Artificial intelligence-based production machine operation data management system and method

Abstract

This invention discloses an AI-based data management system and method for production machinery operation, belonging to the field of industrial data management technology. The method includes: real-time acquisition of raw time-series data and derived data during the operation of production machinery, and extraction of data feature information; construction of a data value assessment model to score the total value of the data and predict access popularity, generating storage tags; construction of a data hierarchical decision model based on the storage tags, comprehensively analyzing real-time data access patterns, storage costs, and lifecycle stages, generating dynamic hierarchical instructions to dynamically allocate data to corresponding storage levels; real-time monitoring of data access patterns, dynamically adjusting storage tags, and triggering data migration operations to optimize storage performance and access efficiency. This invention improves the access speed of critical data and reduces the storage cost of low-value data through AI-driven dynamic data hierarchical storage.

Description

Technical Field

[0001] This invention relates to the field of industrial data management technology, specifically to a production machinery operation data management system and method based on artificial intelligence. Background Technology

[0002] In modern industrial production systems, production machinery serves as the core carrier of manufacturing activities, and the massive amounts of data generated during its operation are a crucial foundation for achieving intelligent management, predictive maintenance, and process optimization. This data encompasses raw time-series data such as vibration and temperature directly collected by equipment sensors, as well as derived information such as equipment health indicators and performance reports obtained after analysis and processing. Together, these constitute digital assets that provide insights into production status and support decision-making.

[0003] However, most existing data management systems typically employ a uniform, extensive storage and management strategy for all types of data. This "one-size-fits-all" approach fails to effectively differentiate the significant differences in data value density and access frequency throughout its lifecycle. As a result, core derived data with long-term analytical value, requiring frequent and rapid access, such as health indicators and optimization parameters, are submerged in a massive "sea" of raw data whose value decays rapidly. This leads to long access paths and slow response times for critical data, severely hindering the real-time nature of decision support. Simultaneously, storage systems continuously incur high performance and cost costs for large amounts of low-frequency raw data. Therefore, the "storage performance bottleneck" caused by uneven data value density and differences in access patterns has become a core obstacle restricting the in-depth mining and efficient utilization of production data value. Summary of the Invention

[0004] The purpose of this invention is to provide an artificial intelligence-based production machinery operation data management system and method to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for managing production machinery operation data based on artificial intelligence, the method comprising: Step S1: During the operation of the production equipment, raw time-series data from the equipment sensors are collected in real time, and derived data generated after analysis and processing is obtained; data feature extraction is performed on the raw time-series data and derived data to obtain data feature information; Step S2: Construct a data value assessment model based on the data feature information; use the data value assessment model to perform total value scoring and access popularity prediction on the original time-series data and derived data respectively; generate storage tags for various types of data based on the total value scoring and access popularity prediction results. Step S3: Construct a data tiering decision model based on the storage tags; use the data tiering decision model to comprehensively analyze real-time data access patterns, storage costs, and lifecycle stages to generate dynamic tiering instructions; dynamically allocate various types of data to corresponding storage tiers according to the dynamic tiering instructions. Step S4: Monitor the access patterns of data in each storage tier in real time, dynamically adjust the storage tags according to the access patterns, and trigger data migration operations to optimize storage performance and access efficiency.

[0006] Furthermore, step S1 includes: Step S1-1: Collect raw time-series data generated during operation in real time using various sensors deployed on the production equipment. The raw time-series data includes vibration data, temperature data, pressure data, operating speed data, and energy consumption data. Collect derived data generated after the raw time-series data is analyzed by algorithms, statistically processed, or logically operated. The derived data includes equipment health index, operating efficiency report, fault warning parameters, and process optimization suggestion data. Step S1-2: The collected raw time-series data and derived data are identified by data type. The data category identification is based on the dimension of the production equipment operation status reflected by the data to determine the data category corresponding to each type of data. The data categories include status monitoring, performance evaluation, fault early warning and process optimization. Steps S1-3: Select a preset time period, count the total number of times the corresponding category of data is generated within the time period, and divide the total number of generation by the duration of the time period to obtain the data generation frequency of each type of data; record the total number of times each type of data is queried and called within a preset access time period, and divide the total number of times by the duration of the access time period to obtain the data access frequency of each type of data. Step S1-4: Based on the data categories determined in step S1-2, analyze the data correlation indicators between the data categories. The data correlation indicators include the overlap of generation time, the correlation of data content, and the correlation of usage scenarios. In the above steps, "derived data" refers to data that is further processed using algorithms (such as machine learning models and statistical analysis) based on the collected raw time-series data (such as vibration and temperature) to generate more valuable derived data. For example, raw vibration data may be analyzed using Fast Fourier Transform (FFT) to generate equipment health indices, or fault warning parameters may be generated using statistical models. Derived data usually has higher decision support value. "Determining the data category corresponding to each type of data based on the dimensions of the production machinery operating status reflected by the data" means classifying the data into predefined categories according to the different dimensions of the equipment operating status described by the data. For example, "condition monitoring" data directly reflects the real-time status of the equipment (such as temperature and pressure), "performance evaluation" data evaluates equipment efficiency (such as energy consumption reports), "fault warning" data predicts potential faults (such as abnormal vibration patterns), and "process optimization" data provides production optimization suggestions (such as parameter adjustment guidelines). Therefore, the data classification is mainly based on the content and application purpose of the data.

[0007] Furthermore, steps S1-4 include: Select any two types of data to be analyzed, extract all generation time points of the two types of data within a specified historical time period, determine the generation time interval of each type of data based on the generation time points, calculate the intersection duration of the time intervals of the two types of data, and divide the intersection duration by the union duration of the time intervals of the two types of data to obtain the generation time overlap. Extract descriptive information and feature parameters from the two types of data, count the number of descriptive information and feature parameters that are common to the two types of data, and divide the number of common descriptive information and feature parameters by the total number of descriptive information and feature parameters of the two types of data to obtain the data content correlation degree. The application scenarios of the two types of data are obtained, including production management, equipment maintenance, and process optimization. The number of application scenarios shared by the two types of data is counted, and the correlation of the application scenarios is obtained by dividing the number of shared application scenarios by the total number of application scenarios of the two types of data. Based on the distribution of data correlation indicators within a preset correlation level range, the correlation level between various types of data is determined; if the data correlation indicators are distributed across ranges, the average value of all data correlation indicators is taken, and the correlation level between the data is determined based on the range in which the average value is located. In the above steps, when analyzing the correlation between two types of data content, it is necessary to extract the descriptive information (such as data name, descriptive fields, etc.) and key numerical parameters (such as feature values, thresholds, etc.) of these two types of data. For example, for "vibration data" and "fault warning parameters", the descriptive information may include "equipment ID" and "sampling frequency", and the feature parameters may include "vibration peak value" and "frequency". In the process of calculating the correlation of usage scenarios, different types of data may have common application scenarios. For example, "vibration data" and "fault warning parameters" may be used together in the "equipment maintenance" scenario, while "energy consumption data" and "process optimization suggestions" may be used together in the "process optimization" scenario. Calculating the correlation of usage scenarios reflects the synergy of data in business.

[0008] Furthermore, step S2 includes: Step S2-1: Using the data feature information obtained in step S1 as model input parameters, construct a data value assessment model; wherein, the data feature information includes data type, data generation frequency, data access frequency, and data correlation level, and the data value assessment model includes a value scoring algorithm for calculating the total value of the data and an access popularity prediction algorithm for predicting the access frequency of the data. Step S2-2: Calculate the total value of the data using a data value assessment model. The specific process is as follows: Based on the importance of different data categories to the management of production equipment, a corresponding basic score range is set for each data category; Set the weight ratio of data access frequency and data generation frequency. Based on the basic score range of each type of data access frequency and data generation frequency, determine the corresponding bonus items. Multiply the bonus items by their corresponding weights and sum them to obtain the frequency-weighted score. Based on the data correlation level obtained in steps S1-4, a corresponding correction score is set for different correlation levels. The higher the correlation level, the higher the correction score, and the correlation correction score is obtained. The total value score of the data is obtained by summing the basic score of the data type, the frequency weighted score, and the correlation correction score. The total value score is then divided into levels according to a preset value range. In the above steps, when assessing the value of data, it is necessary to consider the differences in importance of different data categories in equipment management. For example, fault early warning data is usually more important than condition monitoring data because the former is directly related to equipment safety and downtime costs. Importance can be determined through expert experience or historical impact analysis. Step S2-3: Predict the access popularity of the data based on the data value assessment model. The specific process includes: Obtain access frequency data for various types of data within a specified historical time period, and analyze the changing patterns of access frequency using a trend fitting algorithm to obtain historical access trends; Collect historical operating condition fluctuation data of production machinery and future production task scheduling plans, analyze the impact of different operating conditions and different production tasks on data access frequency, and obtain the correction coefficients of operating conditions and scheduling on access frequency. Based on historical access trends, the basic access popularity is predicted, and the basic access popularity is adjusted by combining the correction coefficients of working conditions and scheduling to obtain the access popularity prediction results of various types of data within a specified time period in the future, and the prediction results are matched with the corresponding preset popularity level. Step S2-4: Generate storage tags based on the total value score and access popularity prediction results; the storage tags include a first storage tag, a second storage tag, and a third storage tag; In the above steps, storage labels are generated based on the total value score and access popularity prediction results. For example: when the total value score of the data is in the high value range and the access popularity prediction is high, a priority storage label is generated; when the total value score of the data is in the medium value range and the access popularity prediction is medium, a standard storage label is generated; when the total value score of the data is in the low value range and the access popularity prediction is low, an archive storage label is generated; if the score and prediction result are in different ranges, the corresponding storage label is generated according to the "total value score priority" principle.

[0009] Furthermore, step S3 includes: Step S3-1: Based on the storage tags generated in step S2, and combined with the performance parameters of the storage system, the cost differences of different storage media, and data lifecycle management rules, a data stratification decision model is constructed. The data stratification decision model is used to output the mapping relationship between data and the target storage level. Step S3-2: Perform multi-dimensional analysis of the data using a data-layered decision-making model, including: Select a preset statistical period and count the number of accesses, peak access periods, and duration of a single access for various types of data within that period, as real-time access characteristics of the data. Calculate the hardware procurement cost, daily operation and maintenance cost, and energy consumption cost corresponding to different storage tiers, and determine the cost priority of each storage tier. Based on the data generation time, access frequency, and relevance to current production and business needs, data is divided into several stages of its lifecycle. In the above steps, the statistical period needs to be set according to the business characteristics. For example, in a continuous production environment, a shorter period (such as 1 hour) may be selected to capture real-time changes. Step S3-3: The data stratification decision model integrates real-time data access characteristics, cost priority of each storage level, and data lifecycle stage to generate dynamic stratification instructions, in which the target storage level corresponding to each type of data is clearly defined; In the above steps, the storage hierarchy includes a first storage layer for storing high-value data that is accessed frequently, a second storage layer for storing medium-value data that is accessed regularly, and a third storage layer for storing low-value data that is accessed infrequently. According to the dynamic tiering instruction, data with priority storage tags is allocated to the first storage layer, data with standard storage tags is allocated to the second storage layer, and data with archive storage tags is allocated to the third storage layer, thus completing the initial tiered storage of data.

[0010] Furthermore, step S4 includes: Step S4-1: Through the monitoring components of the storage system, collect access logs of various types of data in each storage level in real time, extract access time, access count, access duration, and access subject information from the access logs, generate access pattern statistical reports according to preset cycles, and continuously track changes in data access patterns. In the above steps, access subject information refers to the data accesser identity information recorded in the access log, such as user ID, application name, IP address, etc. This helps to analyze the source of access patterns, such as distinguishing between manual queries and system access, thereby more accurately predicting access popularity; the access pattern statistics report is used to summarize the data access situation in each storage layer, including access count, peak time period, distribution of access subjects, etc. This report is used to monitor data usage trends and provide a basis for dynamic adjustment; Step S4-2: Set a preset threshold for the magnitude of changes in access patterns. When the amount of changes in data access patterns tracked in step S4-1 reaches or exceeds the preset threshold, re-evaluate the total value score of the data according to the method in step S2-2, predict the access popularity of the data according to the method in step S2-3, and dynamically adjust the storage tag corresponding to the data based on the re-evaluation of the total value score and the predicted access popularity results. In the above steps, when the access pattern changes exceed the threshold, the value assessment and popularity prediction are re-executed, and the storage tags are updated. This ensures that the storage policy adapts to the data usage and avoids the inefficiency caused by static policies. Step S4-3: Triggering the data migration operation based on the storage tags adjusted in step S4-2 includes: Data migration is triggered within a preset time after the access pattern change is determined; during the migration process, data is sorted from low to high according to its current access frequency, and data with low access frequency is migrated first to reduce the impact on normal access to active data. Data is migrated from the current storage tier to the target storage tier corresponding to the adjusted storage tag according to the migration priority. During the migration process, the migration progress and data integrity are monitored in real time. After the migration is completed, the data storage location information is updated to the system metadata database to ensure that subsequent access can be accurately located, thereby optimizing storage performance and access efficiency. In the above steps, after data migration, the metadata database (a database that records the location of data storage) is updated so that the system can correctly route access requests. This is a key step in achieving dynamic management, ensuring that access is not interrupted after data migration and that storage resources are used optimally.

[0011] Furthermore, to better implement the above methods, an artificial intelligence-based production machinery operation data management system is also provided. This management system includes: a data acquisition module, a data feature extraction module, a data value assessment module, a data hierarchical decision-making module, a storage allocation module, an access monitoring module, a label adjustment module, and a data migration module. The data acquisition module is used to collect raw time-series data from the device's sensors in real time and obtain derived data generated after analysis and processing. The data extraction module is used to receive data output from the data acquisition module, perform data type identification, frequency calculation, and correlation analysis on the raw time series data and derived data, and obtain data feature information. The data value assessment module is used to build a data value assessment model based on the data feature information output by the data feature extraction module, to score the total value of the data and predict its access popularity, and to generate storage tags. The data stratification decision module is used to build a data stratification decision model based on the storage tags output by the data value assessment module, and to generate dynamic stratification instructions by comprehensively analyzing real-time data access patterns, storage costs and lifecycle stages. The storage allocation module is used to dynamically allocate various types of data to the corresponding storage levels according to the dynamic stratification instructions output by the data stratification decision module. The access monitoring module is used to collect access logs of data in each storage tier in real time and track changes in data access patterns. The tag adjustment module is used to reassess the total data value score and access popularity based on changes in access patterns tracked by the access monitoring module, and dynamically adjust the storage tags. The data migration module is used to trigger and execute data migration operations based on the adjusted storage tags output by the tag adjustment module.

[0012] Furthermore, the data feature extraction module includes a data type identification unit, a frequency calculation unit, and a correlation analysis unit; The data type identification unit is used to classify the data output by the data acquisition module and determine the data category corresponding to each type of data. The frequency calculation unit is used to set the statistical period, read data generation records and access logs, and calculate the data generation frequency and data access frequency of various types of data. The correlation analysis unit is used to calculate the overlap of generation time, correlation of data content, and correlation of usage scenarios of different types of data, and to determine the correlation level of data.

[0013] Furthermore, the data value assessment module includes a model building unit, a value scoring unit, a popularity prediction unit, and a tag generation unit; The model building unit is used to build a data value scoring model that includes a total value scoring algorithm and an access popularity prediction algorithm, using data feature information as input parameters. The value scoring unit is used to calculate the total value score of the data through the data value assessment model and to divide the data into value ranges. The popularity prediction unit is used to predict the access popularity of data through a data value assessment model and classify the popularity level. The tag generation unit is used to generate corresponding storage tags based on the value range output by the value scoring unit and the popularity level output by the popularity prediction unit.

[0014] Furthermore, the data layering strategy module includes a layered model construction unit, a multi-dimensional analysis unit, and an instruction generation unit; The hierarchical model building unit is used to build a data hierarchical decision model by combining storage tags, storage system performance parameters, storage media cost differences and data lifecycle management rules; The multi-dimensional analysis unit is used to analyze real-time data access characteristics, calculate the cost of each storage level, and determine the data lifecycle stage. The instruction generation unit is used to integrate the analysis results of the multi-dimensional analysis unit to generate dynamic hierarchical instructions that include data identifiers and target storage level identifiers.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. By constructing a data value assessment model, which integrates data characteristics, access frequency, and lifecycle stage, intelligent data classification and hierarchical storage are achieved. This ensures that high-value, high-access-frequency data is preferentially stored in the high-performance layer, significantly improving the efficiency of critical data access and system response speed.

[0016] 2. By monitoring changes in data access patterns in real time, storage tags are dynamically adjusted and data migration is triggered, enabling storage strategies to adapt to business needs, avoiding resource waste caused by static strategies, effectively reducing storage costs, and improving storage resource utilization.

[0017] 3. By combining data generation, access patterns, and business relevance, we can achieve refined management of data throughout its entire lifecycle from generation to archiving, enhance the overall efficiency of the system, support predictive maintenance and process optimization, and improve the level of intelligence in the operation of production machinery. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the production machinery operation data management system and method based on artificial intelligence of the present invention. Figure 2 This is a schematic diagram of the system structure of the production machinery operation data management system and method based on artificial intelligence of the present invention. Detailed Implementation

[0019] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: As Figure 1 As shown, the present invention provides a technical solution: a method for managing production machinery operation data based on artificial intelligence, the method comprising: Step S1: During the operation of the production equipment, raw time-series data from the equipment sensors are collected in real time, and derived data generated after analysis and processing is obtained; data feature extraction is performed on the raw time-series data and derived data to obtain data feature information; Step S2: Construct a data value assessment model based on the data feature information; use the data value assessment model to perform total value scoring and access popularity prediction on the original time-series data and derived data respectively; generate storage tags for various types of data based on the total value scoring and access popularity prediction results. Step S3: Construct a data tiering decision model based on the storage tags; use the data tiering decision model to comprehensively analyze real-time data access patterns, storage costs, and lifecycle stages to generate dynamic tiering instructions; dynamically allocate various types of data to corresponding storage tiers according to the dynamic tiering instructions. Step S4: Monitor the access patterns of data in each storage tier in real time, dynamically adjust the storage tags according to the access patterns, and trigger data migration operations to optimize storage performance and access efficiency. Step S1 includes: Step S1-1: Collect raw time-series data generated during operation in real time using various sensors deployed on the production equipment. The raw time-series data includes vibration data, temperature data, pressure data, operating speed data, and energy consumption data. Collect derived data generated after the raw time-series data is analyzed by algorithms, statistically processed, or logically operated. The derived data includes equipment health index, operating efficiency report, fault warning parameters, and process optimization suggestion data. Step S1-2: The collected raw time-series data and derived data are identified by data type. The data category identification is based on the dimension of the production equipment operation status reflected by the data to determine the data category corresponding to each type of data. The data categories include status monitoring, performance evaluation, fault early warning and process optimization. Steps S1-3: Select a preset time period, count the total number of times the corresponding category of data is generated within the time period, and divide the total number of generation by the duration of the time period to obtain the data generation frequency of each type of data; record the total number of times each type of data is queried and called within a preset access time period, and divide the total number of times by the duration of the access time period to obtain the data access frequency of each type of data. Step S1-4: Based on the data categories determined in step S1-2, analyze the data correlation indicators between the data categories. The data correlation indicators include the overlap of generation time, the correlation of data content, and the correlation of usage scenarios. Steps S1-4 include: Select any two types of data to be analyzed, extract all generation time points of the two types of data within a specified historical time period, determine the generation time interval of each type of data based on the generation time points, calculate the intersection duration of the time intervals of the two types of data, and divide the intersection duration by the union duration of the time intervals of the two types of data to obtain the generation time overlap. Extract descriptive information and feature parameters from the two types of data, count the number of descriptive information and feature parameters that are common to the two types of data, and divide the number of common descriptive information and feature parameters by the total number of descriptive information and feature parameters of the two types of data to obtain the data content correlation degree. The application scenarios of the two types of data are obtained, including production management, equipment maintenance, and process optimization. The number of application scenarios shared by the two types of data is counted, and the correlation of the application scenarios is obtained by dividing the number of shared application scenarios by the total number of application scenarios of the two types of data. Based on the distribution of data correlation indicators within a preset correlation level range, the correlation level between various types of data is determined; if the data correlation indicators are distributed across ranges, the average value of all data correlation indicators is taken, and the correlation level between the data is determined based on the range in which the average value is located. Step S2 includes: Step S2-1: Using the data feature information obtained in step S1 as model input parameters, construct a data value assessment model; wherein, the data feature information includes data type, data generation frequency, data access frequency, and data correlation level, and the data value assessment model includes a value scoring algorithm for calculating the total value of the data and an access popularity prediction algorithm for predicting the access frequency of the data. Step S2-2: Calculate the total value of the data using a data value assessment model. The specific process is as follows: Based on the importance of different data categories to the management of production equipment, a corresponding basic score range is set for each data category; Set the weight ratio of data access frequency and data generation frequency. Based on the basic score range of each type of data access frequency and data generation frequency, determine the corresponding bonus items. Multiply the bonus items by their corresponding weights and sum them to obtain the frequency-weighted score. Based on the data correlation level obtained in steps S1-4, a corresponding correction score is set for different correlation levels. The higher the correlation level, the higher the correction score, and the correlation correction score is obtained. The total value score of the data is obtained by summing the basic score of the data type, the frequency weighted score, and the correlation correction score. The total value score is then divided into levels according to a preset value range. Step S2-3: Predict the access popularity of the data based on the data value assessment model. The specific process includes: Obtain access frequency data for various types of data within a specified historical time period, and analyze the changing patterns of access frequency using a trend fitting algorithm to obtain historical access trends; Collect historical operating condition fluctuation data of production machinery and future production task scheduling plans, analyze the impact of different operating conditions and different production tasks on data access frequency, and obtain the correction coefficients of operating conditions and scheduling on access frequency. Based on historical access trends, the basic access popularity is predicted, and the basic access popularity is adjusted by combining the correction coefficients of working conditions and scheduling to obtain the access popularity prediction results of various types of data within a specified time period in the future, and the prediction results are matched with the corresponding preset popularity level. Step S2-4: Generate storage tags based on the total value score and access popularity prediction results; the storage tags include a first storage tag, a second storage tag, and a third storage tag; Step S3 includes: Step S3-1: Based on the storage tags generated in step S2, and combined with the performance parameters of the storage system, the cost differences of different storage media, and data lifecycle management rules, a data stratification decision model is constructed. The data stratification decision model is used to output the mapping relationship between data and the target storage level. Step S3-2: Perform multi-dimensional analysis of the data using a data-layered decision-making model, including: Select a preset statistical period and count the number of accesses, peak access periods, and duration of a single access for various types of data within that period, as real-time access characteristics of the data. Calculate the hardware procurement cost, daily operation and maintenance cost, and energy consumption cost corresponding to different storage tiers, and determine the cost priority of each storage tier. Based on the data generation time, access frequency, and relevance to current production and business needs, data is divided into several stages of its lifecycle. Step S3-3: The data stratification decision model integrates real-time data access characteristics, cost priority of each storage level, and data lifecycle stage to generate dynamic stratification instructions, in which the target storage level corresponding to each type of data is clearly defined; Step S4 includes: Step S4-1: Through the monitoring components of the storage system, collect access logs of various types of data in each storage level in real time, extract access time, access count, access duration, and access subject information from the access logs, generate access pattern statistical reports according to preset cycles, and continuously track changes in data access patterns. Step S4-2: Set a preset threshold for the magnitude of changes in access patterns. When the amount of changes in data access patterns tracked in step S4-1 reaches or exceeds the preset threshold, re-evaluate the total value score of the data according to the method in step S2-2, predict the access popularity of the data according to the method in step S2-3, and dynamically adjust the storage tag corresponding to the data based on the re-evaluation of the total value score and the predicted access popularity results. Step S4-3: Triggering the data migration operation based on the storage tags adjusted in step S4-2 includes: Data migration is triggered within a preset time after the access pattern change is determined; during the migration process, data is sorted from low to high according to its current access frequency, and data with low access frequency is migrated first to reduce the impact on normal access to active data. Data is migrated from the current storage level to the target storage level corresponding to the adjusted storage tag according to the migration priority. During the migration process, the migration progress and data integrity are monitored in real time. After the migration is completed, the data storage location information is updated to the system metadata database. Example 2: Figure 2 As shown, in order to better implement the above method, an artificial intelligence-based production machinery operation data management system is also provided. The management system includes: a data acquisition module, a data feature extraction module, a data value assessment module, a data hierarchical decision-making module, a storage allocation module, an access monitoring module, a label adjustment module, and a data migration module. The data acquisition module is used to collect raw time-series data from the device's sensors in real time and obtain derived data generated after analysis and processing. The data extraction module is used to receive data output from the data acquisition module, perform data type identification, frequency calculation, and correlation analysis on the raw time series data and derived data, and obtain data feature information. The data value assessment module is used to build a data value assessment model based on the data feature information output by the data feature extraction module, to score the total value of the data and predict its access popularity, and to generate storage tags. The data stratification decision module is used to build a data stratification decision model based on the storage tags output by the data value assessment module, and to generate dynamic stratification instructions by comprehensively analyzing real-time data access patterns, storage costs and lifecycle stages. The storage allocation module is used to dynamically allocate various types of data to the corresponding storage levels according to the dynamic stratification instructions output by the data stratification decision module. The access monitoring module is used to collect access logs of data in each storage tier in real time and track changes in data access patterns. The tag adjustment module is used to reassess the total data value score and access popularity based on changes in access patterns tracked by the access monitoring module, and dynamically adjust the storage tags. The data migration module is used to trigger and execute data migration operations based on the adjusted storage tags output by the tag adjustment module. The data feature extraction module includes a data type identification unit, a frequency calculation unit, and a correlation analysis unit. The data type identification unit is used to classify the data output by the data acquisition module and determine the data category corresponding to each type of data. The frequency calculation unit is used to set the statistical period, read data generation records and access logs, and calculate the data generation frequency and data access frequency of various types of data. The correlation analysis unit is used to calculate the overlap of generation time, correlation of data content, and correlation of usage scenarios of different types of data, and to determine the correlation level of data. The data value assessment module includes a model building unit, a value scoring unit, a popularity prediction unit, and a tag generation unit. The model building unit is used to build a data value scoring model that includes a total value scoring algorithm and an access popularity prediction algorithm, using data feature information as input parameters. The value scoring unit is used to calculate the total value score of the data through the data value assessment model and to divide the data into value ranges. The popularity prediction unit is used to predict the access popularity of data through a data value assessment model and classify the popularity level. The tag generation unit is used to generate corresponding storage tags based on the value range output by the value scoring unit and the popularity level output by the popularity prediction unit. The data layering strategy module includes a layered model construction unit, a multi-dimensional analysis unit, and an instruction generation unit. The hierarchical model building unit is used to build a data hierarchical decision model by combining storage tags, storage system performance parameters, storage media cost differences and data lifecycle management rules; The multi-dimensional analysis unit is used to analyze real-time data access characteristics, calculate the cost of each storage level, and determine the data lifecycle stage. The instruction generation unit is used to integrate the analysis results of the multi-dimensional analysis unit to generate dynamic hierarchical instructions that include data identifiers and target storage level identifiers. In embodiments of this invention, the operating data of multiple CNC machine tools are managed. First, raw time-series data such as vibration, temperature, and pressure are collected in real time using sensors built into the machine tools, and derived data such as equipment health index and fault warning parameters are generated using analysis algorithms. Subsequently, feature extraction is performed on the data to identify data types (such as condition monitoring and fault warning), the data generation frequency and access frequency are calculated, and the correlation between data is analyzed (such as the high correlation between vibration data and fault warning parameters). Through a data value assessment model, the data is scored for total value and access popularity is predicted, and storage tags are generated: fault warning data is marked as priority storage tag due to its high value and high popularity and is allocated to the high-speed storage layer; while historical temperature data is marked as archive storage tag due to its low access frequency and is allocated to the low-cost storage layer. According to dynamic tiering instructions, the system automatically allocates data to the corresponding storage level and monitors access patterns in real time. When a device enters a maintenance cycle, the access frequency of its related data increases significantly, and the system triggers a reassessment and migration operation to migrate the data from the archive layer to the high-speed layer, ensuring that maintenance personnel can access it quickly, thereby optimizing storage performance and access efficiency.

[0021] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing production machinery operation data based on artificial intelligence, characterized in that: The management method includes: Step S1: During the operation of the production equipment, raw time-series data from the equipment sensors are collected in real time, and derived data generated after analysis and processing is obtained; data feature extraction is performed on the raw time-series data and derived data to obtain data feature information; Step S2: Construct a data value assessment model based on the data feature information; use the data value assessment model to perform total value scoring and access popularity prediction on the original time-series data and derived data respectively; generate storage tags for various types of data based on the total value scoring and access popularity prediction results. Step S3: Construct a data tiering decision model based on the storage tags; use the data tiering decision model to comprehensively analyze real-time data access patterns, storage costs, and lifecycle stages to generate dynamic tiering instructions; dynamically allocate various types of data to corresponding storage tiers according to the dynamic tiering instructions. Step S4: Monitor the access patterns of data in each storage tier in real time, dynamically adjust the storage tags according to the access patterns, and trigger data migration operations to optimize storage performance and access efficiency.

2. The method for managing production machinery operation data based on artificial intelligence according to claim 1, characterized in that: Step S1 includes: Step S1-1: Collect raw time-series data generated during operation in real time using various sensors deployed on the production equipment. The raw time-series data includes vibration data, temperature data, pressure data, operating speed data, and energy consumption data. Collect derived data generated after the raw time-series data is analyzed by algorithms, statistically processed, or logically operated. The derived data includes equipment health index, operating efficiency report, fault warning parameters, and process optimization suggestion data. Step S1-2: The collected raw time-series data and derived data are identified by data type. The data category identification is based on the dimension of the production equipment operation status reflected by the data to determine the data category corresponding to each type of data. The data categories include status monitoring, performance evaluation, fault early warning and process optimization. Steps S1-3: Select a preset time period, count the total number of times the corresponding category of data is generated within the time period, and divide the total number of generation by the duration of the time period to obtain the data generation frequency of each type of data; record the total number of times each type of data is queried and called within a preset access time period, and divide the total number of times by the duration of the access time period to obtain the data access frequency of each type of data. Step S1-4: Based on the data categories determined in step S1-2, analyze the data correlation indicators between the various data categories. The data correlation indicators include the overlap of generation time, the correlation of data content, and the correlation of usage scenarios.

3. The method for managing production machinery operation data based on artificial intelligence according to claim 2, characterized in that: Steps S1-4 include: Select any two types of data to be analyzed, extract all generation time points of the two types of data within a specified historical time period, determine the generation time interval of each type of data based on the generation time points, calculate the intersection duration of the time intervals of the two types of data, and divide the intersection duration by the union duration of the time intervals of the two types of data to obtain the generation time overlap. Extract descriptive information and feature parameters from the two types of data, count the number of descriptive information and feature parameters that are common to the two types of data, and divide the number of common descriptive information and feature parameters by the total number of descriptive information and feature parameters of the two types of data to obtain the data content correlation degree. The application scenarios of the two types of data are obtained, including production management, equipment maintenance, and process optimization. The number of application scenarios shared by the two types of data is counted, and the correlation of the application scenarios is obtained by dividing the number of shared application scenarios by the total number of application scenarios of the two types of data. Based on the distribution of data correlation indicators within a preset correlation level range, the correlation level between various types of data is determined; if the data correlation indicators are distributed across ranges, the average value of all data correlation indicators is taken, and the correlation level between the data is determined based on the range in which the average value is located.

4. The method for managing production machinery operation data based on artificial intelligence according to claim 1, characterized in that: Step S2 includes: Step S2-1: Using the data feature information obtained in step S1 as model input parameters, construct a data value assessment model; wherein, the data feature information includes data type, data generation frequency, data access frequency, and data correlation level, and the data value assessment model includes a value scoring algorithm for calculating the total value of the data and an access popularity prediction algorithm for predicting the access frequency of the data. Step S2-2: Calculate the total value of the data using a data value assessment model. The specific process is as follows: Based on the importance of different data categories to the management of production equipment, a corresponding basic score range is set for each data category; Set the weight ratio of data access frequency and data generation frequency. Based on the basic score range of each type of data access frequency and data generation frequency, determine the corresponding bonus items. Multiply the bonus items by their corresponding weights and sum them to obtain the frequency-weighted score. Based on the data correlation level obtained in steps S1-4, a corresponding correction score is set for different correlation levels. The higher the correlation level, the higher the correction score, and the correlation correction score is obtained. The total value score of the data is obtained by summing the basic score of the data type, the frequency weighted score, and the correlation correction score. The total value score is then divided into levels according to a preset value range. Step S2-3: Predict the access popularity of the data based on the data value assessment model. The specific process includes: Obtain access frequency data for various types of data within a specified historical time period, and analyze the changing patterns of access frequency using a trend fitting algorithm to obtain historical access trends; Collect historical operating condition fluctuation data of production machinery and future production task scheduling plans, analyze the impact of different operating conditions and different production tasks on data access frequency, and obtain the correction coefficients of operating conditions and scheduling on access frequency. Based on historical access trends, the basic access popularity is predicted, and the basic access popularity is adjusted by combining the correction coefficients of working conditions and scheduling to obtain the access popularity prediction results of various types of data within a specified time period in the future, and the prediction results are matched with the corresponding preset popularity level. Step S2-4: Generate storage tags based on the total value score and access popularity prediction results; the storage tags include a first storage tag, a second storage tag, and a third storage tag.

5. The method for managing production machinery operation data based on artificial intelligence according to claim 1, characterized in that: Step S3 includes: Step S3-1: Based on the storage tags generated in step S2, and combined with the performance parameters of the storage system, the cost differences of different storage media, and data lifecycle management rules, a data stratification decision model is constructed. The data stratification decision model is used to output the mapping relationship between data and the target storage level. Step S3-2: Perform multi-dimensional analysis of the data using a data-layered decision-making model, including: Select a preset statistical period and count the number of accesses, peak access periods, and duration of a single access for various types of data within that period, as real-time access characteristics of the data. Calculate the hardware procurement cost, daily operation and maintenance cost, and energy consumption cost corresponding to different storage tiers, and determine the cost priority of each storage tier. Based on the data generation time, access frequency, and relevance to current production and business needs, data is divided into several stages of its lifecycle. Step S3-3: The data stratification decision model integrates real-time data access characteristics, cost priorities of each storage level, and data lifecycle stages to generate dynamic stratification instructions, in which the target storage level corresponding to each type of data is clearly defined.

6. The method for managing production machinery operation data based on artificial intelligence according to claim 1, characterized in that: Step S4 includes: Step S4-1: Through the monitoring components of the storage system, collect access logs of various types of data in each storage level in real time, extract access time, access count, access duration, and access subject information from the access logs, generate access pattern statistical reports according to preset cycles, and continuously track changes in data access patterns. Step S4-2: Set a preset threshold for the magnitude of changes in access patterns. When the amount of changes in data access patterns tracked in step S4-1 reaches or exceeds the preset threshold, re-evaluate the total value score of the data according to the method in step S2-2, predict the access popularity of the data according to the method in step S2-3, and dynamically adjust the storage tag corresponding to the data based on the re-evaluation of the total value score and the predicted access popularity results. Step S4-3: Triggering the data migration operation based on the storage tags adjusted in step S4-2 includes: Data migration is triggered within a preset time after the access pattern change is determined; during the migration process, data is sorted from low to high according to its current access frequency, and data with low access frequency is migrated first to reduce the impact on normal access to active data. Data is migrated from the current storage tier to the target storage tier corresponding to the adjusted storage tag according to the migration priority. The migration progress and data integrity are monitored in real time during the migration process. After the migration is completed, the data storage location information is updated to the system metadata database.

7. An AI-based production machinery operation data management system, used to execute the AI-based production machinery operation data management method according to any one of claims 1-5, characterized in that: The management system includes: a data acquisition module, a data feature extraction module, a data value assessment module, a data hierarchical decision-making module, a storage allocation module, an access monitoring module, a tag adjustment module, and a data migration module; The data acquisition module is used to acquire raw time-series data from the device's sensors in real time and obtain derived data generated after analysis and processing. The data extraction module is used to receive data output by the data acquisition module, and to perform data type identification, frequency calculation, and correlation analysis on the original time-series data and derived data to obtain data feature information. The data value assessment module is used to construct a data value assessment model based on the data feature information output by the data feature extraction module, to score the total value of the data and predict its access popularity, and to generate storage tags. The data stratification decision module is used to construct a data stratification decision model based on the storage tags output by the data value assessment module, comprehensively analyze the real-time data access pattern, storage cost and life cycle stage, and generate dynamic stratification instructions. The storage allocation module is used to dynamically allocate various types of data to the corresponding storage level according to the dynamic stratification instructions output by the data stratification decision module. The access monitoring module is used to collect access logs of data in each storage level in real time and track changes in data access patterns. The tag adjustment module is used to reassess the total data value score and access popularity based on the changes in access patterns tracked by the access monitoring module, and dynamically adjust the storage tags. The data migration module is used to trigger and execute a data migration operation based on the adjusted storage tags output by the tag adjustment module.

8. The production machinery operation data management system based on artificial intelligence according to claim 1, characterized in that: The data feature extraction module includes a data type identification unit, a frequency calculation unit, and a correlation analysis unit; The data type identification unit is used to classify the data output by the data acquisition module and determine the data category corresponding to each type of data. The frequency calculation unit is used to set the statistical period, read data generation records and access logs, and calculate the data generation frequency and data access frequency of various types of data. The correlation analysis unit is used to calculate the overlap of generation time, correlation of data content, and correlation of usage scenarios of different types of data, and to determine the correlation level of data.

9. The production machinery operation data management system based on artificial intelligence according to claim 1, characterized in that: The data value assessment module includes a model building unit, a value scoring unit, a popularity prediction unit, and a tag generation unit; The model building unit is used to build a data value scoring model that includes a total value scoring algorithm and an access popularity prediction algorithm, using data feature information as input parameters. The value scoring unit is used to calculate the total value score of the data through the data value assessment model and to divide the value range. The popularity prediction unit is used to predict the access popularity of data through a data value assessment model and classify the popularity level. The tag generation unit is used to generate corresponding storage tags based on the value range output by the value scoring unit and the popularity level output by the popularity prediction unit.

10. The production machinery operation data management system based on artificial intelligence according to claim 1, characterized in that: The data layering strategy module includes a layered model construction unit, a multi-dimensional analysis unit, and an instruction generation unit; The hierarchical model construction unit is used to build a data hierarchical decision model by combining storage tags, storage system performance parameters, storage medium cost differences and data lifecycle management rules. The multi-dimensional analysis unit is used to analyze real-time data access characteristics, calculate the cost of each storage level, and determine the data lifecycle stage. The instruction generation unit is used to integrate the analysis results of the multi-dimensional analysis unit to generate dynamic hierarchical instructions that include data identifiers and target storage level identifiers.

Images