data sources according to the comparison results, and obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources;
[0042] The hierarchical storage module is used for distributed storage of data for the first-level importance data source, the second-level importance data source and the third-level importance data source.
[0043] Furthermore, the importance level classification module includes:
[0044] A coefficient comparison module, used for comparing the comprehensive weight coefficient with a pre-examination first weight coefficient threshold and a second weight coefficient threshold;
[0045] A first-level importance data source determination module, used for determining that when the comprehensive weight coefficient exceeds a preset first weight coefficient threshold, the data source corresponding to the comprehensive weight coefficient exceeding the preset first weight coefficient threshold is a first-level importance data source;
[0046] A secondary importance data source determination module, for determining that the data source corresponding to the comprehensive weight coefficient not exceeding the preset first weight coefficient threshold but exceeding the preset second weight coefficient threshold is a secondary importance data source when the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold but exceeds the preset second weight coefficient threshold;
[0047] The third-level importance data source determination module is used to determine that the data source corresponding to the comprehensive weight coefficient not exceeding the preset second weight coefficient threshold is a third-level importance data source when the comprehensive weight coefficient does not exceed the preset second weight coefficient threshold.
[0048] Furthermore, the fault analysis module includes:
[0049] A fault identification model retrieval module is used to retrieve a trained fault identification model from a database; wherein the fault identification model corresponds to a data source;
[0050] The abnormality identification module is used to input the stored data information corresponding to the data source into the fault identification model for fault identification, and when it is determined that the stored data information corresponding to the data source has abnormal operation data, issue an abnormal data source operation alarm.
[0051] Beneficial effects of the present invention:
[0052] The present invention proposes a smart factory management method and system based on AI big data, which can timely discover and solve problems in the production process, reduce downtime and resource waste, and improve production efficiency and output by real-time monitoring and analysis of production data. Through intelligent analysis of production data, production parameters and quality standards can be controlled more accurately, unqualified products can be discovered and eliminated in time, and the overall quality of products can be improved. By optimizing data storage and processing strategies, the demand for hardware equipment and network bandwidth can be reduced; at the same time, predictive maintenance can be used to reduce equipment failures and downtime losses, further reducing operating costs. Provide rich data visualization tools and reporting functions to help managers fully understand the factory operation status and market trends, and provide strong support for decision-making. Through real-time monitoring and early warning mechanisms, potential safety risks and equipment failures can be discovered and responded to in a timely manner to ensure safe production and stable operation of the factory. BRIEF DESCRIPTION OF THE DRAWINGS
[0053] Figure 1 is a flow chart of the method of the present invention;
[0054] Figure 2 The system block diagram of the system of the present invention. DETAILED DESCRIPTION
[0055] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.
[0056] The embodiment of the present invention proposes a smart factory management method based on AI big data, such as Figure 1 As shown, the smart factory management method based on AI big data includes:
[0057] S1. Real-time identification of data sources within the factory, and data preprocessing of data information obtained from the data sources to obtain preprocessed data information;
[0058] S2. Distributed storage of the processed data information according to the importance and usage frequency of the data source;
[0059] S3. Perform real-time intelligent analysis on the stored data information, obtain intelligent analysis results, and generate early warning data information of factory production based on the intelligent analysis results.
[0060] The working principle of the above technical solution is: using the Internet of Things (IoT) technology, smart factories can identify and access various data sources in real time, including production equipment sensors, video surveillance, personnel positioning systems, etc. The data generated by these data sources is a direct reflection of the factory's operating status. The acquired raw data often contains noise, redundant information or incomplete data, and it is necessary to improve the quality and availability of the data through preprocessing steps such as data cleaning, deduplication, and normalization. The preprocessed data is more accurate and standardized, providing support for subsequent analysis and decision-making.
[0061] Evaluate the importance and frequency of use of data sources: Classify pre-processed data according to the importance of the data source to factory production and the frequency of data use. Highly important and frequently used data needs to be stored first, and its high availability and fast access capabilities must be ensured. Use big data storage technologies such as Hadoop and Spark to store processed data in a distributed manner on multiple nodes. This approach not only increases the storage capacity of data, but also increases the speed of data access and processing through parallel processing. At the same time, according to the importance and frequency of use of the data source, adopt appropriate storage strategies and backup mechanisms to ensure data security and reliability.
[0062] Based on artificial intelligence (AI) technologies, such as machine learning and deep learning, real-time intelligent analysis of stored data is performed. By building various analytical models, such as prediction models, classification models, clustering models, etc., hidden information and value in the data are mined. Based on the results of intelligent analysis, the system automatically identifies abnormal conditions or potential risks in the production process, such as equipment failure, product quality problems, and reduced production efficiency. Once an abnormal condition is detected, the system immediately generates an early warning message and notifies relevant personnel through a visual interface, SMS, email, etc.
[0063] The effects of the above technical solutions are: through real-time monitoring and analysis of production data, timely discover and solve problems in the production process, reduce downtime and resource waste, and improve production efficiency and output. Through intelligent analysis of production data, production parameters and quality standards can be controlled more accurately, unqualified products can be discovered and eliminated in time, and the overall quality of products can be improved. By optimizing data storage and processing strategies, the demand for hardware equipment and network bandwidth can be reduced; at the same time, predictive maintenance can be used to reduce equipment failures and downtime losses, further reducing operating costs. Provide a wealth of data visualization tools and reporting functions to help managers fully understand the factory operation status and market trends, and provide strong support for decision-making. Through real-time monitoring and early warning mechanisms, potential safety risks and equipment failures can be discovered and responded to in a timely manner to ensure safe production and stable operation of the factory.
[0064] In summary, the smart factory management method based on AI big data realizes real-time monitoring, intelligent analysis and early warning management of the production process by integrating the Internet of Things, big data analysis and artificial intelligence technologies, significantly improving the factory's production efficiency, product quality and operational benefits.
[0065] In one embodiment of the present invention, a data source in a factory is identified in real time, and data information acquired from the data source is preprocessed. The preprocessed data information is acquired including:
[0066] S101, identifying data sources in the factory in real time, wherein the data sources include equipment sensors and production line status monitoring data;
[0067] S102, collecting data information for each data source in real time, and obtaining data information of each data source corresponding to a unit time;
[0068] S103, performing data preprocessing on the data information acquired by the data source to obtain the processed data information; wherein the data preprocessing includes timestamp alignment processing, data format unification processing and noise removal processing.
[0069] The working principle of the above technical solution is: the system uses the Internet of Things (IoT) technology to identify various data sources in the factory in real time. These data sources mainly include equipment sensors and production line status monitoring data. Equipment sensors such as temperature sensors, pressure sensors, vibration sensors, etc. are used to monitor the operating status and performance parameters of the equipment; production line status monitoring data includes production speed, output, energy consumption, etc., which are used to reflect the overall operating status of the production line.
[0070] For each identified data source, the system collects data information in real time. By setting appropriate sampling frequency and collection rules, it ensures that the data information corresponding to each data source can be obtained within a unit of time. This data information reflects the real-time status of the factory production process and is the basis for subsequent data preprocessing and analysis.
[0071] Since the data collection time of different data sources may be different, the system first performs timestamp alignment on the data information. By unifying the time base, the consistency and comparability of all data in the time dimension are ensured. The data formats generated by different data sources may be different, such as numeric, text, image, etc. The system unifies the data format and converts data in different formats into a unified standard format to facilitate subsequent data analysis and processing. The original data may contain noise or outliers, which will interfere with subsequent analysis. The system removes noise from the data through algorithms such as smoothing, filtering, and denoising to improve the quality and accuracy of the data.
[0072] The effect of the above technical solution is: by real-time identification of data sources and collection of data information, the system can timely reflect the real-time status of the factory production process and provide support for rapid response to production changes. Through data preprocessing steps, including timestamp alignment, data format unification and noise removal, the system can significantly improve the accuracy and availability of data, and provide support for subsequent data analysis and decision-making. Unified data formats and high-quality data after noise removal can speed up data analysis and improve the efficiency of data analysis. At the same time, real-time also ensures that the data analysis results can be applied to production practice in a timely manner. This technical solution has strong scalability and can flexibly adapt to factory environments of different sizes and types. With the expansion of factory scale and the increase of production equipment, the system can easily expand the data source and data collection scope. It lays the foundation for subsequent intelligent analysis and early warning management. By real-time collection and preprocessing of high-quality data information, the system can more accurately identify abnormal conditions and potential risks in the production process, and provide strong support for the realization of smart factories.
[0073] In one embodiment of the invention, the processed data information is distributedly stored according to the importance and usage frequency of the data source, including:
[0074] S201, extracting the first-level weight coefficient corresponding to each data source;
[0075] S202, extracting the secondary weight coefficient and data usage frequency corresponding to the data information obtained by each data source;
[0076] S203. Obtain a comprehensive weight coefficient corresponding to each data source by using the first-level weight coefficient, the second-level weight coefficient and the data usage frequency; wherein the comprehensive weight coefficient is obtained by the following formula:
[0077]
[0078] Among them, QZ is the comprehensive weight coefficient, α, β and γ are adjustment coefficients, W1 is the primary weight coefficient of data source 1, n is the total number of data sources, W 1i is the primary weight coefficient of the i-th data source, W2 is the secondary weight coefficient of data source 1, and W 2i is the secondary weight coefficient of the i-th data source, F3 is the data usage frequency of data source 1, and F 3i is the data usage frequency of the ith data source. The value range of the adjustment coefficient is generally 0-1, and can be flexibly adjusted in special cases.
[0079] S204, comparing the comprehensive weight coefficient with a pre-examination first weight coefficient threshold and a second weight coefficient threshold to obtain a comparison result; wherein the value corresponding to the first weight coefficient threshold is greater than the value corresponding to the second weight coefficient threshold;
[0080] S205, classifying the data sources according to the comparison results, and obtaining a primary importance data source, a secondary importance data source, and a tertiary importance data source;
[0081] S206. Distribute and store data for the primary importance data source, the secondary importance data source, and the tertiary importance data source.
[0082] The working principle of the above technical solution is as follows: the first-level weight coefficient reflects the importance of the data source in the factory production process. It may be determined based on factors such as the type of data source (such as key equipment sensors, core monitoring points of the production line, etc.), historical failure rate, and the degree of impact on the production process. The second-level weight coefficient further refines the importance of different data items within the data source. It may be evaluated based on factors such as the contribution of data items to production decisions, real-time requirements, and data quality. The frequency of data use reflects the frequency with which a specific data item is queried, analyzed, or used for decision-making. Data items with high usage frequency may require faster access speeds and higher storage reliability.
[0083] By weighted summation or more complex algorithms, the first-level weight coefficient, the second-level weight coefficient and the frequency of data use are combined to obtain a comprehensive indicator that reflects the overall importance and frequency of use of the data source. The comprehensive weight coefficient is compared with the pre-examination first weight coefficient threshold and the second weight coefficient threshold to obtain the comparison result. According to the actual needs and resource conditions of the factory, two weight coefficient thresholds are pre-set (the first weight coefficient threshold is greater than the second weight coefficient threshold). These two thresholds are used to divide data sources into different levels of importance.
[0084] According to the comparison results, the data sources are divided into importance levels, and the first-level importance data sources, the second-level importance data sources, and the third-level importance data sources are obtained. According to the comparison results of the comprehensive weight coefficient and the threshold, the data sources are divided into first-level importance data sources (highest importance), second-level importance data sources, and third-level importance data sources (lowest importance).
[0085] Different storage strategies are used for distributed storage according to the importance level of the data source. For example, a data source of level 1 importance may adopt a high-availability, low-latency storage solution and set up multiple backups; while a data source of level 3 importance may adopt a lower-cost storage solution and the number of backups is reduced accordingly.
[0086] The effect of the above technical solution is: by dividing the importance, the storage resources are reasonably allocated to ensure that the key data is given priority, while reducing the storage cost of non-key data. For data with high usage frequency and importance, a high-performance storage solution is adopted to improve the data access speed and meet the real-time requirements. Through multiple backup and redundant storage strategies, it is ensured that the key data can still be reliably accessed in the event of a single point failure, and the fault tolerance and stability of the system are improved. The distributed storage architecture supports the dynamic expansion of storage capacity and processing capacity. With the expansion of the factory scale and the increase of production demand, storage nodes can be easily added to meet the needs of future development. It lays a solid data foundation for subsequent intelligent analysis and early warning management, and supports more complex intelligent application scenarios through efficient and reliable data storage solutions. The influence of weights is strengthened by using the above formula, so that high-weight data sources contribute more to the overall results. This helps to more clearly distinguish the influence of high-weight and low-weight data sources. The secondary weight coefficient can help mitigate the possible extreme changes in the original weight value and make the weight growth smoother. For the frequency of use, the importance of frequently used data sources can be emphasized. The influence of a single data source relative to all data sources is strengthened. This helps to achieve detailed analysis of individual data sources and integrated evaluation of overall data sources in the evaluation. Using nonlinear functions can better adapt to different data distribution characteristics, such as processing highly skewed data or data with long tail characteristics. Such processing can make the model more robust and adaptable to complex data environments. By introducing nonlinear elements into the model, the model's ability to capture complex data relationships can be enhanced, which is critical for multi-dimensional and complex evaluation of data source quality and influence.
[0087] In one embodiment of the present invention, data sources are divided into importance categories according to comparison results to obtain primary importance data sources, secondary importance data sources, and tertiary importance data sources, including:
[0088] S2051, comparing the comprehensive weight coefficient with the first weight coefficient threshold and the second weight coefficient threshold of the preliminary examination;
[0089] S2052: When the comprehensive weight coefficient exceeds a preset first weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient exceeding the preset first weight coefficient threshold is a first-level importance data source;
[0090] S2053: When the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold is a secondary importance data source;
[0091] S2054. When the comprehensive weight coefficient does not exceed the preset second weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient not exceeding the preset second weight coefficient threshold is a third-level importance data source.
[0092] The working principle of the above technical solution is as follows: the comprehensive weight coefficient of each data source is compared with two preset weight coefficient thresholds (the first weight coefficient threshold and the second weight coefficient threshold). These two thresholds represent different levels of importance of the data source. If the comprehensive weight coefficient of the data source exceeds the preset first weight coefficient threshold, then the data source is determined to be a first-level importance data source. The first-level importance data source represents the most critical and important data source in the factory production process, and has a vital impact on production decisions and operations.
[0093] If the comprehensive weight coefficient of the data source does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold, then the data source is determined to be a secondary importance data source. Although the secondary importance data source is not as critical as the primary data source, it still has an important impact on the production and operation of the factory.
[0094] If the comprehensive weight coefficient of the data source does not exceed the preset second weight coefficient threshold, the data source is determined to be a third-level importance data source. The third-level importance data source is a relatively less critical data source in the factory production process, but still provides valuable information, but is relatively low in importance and frequency of use.
[0095] The effect of the above technical solution is: by comparing the comprehensive weight coefficient and the preset threshold, the data source can be clearly divided into primary, secondary and tertiary importance data sources, which provides a clear direction for subsequent storage, processing and analysis. According to the importance level of the data source, a more reasonable storage and processing strategy can be formulated. For example, for the primary importance data source, a high-performance and high-reliability storage solution can be adopted, and processing and analysis can be given priority; while for the tertiary importance data source, a relatively economical and inefficient storage solution can be adopted. By clarifying the importance level of the data source, data can be managed and utilized more effectively. For example, when querying, analyzing and visualizing data, priority can be given to primary and secondary importance data sources to obtain more critical and valuable information. Clear data source importance division and optimized storage and processing strategies provide a solid foundation for subsequent intelligent analysis and early warning management. This helps factories achieve more intelligent and efficient production and operation decisions.
[0096] In one embodiment of the present invention, real-time intelligent analysis is performed on the stored data information to obtain the intelligent analysis results, and early warning data information of factory production is generated according to the intelligent analysis results, including:
[0097] S301, retrieving a trained fault identification model from a database; wherein the fault identification model corresponds to a data source;
[0098] S302: input the stored data information corresponding to the data source into a fault identification model for fault identification, and when it is determined that the stored data information corresponding to the data source has abnormal operation data, issue an abnormal operation alarm for the data source.
[0099] The working principle of the above technical solution is as follows: In the early stage, the historical fault data in the factory, expert knowledge and machine learning algorithms are used to train fault identification models corresponding to different data sources. These models can learn and identify possible fault modes in specific data sources. The trained fault identification models are stored in the database so that they can be quickly retrieved when needed. Each fault identification model corresponds to a specific data source to ensure the accuracy and pertinence of the analysis process.
[0100] Retrieve the stored data from the database in real time. The data comes from various data sources in the factory, such as equipment sensors, production line status monitoring systems, etc. Input the retrieved data into the corresponding fault identification model. The model uses the learned fault mode to analyze the data to determine whether there is an operational anomaly. If the model identifies an operational anomaly in the data information, it means that there may be a problem in a certain link in the factory production process, which requires further attention and processing. Once it is determined that the stored data corresponding to the data source has operational anomaly data, the system will automatically trigger the alarm mechanism. The alarm information will detail the specific content of the anomaly, the location of occurrence, and the possible scope of impact, providing clear guidance for factory production personnel. Production personnel quickly locate the source of the problem based on the alarm information and take corresponding measures to deal with it to prevent the fault from expanding and resume production as soon as possible.
[0101] The effect of the above technical solution is: through real-time intelligent analysis of the stored data information, the system can timely discover potential problems in the factory production process and issue early warnings, providing sufficient response time for production personnel. The fault identification model trained by machine learning algorithm has a high recognition accuracy, which can reduce false alarms and missed alarms and improve the reliability of fault identification. By timely discovering and handling abnormal situations in the production process, it is possible to avoid the failure from having a greater impact on the production process, thereby optimizing the production process and improving production efficiency. The application of intelligent analysis system can reduce the workload of manual monitoring and reduce operation and maintenance costs. At the same time, the system's automated processing function also improves the efficiency and accuracy of fault handling. Real-time intelligent analysis results provide accurate and comprehensive data support for factory production personnel, helping them make more informed decisions to cope with various challenges in the production process.
[0102] The embodiment of the present invention proposes a smart factory management system based on AI big data, such as Figure 2 As shown, the smart factory management system based on AI big data includes:
[0103] A real-time identification module is used to identify the data source in the factory in real time, and perform data preprocessing on the data information obtained from the data source to obtain the preprocessed data information;
[0104] A distributed storage module, used for performing distributed storage of the processed data information according to the importance and usage frequency of the data source;
[0105] The fault analysis module is used to perform real-time intelligent analysis on the stored data information, obtain the intelligent analysis results, and generate early warning data information of the factory production according to the intelligent analysis results.
[0106] The working principle of the above technical solution is: using the Internet of Things (IoT) technology, smart factories can identify and access various data sources in real time, including production equipment sensors, video surveillance, personnel positioning systems, etc. The data generated by these data sources is a direct reflection of the factory's operating status. The acquired raw data often contains noise, redundant information or incomplete data, and it is necessary to improve the quality and availability of the data through preprocessing steps such as data cleaning, deduplication, and normalization. The preprocessed data is more accurate and standardized, providing support for subsequent analysis and decision-making.
[0107] Evaluate the importance and frequency of use of data sources: Classify pre-processed data according to the importance of the data source to factory production and the frequency of data use. Highly important and frequently used data needs to be stored first, and its high availability and fast access capabilities must be ensured. Use big data storage technologies such as Hadoop and Spark to store processed data in a distributed manner on multiple nodes. This approach not only increases the storage capacity of data, but also increases the speed of data access and processing through parallel processing. At the same time, according to the importance and frequency of use of the data source, adopt appropriate storage strategies and backup mechanisms to ensure data security and reliability.
[0108] Based on artificial intelligence (AI) technologies, such as machine learning and deep learning, real-time intelligent analysis of stored data is performed. By building various analytical models, such as prediction models, classification models, clustering models, etc., hidden information and value in the data are mined. Based on the results of intelligent analysis, the system automatically identifies abnormal conditions or potential risks in the production process, such as equipment failure, product quality problems, and reduced production efficiency. Once an abnormal condition is detected, the system immediately generates an early warning message and notifies relevant personnel through a visual interface, SMS, email, etc.
[0109] The effects of the above technical solutions are: through real-time monitoring and analysis of production data, timely discover and solve problems in the production process, reduce downtime and resource waste, and improve production efficiency and output. Through intelligent analysis of production data, production parameters and quality standards can be controlled more accurately, unqualified products can be discovered and eliminated in time, and the overall quality of products can be improved. By optimizing data storage and processing strategies, the demand for hardware equipment and network bandwidth can be reduced; at the same time, predictive maintenance can be used to reduce equipment failures and downtime losses, further reducing operating costs. Provide a wealth of data visualization tools and reporting functions to help managers fully understand the factory operation status and market trends, and provide strong support for decision-making. Through real-time monitoring and early warning mechanisms, potential safety risks and equipment failures can be discovered and responded to in a timely manner to ensure safe production and stable operation of the factory.
[0110] In summary, the smart factory management method based on AI big data realizes real-time monitoring, intelligent analysis and early warning management of the production process by integrating the Internet of Things, big data analysis and artificial intelligence technologies, significantly improving the factory's production efficiency, product quality and operational benefits.
[0111] In one embodiment of the present invention, the real-time recognition module includes:
[0112] A data source identification module, used to identify data sources within the factory in real time, wherein the data sources include equipment sensors and production line status monitoring data;
[0113] The data collection module is used to collect data information from each data source in real time and obtain the data information of each data source corresponding to a unit time;
[0114] The data preprocessing module is used to perform data preprocessing on the data information obtained by the data source to obtain the processed data information; wherein the data preprocessing includes timestamp alignment processing, data format unification processing and noise removal processing.
[0115] The working principle of the above technical solution is: the system uses the Internet of Things (IoT) technology to identify various data sources in the factory in real time. These data sources mainly include equipment sensors and production line status monitoring data. Equipment sensors such as temperature sensors, pressure sensors, vibration sensors, etc. are used to monitor the operating status and performance parameters of the equipment; production line status monitoring data includes production speed, output, energy consumption, etc., which are used to reflect the overall operating status of the production line.
[0116] For each identified data source, the system collects data information in real time. By setting appropriate sampling frequency and collection rules, it ensures that the data information corresponding to each data source can be obtained within a unit of time. This data information reflects the real-time status of the factory production process and is the basis for subsequent data preprocessing and analysis.
[0117] Since the data collection time of different data sources may be different, the system first performs timestamp alignment on the data information. By unifying the time base, the consistency and comparability of all data in the time dimension are ensured. The data formats generated by different data sources may be different, such as numeric, text, image, etc. The system unifies the data format and converts data in different formats into a unified standard format to facilitate subsequent data analysis and processing. The original data may contain noise or outliers, which will interfere with subsequent analysis. The system removes noise from the data through algorithms such as smoothing, filtering, and denoising to improve the quality and accuracy of the data.
[0118] The effect of the above technical solution is: by real-time identification of data sources and collection of data information, the system can timely reflect the real-time status of the factory production process and provide support for rapid response to production changes. Through data preprocessing steps, including timestamp alignment, data format unification and noise removal, the system can significantly improve the accuracy and availability of data, and provide support for subsequent data analysis and decision-making. Unified data formats and high-quality data after noise removal can speed up data analysis and improve the efficiency of data analysis. At the same time, real-time also ensures that the data analysis results can be applied to production practice in a timely manner. This technical solution has strong scalability and can flexibly adapt to factory environments of different sizes and types. With the expansion of factory scale and the increase of production equipment, the system can easily expand the data source and data collection scope. It lays the foundation for subsequent intelligent analysis and early warning management. By real-time collection and preprocessing of high-quality data information, the system can more accurately identify abnormal conditions and potential risks in the production process, and provide strong support for the realization of smart factories.
[0119] In one embodiment of the present invention, the distributed storage module includes:
[0120] A first-level weight coefficient extraction module is used to extract the first-level weight coefficient corresponding to each data source;
[0121] A secondary weight coefficient extraction module is used to extract the secondary weight coefficient and data usage frequency corresponding to the data information obtained by each data source;
[0122] A comprehensive weight coefficient acquisition module is used to obtain a comprehensive weight coefficient corresponding to each data source by using the first-level weight coefficient, the second-level weight coefficient and the data usage frequency;
[0123] A comparison module, configured to compare the comprehensive weight coefficient with a first weight coefficient threshold and a second weight coefficient threshold of a preliminary examination to obtain a comparison result; wherein a value corresponding to the first weight coefficient threshold is greater than a value corresponding to the second weight coefficient threshold;
[0124] An importance level classification module is used to classify the importance of data sources according to the comparison results, and obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources;
[0125] The hierarchical storage module is used for distributed storage of data for the first-level importance data source, the second-level importance data source and the third-level importance data source.
[0126] The working principle of the above technical solution is as follows: the first-level weight coefficient reflects the importance of the data source in the factory production process. It may be determined based on factors such as the type of data source (such as key equipment sensors, core monitoring points of the production line, etc.), historical failure rate, and the degree of impact on the production process. The second-level weight coefficient further refines the importance of different data items within the data source. It may be evaluated based on factors such as the contribution of data items to production decisions, real-time requirements, and data quality. The frequency of data use reflects the frequency with which a specific data item is queried, analyzed, or used for decision-making. Data items with high usage frequency may require faster access speeds and higher storage reliability.
[0127] By weighted summation or more complex algorithms, the first-level weight coefficient, the second-level weight coefficient and the frequency of data use are combined to obtain a comprehensive indicator that reflects the overall importance and frequency of use of the data source. The comprehensive weight coefficient is compared with the pre-examination first weight coefficient threshold and the second weight coefficient threshold to obtain the comparison result. According to the actual needs and resource conditions of the factory, two weight coefficient thresholds are pre-set (the first weight coefficient threshold is greater than the second weight coefficient threshold). These two thresholds are used to divide data sources into different levels of importance.
[0128] According to the comparison results, the data sources are divided into importance levels, and the first-level importance data sources, the second-level importance data sources, and the third-level importance data sources are obtained. According to the comparison results of the comprehensive weight coefficient and the threshold, the data sources are divided into first-level importance data sources (highest importance), second-level importance data sources, and third-level importance data sources (lowest importance).
[0129] Different storage strategies are used for distributed storage according to the importance level of the data source. For example, a data source of level 1 importance may adopt a high-availability, low-latency storage solution and set up multiple backups; while a data source of level 3 importance may adopt a lower-cost storage solution and the number of backups is reduced accordingly.
[0130] The effects of the above technical solutions are: by dividing the importance and reasonably allocating storage resources, key data is ensured to be given priority, while reducing the storage cost of non-key data. For data with high usage frequency and importance, a high-performance storage solution is adopted to improve data access speed and meet real-time requirements. Through multiple backup and redundant storage strategies, key data can be reliably accessed in the event of a single point of failure, improving the fault tolerance and stability of the system. The distributed storage architecture supports dynamic expansion of storage capacity and processing power. As the scale of the factory expands and production needs increase, storage nodes can be easily added to meet the needs of future development. It lays a solid data foundation for subsequent intelligent analysis and early warning management, and supports more complex intelligent application scenarios through efficient and reliable data storage solutions.
[0131] In one embodiment of the present invention, the importance level classification module includes:
[0132] A coefficient comparison module, used for comparing the comprehensive weight coefficient with a pre-examination first weight coefficient threshold and a second weight coefficient threshold;
[0133] A first-level importance data source determination module, used for determining that when the comprehensive weight coefficient exceeds a preset first weight coefficient threshold, the data source corresponding to the comprehensive weight coefficient exceeding the preset first weight coefficient threshold is a first-level importance data source;
[0134] A secondary importance data source determination module, for determining that the data source corresponding to the comprehensive weight coefficient not exceeding the preset first weight coefficient threshold but exceeding the preset second weight coefficient threshold is a secondary importance data source when the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold but exceeds the preset second weight coefficient threshold;
[0135] The third-level importance data source determination module is used to determine that the data source corresponding to the comprehensive weight coefficient not exceeding the preset second weight coefficient threshold is a third-level importance data source when the comprehensive weight coefficient does not exceed the preset second weight coefficient threshold.
[0136] The working principle of the above technical solution is as follows: the comprehensive weight coefficient of each data source is compared with two preset weight coefficient thresholds (the first weight coefficient threshold and the second weight coefficient threshold). These two thresholds represent different levels of importance of the data source. If the comprehensive weight coefficient of the data source exceeds the preset first weight coefficient threshold, then the data source is determined to be a first-level importance data source. The first-level importance data source represents the most critical and important data source in the factory production process, and has a vital impact on production decisions and operations.
[0137] If the comprehensive weight coefficient of the data source does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold, then the data source is determined to be a secondary importance data source. Although the secondary importance data source is not as critical as the primary data source, it still has an important impact on the production and operation of the factory.
[0138] If the comprehensive weight coefficient of the data source does not exceed the preset second weight coefficient threshold, the data source is determined to be a third-level importance data source. The third-level importance data source is a relatively less critical data source in the factory production process, but still provides valuable information, but is relatively low in importance and frequency of use.
[0139] The effect of the above technical solution is: by comparing the comprehensive weight coefficient and the preset threshold, the data source can be clearly divided into primary, secondary and tertiary importance data sources, which provides a clear direction for subsequent storage, processing and analysis. According to the importance level of the data source, a more reasonable storage and processing strategy can be formulated. For example, for the primary importance data source, a high-performance and high-reliability storage solution can be adopted, and processing and analysis can be given priority; while for the tertiary importance data source, a relatively economical and inefficient storage solution can be adopted. By clarifying the importance level of the data source, data can be managed and utilized more effectively. For example, when querying, analyzing and visualizing data, priority can be given to primary and secondary importance data sources to obtain more critical and valuable information. Clear data source importance division and optimized storage and processing strategies provide a solid foundation for subsequent intelligent analysis and early warning management. This helps factories achieve more intelligent and efficient production and operation decisions.
[0140] In one embodiment of the present invention, the fault analysis module includes:
[0141] A fault identification model retrieval module is used to retrieve a trained fault identification model from a database; wherein the fault identification model corresponds to a data source;
[0142] The abnormality identification module is used to input the stored data information corresponding to the data source into the fault identification model for fault identification, and when it is determined that the stored data information corresponding to the data source has abnormal operation data, issue an abnormal data source operation alarm.
[0143] The working principle of the above technical solution is as follows: In the early stage, the historical fault data in the factory, expert knowledge and machine learning algorithms are used to train fault identification models corresponding to different data sources. These models can learn and identify possible fault modes in specific data sources. The trained fault identification models are stored in the database so that they can be quickly retrieved when needed. Each fault identification model corresponds to a specific data source to ensure the accuracy and pertinence of the analysis process.
[0144] Retrieve the stored data from the database in real time. The data comes from various data sources in the factory, such as equipment sensors, production line status monitoring systems, etc. Input the retrieved data into the corresponding fault identification model. The model uses the learned fault mode to analyze the data to determine whether there is an operational anomaly. If the model identifies an operational anomaly in the data information, it means that there may be a problem in a certain link in the factory production process, which requires further attention and processing. Once it is determined that the stored data corresponding to the data source has operational anomaly data, the system will automatically trigger the alarm mechanism. The alarm information will detail the specific content of the anomaly, the location of occurrence, and the possible scope of impact, providing clear guidance for factory production personnel. Production personnel quickly locate the source of the problem based on the alarm information and take corresponding measures to deal with it to prevent the fault from expanding and resume production as soon as possible.
[0145] The effect of the above technical solution is: through real-time intelligent analysis of the stored data information, the system can timely discover potential problems in the factory production process and issue early warnings, providing sufficient response time for production personnel. The fault identification model trained by machine learning algorithm has a high recognition accuracy, which can reduce false alarms and missed alarms and improve the reliability of fault identification. By timely discovering and handling abnormal situations in the production process, it is possible to avoid the failure from having a greater impact on the production process, thereby optimizing the production process and improving production efficiency. The application of intelligent analysis system can reduce the workload of manual monitoring and reduce operation and maintenance costs. At the same time, the system's automated processing function also improves the efficiency and accuracy of fault handling. Real-time intelligent analysis results provide accurate and comprehensive data support for factory production personnel, helping them make more informed decisions to cope with various challenges in the production process.
[0146] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A smart factory management method based on AI big data, characterized in that: The smart factory management method based on AI big data includes: Identify the data source in the factory in real time, and perform data preprocessing on the data information obtained from the data source to obtain the preprocessed data information; Distributed storage of the processed data information is performed according to the importance and usage frequency of the data source; Perform real-time intelligent analysis on the stored data information, obtain intelligent analysis results, and generate early warning data information for factory production based on the intelligent analysis results.
2. The smart factory management method based on AI big data according to claim 1 is characterized in that: Identify the data source in the factory in real time and pre-process the data information obtained from the data source. The pre-processed data information includes: Identify data sources within the factory in real time, including equipment sensors and production line status monitoring data; Collect data information for each data source in real time to obtain the data information of each data source corresponding to a unit of time; Data preprocessing is performed on the data information obtained from the data source to obtain the processed data information; wherein the data preprocessing includes timestamp alignment processing, data format unification processing and noise removal processing.
3. The smart factory management method based on AI big data according to claim 1 is characterized in that: The processed data information is stored in a distributed manner according to the importance and usage frequency of the data source, including: Extract the first-level weight coefficient corresponding to each data source; Extract the secondary weight coefficient and data usage frequency corresponding to the data information obtained by each data source; Using the first-level weight coefficient, the second-level weight coefficient and the data usage frequency, obtain a comprehensive weight coefficient corresponding to each data source; Compare the comprehensive weight coefficient with a pre-examination first weight coefficient threshold and a second weight coefficient threshold to obtain a comparison result; wherein the value corresponding to the first weight coefficient threshold is greater than the value corresponding to the second weight coefficient threshold; According to the comparison results, the data sources are divided into importance categories to obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources; Data is distributedly stored for primary importance data sources, secondary importance data sources, and tertiary importance data sources.
4. The smart factory management method based on AI big data according to claim 3 is characterized in that: According to the comparison results, the data sources are divided into importance categories to obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources, including: comparing the comprehensive weight coefficient with a pre-screened first weight coefficient threshold and a second weight coefficient threshold; When the comprehensive weight coefficient exceeds the preset first weight coefficient threshold, the data source corresponding to the comprehensive weight coefficient exceeding the preset first weight coefficient threshold is determined to be a first-level importance data source; When the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold, but exceeds the preset second weight coefficient threshold is a secondary importance data source; When the comprehensive weight coefficient does not exceed the preset second weight coefficient threshold, it is determined that the data source corresponding to the comprehensive weight coefficient not exceeding the preset second weight coefficient threshold is a third-level importance data source.
5. The smart factory management method based on AI big data according to claim 1 is characterized in that: Perform real-time intelligent analysis on the stored data information, obtain intelligent analysis results, and generate early warning data information of factory production based on the intelligent analysis results, including: Retrieving a trained fault identification model from a database; wherein the fault identification model corresponds to a data source; The stored data information corresponding to the data source is input into the fault identification model for fault identification, and when it is determined that the stored data information corresponding to the data source has abnormal operation data, an abnormal operation alarm of the data source is issued.
6. A smart factory management system based on AI big data, characterized in that: The smart factory management system based on AI big data includes: A real-time identification module is used to identify the data source in the factory in real time, and perform data preprocessing on the data information obtained from the data source to obtain the preprocessed data information; A distributed storage module, used for performing distributed storage of the processed data information according to the importance and usage frequency of the data source; The fault analysis module is used to perform real-time intelligent analysis on the stored data information, obtain the intelligent analysis results, and generate early warning data information of the factory production according to the intelligent analysis results.
7. The smart factory management system based on AI big data according to claim 6 is characterized in that: The real-time recognition module comprises: A data source identification module, used to identify data sources within the factory in real time, wherein the data sources include equipment sensors and production line status monitoring data; The data collection module is used to collect data information from each data source in real time and obtain the data information of each data source corresponding to a unit time; The data preprocessing module is used to perform data preprocessing on the data information obtained by the data source to obtain the processed data information; wherein the data preprocessing includes timestamp alignment processing, data format unification processing and noise removal processing.
8. The smart factory management system based on AI big data according to claim 6 is characterized in that: The distributed storage module includes: A first-level weight coefficient extraction module is used to extract the first-level weight coefficient corresponding to each data source; A secondary weight coefficient extraction module is used to extract the secondary weight coefficient and data usage frequency corresponding to the data information obtained by each data source; A comprehensive weight coefficient acquisition module is used to obtain a comprehensive weight coefficient corresponding to each data source by using the first-level weight coefficient, the second-level weight coefficient and the data usage frequency; A comparison module, configured to compare the comprehensive weight coefficient with a first weight coefficient threshold and a second weight coefficient threshold of a preliminary examination to obtain a comparison result; wherein a value corresponding to the first weight coefficient threshold is greater than a value corresponding to the second weight coefficient threshold; An importance level classification module is used to classify the importance of data sources according to the comparison results, and obtain first-level importance data sources, second-level importance data sources, and third-level importance data sources; The hierarchical storage module is used for distributed storage of data for the first-level importance data source, the second-level importance data source and the third-level importance data source.
9. The smart factory management system based on AI big data according to claim 8 is characterized in that: The importance level classification module includes: A coefficient comparison module, used for comparing the comprehensive weight coefficient with a pre-examination first weight coefficient threshold and a second weight coefficient threshold; A first-level importance data source determination module, used for determining that when the comprehensive weight coefficient exceeds a preset first weight coefficient threshold, the data source corresponding to the comprehensive weight coefficient exceeding the preset first weight coefficient threshold is a first-level importance data source; A secondary importance data source determination module, for determining that the data source corresponding to the comprehensive weight coefficient not exceeding the preset first weight coefficient threshold but exceeding the preset second weight coefficient threshold is a secondary importance data source when the comprehensive weight coefficient does not exceed the preset first weight coefficient threshold but exceeds the preset second weight coefficient threshold; The third-level importance data source determination module is used to determine that the data source corresponding to the comprehensive weight coefficient not exceeding the preset second weight coefficient threshold is a third-level importance data source when the comprehensive weight coefficient does not exceed the preset second weight coefficient threshold.
10. The smart factory management system based on AI big data according to claim 6 is characterized in that: The fault analysis module comprises: A fault identification model retrieval module is used to retrieve a trained fault identification model from a database; wherein the fault identification model corresponds to a data source; The abnormality identification module is used to input the stored data information corresponding to the data source into the fault identification model for fault identification, and when it is determined that the stored data information corresponding to the data source has abnormal operation data, issue an abnormal data source operation alarm.
