Industrial big data system for industrial data analysis
By integrating the uncertainty quantization engine and adaptive model training mechanism, the data uncertainty problem in industrial big data systems is solved, high-precision prediction in complex environments is achieved, equipment failure downtime is reduced, and system adaptability and prediction accuracy are improved.
Patent Information
- Application Number
- CN202510377962.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-28
- Publication Date
- 2025-06-20
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing industrial big data systems have shortcomings in processing data uncertainty, especially when dealing with measurement errors and external interference in the input data, which affects the accuracy and reliability of the prediction model. Data loss or delay increases data uncertainty, making it difficult to maintain high-precision prediction results under complex operating conditions.
The input data is evaluated and quantified by the integrated uncertainty quantization engine, a detailed uncertainty report is generated, and an adaptive model training mechanism is used to dynamically adjust the analysis model parameters or select appropriate algorithms. At the same time, a multi-model fusion strategy is used to allocate weights based on the performance of each model on historical data to improve prediction accuracy.
High-precision prediction can still be maintained in highly uncertain data environments, reducing downtime caused by equipment failures, and improving the system's adaptability and prediction accuracy in complex and changing environments.
Smart Images
Figure CN120180041A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of industrial big data, and specifically, to an industrial big data system for industrial data analysis. Background Art
[0002] In modern industrial production, the wide application of Internet of Things technology enables devices and production lines to generate a large amount of data in real time, covering key information such as device operation status, environmental parameters, and production process indicators. However, due to factors such as sensor failures, environmental interference, and data transmission errors, the quality of the data is uneven, which poses a huge challenge to predictive maintenance that relies on high-quality data for analysis and decision-making. Predictive maintenance monitors the device status in real time and analyzes historical data to identify potential failures in advance, thereby reducing downtime and maintenance costs. However, data quality issues seriously affect the accuracy and reliability of the prediction model.
[0003] Current industrial big data systems have certain deficiencies in dealing with data uncertainty, especially when dealing with measurement errors and external interference in input data. When the input data contains measurement errors or is affected by external interference, traditional methods are insufficient in accurately reflecting these uncertainties, which directly affects the reliability and accuracy of the prediction model. At the same time, data loss or delay further increases the level of data uncertainty, and the existing systems lack the ability to compensate for these problems, resulting in possible deviations in the final analysis results. These situations together increase the overall data uncertainty, making it difficult for the existing systems to maintain high-precision prediction results under complex working conditions. Summary of the Invention
[0004] The purpose of the present invention is to provide an industrial big data system for industrial data analysis to solve the problem of insufficient prediction accuracy caused by data uncertainty and low-quality data in industrial data analysis. Specifically, the data analysis module integrates an uncertainty quantification engine to evaluate and quantify the noise and uncertainty in the input data, generating a detailed uncertainty report. Based on these reports, the adaptive model training mechanism can dynamically adjust the analysis model parameters or select a more appropriate algorithm to ensure high-precision prediction even in a highly uncertain data environment. In addition, the module also adopts a multi-model fusion strategy, assigning weights according to the performance of each model on historical data to further improve the prediction accuracy. Through this series of measures, the system can provide stable and reliable prediction results in a complex and changeable industrial environment, effectively supporting the status monitoring and fault warning of production equipment.
[0005] To achieve the above object, an industrial big data system for industrial data analysis is provided, including a data acquisition module. The data acquisition module obtains data streams in real time from a variety of sensors, and performs preliminary cleaning and verification through edge computing nodes to generate original data packets. The data acquisition module transfers the original data packets to the data storage module. The data storage module receives the original data packets, and organizes and stores the original data packets using a hybrid storage strategy combining a distributed file system and a relational database, and converts them into standardized data streams. It also includes:
[0006] A data analysis module. The data analysis module includes an uncertainty quantification engine. The uncertainty quantification engine receives the standardized data stream and performs a detailed uncertainty assessment on it to generate an uncertainty report. The data analysis module also includes an adaptive model training mechanism. The adaptive model training mechanism dynamically adjusts the analysis model parameters based on the uncertainty report and generates an optimized model parameter set to improve the accuracy and reliability of equipment status prediction. The data analysis module transfers the optimized model parameter set and the uncertainty report to the visualization and warning module for real-time monitoring of the production equipment status, thereby supporting the decision-making process of the staff.
[0007] As a further improvement of this technical solution, the data acquisition module performs preliminary processing on the received sensor data, including removing noise, filling in missing values, and performing format standardization operations.
[0008] As a further improvement of this technical solution, the original data packet includes a timestamp, a sensor ID, a measurement value, and a preliminary processing status, which are used to ensure the accuracy and reliability of the data in the subsequent processing stage.
[0009] As a further improvement of this technical solution, the data storage module converts unstructured sensor data into a format suitable for long-term storage and efficient query.
[0010] As a further improvement of this technical solution, the data storage module uses a relational database management system to store the sensor ID and timestamp, and establishes an indexing mechanism to speed up the query speed.
[0011] As a further improvement of this technical solution, the uncertainty quantification engine evaluates data uncertainty by calculating the standard deviation for data following a normal distribution.
[0012] As a further improvement of this technical solution, the adaptive model training mechanism realizes parameter adjustment by applying the Bayesian optimization algorithm, and uses the Bayesian optimization algorithm to find the optimal parameter set.
[0013] As a further improvement of this technical solution, the data analysis module outputs more accurate and reliable equipment status prediction results through a multi-model fusion strategy and weight assignment based on historical performance.
[0014] As a further improvement of this technical solution, the visualization and warning module receives the optimized model parameter set and uncertainty report from the data analysis module, and converts them into intuitive and easy-to-understand charts and alarm information for supporting real-time monitoring and fault warning.
[0015] As a further improvement of this technical solution, the visualization and warning module displays the uncertainty level of data through dynamically updated trend charts, bar charts and color coding, and automatically triggers an alarm and recommends countermeasures when detecting potential problems.
[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0017] 1. In the industrial big data system for industrial data analysis, through integrating the uncertainty quantification engine, the data analysis module converts the standardized data stream into a detailed uncertainty report. By conducting meticulous statistical analysis and evaluation on the input data, this engine can accurately identify the uncertainty level of each part within the data set and its possible sources. Based on these uncertainty reports, the adaptive model training mechanism can dynamically adjust the analysis model parameters or select a more suitable algorithm to ensure high-precision prediction even in a highly uncertain data environment. In addition, by adopting the multi-model fusion strategy and assigning weights according to the performance of each model on historical data, the quality of the prediction results is further optimized. This not only improves the prediction accuracy of a single model but also enhances the adaptability of the system in a complex and changeable data environment, enabling the system to provide stable and reliable predictions in a highly uncertain data environment and greatly reducing the downtime caused by equipment failures.
[0018] 2. In the industrial big data system for industrial data analysis, the data acquisition module uses edge computing nodes for preliminary cleaning and verification to ensure the quality and availability of the original data. The data storage module adopts a method combining a distributed file system and a relational database to effectively manage and store a large amount of structured and unstructured data, and converts it into a standardized data stream for subsequent analysis. The data analysis module deeply processes these standardized data streams to generate an optimized model parameter set and an uncertainty report, which are transmitted to the visualization and warning module. This module converts complex analysis results into intuitive and easy-to-understand charts and alarm information to help operators quickly understand the equipment status and take corresponding measures. This seamless data processing flow not only simplifies the operation steps but also greatly improves the response speed and decision-making support ability of the system. Description of the Drawings
[0019] Figure 1 It is a schematic diagram of the process structure of the industrial big data system for industrial data analysis of the present invention;
[0020] Figure 2 Schematic diagram of the process structure of the data storage module of the present invention;
[0021] Figure 3 Schematic diagram of the process structure of the data analysis module of the present invention;
[0022] Figure 4 Schematic diagram of the process structure of the visualization and warning module of the present invention.
[0023] The meanings of the various labels in the figure are as follows:
[0024] Among them: 100, data acquisition module; 200, data storage module; 300, data analysis module; 400, visualization and warning module. Specific implementation manners
[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0026] At the same time, some technical terms are explained here:
[0027] The distributed file system is used to store a large amount of unstructured or semi-structured sensor data. By converting the original data into an efficient storage format (such as Parquet or ORC), it reduces the storage space requirements and improves the reading speed. It has high scalability and fault tolerance, and can easily handle the growth of data volume and node failures, ensuring the reliability and persistence of data;
[0028] The relational database is used to store structured metadata that needs to be frequently queried and updated, such as sensor IDs, timestamps, etc. It supports complex query operations and transaction management, ensures data consistency and integrity, and significantly speeds up the query response time by establishing an indexing mechanism, providing efficient data access and management functions;
[0029] Parquet is a columnar storage format suitable for efficient reading and writing of large-scale data sets. It optimizes query performance by storing data column by column, especially outstanding when dealing with queries that need to access specific columns rather than entire rows of data. Parquet also supports complex nested data structures and reduces storage space requirements through compression technology, improving I / O efficiency, making big data analysis tasks more efficient;
[0030] ORC (Optimized Row Columnar) is also an efficient columnar storage format, specifically designed to optimize the reading, writing, and processing speeds of large datasets. The ORC file format can significantly reduce storage space and accelerate query operations through built-in indexing, lightweight compression algorithms, and support for complex data types. Additionally, the ORC format is highly efficient in handling complex nested data structures, making it an ideal choice for high-performance data processing and analysis.
[0031] See Figure 1 As shown, the purpose of this embodiment is to provide an industrial big data system for industrial data analysis, aiming to provide efficient and accurate data processing and analysis capabilities for industrial predictive maintenance by integrating a data acquisition module 100, a data storage module 200, a data analysis module 300, and a visualization and warning module 400, thereby achieving real-time monitoring and predictive maintenance of the status of production equipment and ultimately reducing downtime caused by equipment failures.
[0032] The data acquisition module 100 is the fundamental link for realizing real-time monitoring and predictive maintenance of the status of production equipment. This module is responsible for obtaining data streams in real time from various sensors (such as temperature, pressure, vibration, etc.) and performing preliminary cleaning and verification through edge computing nodes to ensure the quality and usability of the data. Specifically, sensors are deployed at key positions of production equipment to monitor the operating status of the equipment. Edge computing nodes play an important role in this process. They not only remove noise, fill in missing values, but also perform format standardization operations to generate raw data packets. Each raw data packet contains metadata such as a timestamp, a sensor ID, a measurement value, and a preliminary processing status.
[0033] To better understand the preliminary processing steps in the data acquisition process, we introduce a formula to describe the noise removal process in data preprocessing. The clean data obtained after the noise removal algorithm of the edge computing node is . Noise removal can be achieved through the following formula:
[0034] ;
[0035] In the formula, represents the raw measurement value at the th time point;
[0036] represents the clean data after denoising;
[0037] is a regulation coefficient (usually taking values between 0 and 1) used to control the denoising intensity;
[0038] Represents the window size, which determines the number of data points used to calculate the median;
[0039] The function represents the process of calculating the median.
[0040] This formula is based on the principle of the median filter and can effectively remove random noise in the sensor data, improving data quality. The original data packet after the above processing is transmitted to the data storage module 200, providing high-quality basic data for subsequent data analysis. In this way, the data acquisition module 100 not only ensures the accuracy of the data but also lays a solid foundation for the efficient operation of the entire system. The effective processing at this stage is crucial for the accuracy and reliability of subsequent modules because high-quality data input is a prerequisite for ensuring the reliability of the final analysis results.
[0041] See Figure 2 As shown, the data storage module 200 plays a key role in the system, responsible for receiving the original data packets from the data acquisition module 100 and effectively organizing and storing these data to support subsequent data analysis and processing. To ensure data integrity and efficient utilization, this module adopts a hybrid storage strategy combining a distributed file system and a relational database.
[0042] Specifically, for the large amount of unstructured sensor data received from the data acquisition module 100, we first convert it into a format suitable for long-term storage and efficient query (such as Parquet or ORC). This not only helps reduce the storage space requirements but also significantly improves the data reading speed. The compression ratio can be expressed by the following formula:
[0043] ;
[0044] In the formula, is the compression ratio, representing the ratio of the compressed data volume to the original data volume, reflecting the compression efficiency;
[0045] is the original uncompressed data volume;
[0046] is the data volume after format conversion and compression.
[0047] This calculation method shows that by adopting appropriate data formats and compression technologies, the storage cost can be significantly reduced and the data access efficiency can be improved.
[0048] For metadata that needs to be frequently queried and updated (such as sensor IDs, timestamps, etc.), we use a relational database management system for storage. This approach can support complex query operations and transaction management, ensuring data consistency and integrity. In addition, to further speed up query speed, an indexing mechanism is established for common query paths, thereby reducing query response time and improving the overall performance of the system.
[0049] Overall, by applying a hybrid storage strategy and converting the received raw data packets into a standardized data stream, the data storage module 200 not only realizes the effective management and storage of massive raw data, but also greatly improves the scalability and reliability of the system. This module first preprocesses the received raw data and converts it into a standard data stream in a unified format, laying a solid foundation for subsequent data analysis. This design ensures that the system can still operate efficiently in the face of large-scale data, supports quickly and accurately extracting valuable information from massive data, so as to realize real-time monitoring and predictive maintenance of the production equipment status. By optimizing data access efficiency and ensuring data consistency and integrity, the data storage module 200 provides a high-quality data foundation for data analysis, enabling the entire system to maintain efficient and stable operation in a complex and changing industrial environment, and further accurately evaluating the equipment condition and predicting potential failures.
[0050] Although the data storage module 200 effectively solves the problems of storing and managing large-scale data by adopting a hybrid strategy of a distributed file system and a relational database, there is still a core challenge when facing a complex and changing industrial environment: how to accurately process and quantify the uncertainty in the data. For example, there may be measurement errors in sensor data due to hardware failures, external interferences, or transmission errors, and existing storage solutions are difficult to directly identify and process these uncertainties. This limitation directly affects the effect of in-depth data analysis. Especially in supporting predictive maintenance, traditional methods are unable to accurately quantify uncertainties and dynamically adjust analysis models, resulting in analysis results that may not be accurate or reliable enough, thus affecting the quality of the final decision.
[0051] To address this key issue, the present invention introduces the data analysis module 300. As the main improvement point, this module focuses on enhancing the ability to handle data uncertainty. By integrating an uncertainty quantification engine and an adaptive model training mechanism, the data analysis module 300 can effectively evaluate and quantify the uncertainty in the input data, and dynamically adjust the parameters of the analysis model or select a more appropriate algorithm based on this information. This design not only makes up for the deficiencies of the prior art in dealing with data uncertainty and dynamic adjustment, but also significantly enhances the prediction ability and reliability of the system, ensuring accurate analysis results can be provided even in a highly uncertain data environment. Therefore, by optimizing the management of uncertainty and the adaptive ability of the analysis model, the present invention lays a solid foundation for achieving efficient and accurate predictive maintenance. Next, we will introduce the design and implementation of this key module in detail.
[0052] See Figure 3 As shown, the data analysis module 300 receives the standardized data stream from the data storage module 200 and conducts in-depth analysis based on this high-quality data. First, the uncertainty quantification engine processes the received standardized data stream, evaluates, and quantifies the uncertainty level therein. Suppose we have a set of standardized sensor data , each representing a measurement value at a certain time point. To measure the uncertainty in these data, methods such as standard deviation or entropy measure can be used. For example, for data following a normal distribution, its uncertainty can be expressed by the following formula:
[0053] ;
[0054] In the formula, represents the standard deviation, which is used to measure the deviation of the values in the data set from their average value;
[0055] represents the number of samples in the data set. In the formula, represents that there are measurement values in total;
[0056] represents the th measurement value, representing each observation in the data set;
[0057] represents the average value of the data set.
[0058] By calculating these statistics, the uncertainty quantification engine generates a detailed uncertainty report that describes in detail the uncertainty levels of various parts within the dataset and their possible sources. This step not only provides a key basis for subsequent steps but also helps identify which data requires special attention or further processing.
[0059] Next, the adaptive model training mechanism dynamically adjusts and optimizes the analysis model based on the uncertainty report. When it detects a significant increase in the uncertainty of a certain set of sensor data (i.e., or the value exceeds a preset threshold), the system will automatically select or retrain a model that is more suitable for the current data characteristics. This process can achieve parameter adjustment by applying the Bayesian optimization algorithm, with the goal of finding the optimal set of model parameters , satisfying:
[0060] ;
[0061] In the formula, represents the conditional probability distribution, indicating the optimal estimate of the model parameter under the condition of the given data ;
[0062] represents the maximum point of the solution function. Specifically, it is to find the parameter value that makes the largest among all possible parameters ;
[0063] represents the conditional probability distribution, indicating the probability distribution of the model parameter under the condition of the given data . This is the posterior probability distribution in Bayesian statistics.
[0064] Based on the uncertainty report provided by the uncertainty quantification engine, the adaptive model training mechanism can specifically generate an optimized set of model parameters to ensure high prediction accuracy even in a highly uncertain data environment. For example, in the face of highly uncertain data, the system may choose to use a more complex machine learning model, such as a deep neural network, or fine-tune the existing model to better capture the data characteristics.
[0065] In addition, to further improve the performance of the system, the data analysis module 300 also adopts a multi-model fusion strategy, integrating the output results of different types of prediction models. The weight allocation is based on the performance of each model on historical data, especially their adaptability to highly uncertain data. For example, if a certain model has performed better than other models on highly uncertain data in the past, it will be given a higher weight when processing similar data in the future. This strategy not only improves the accuracy of prediction but also enhances the robustness of the system, enabling it to operate stably under various complex conditions.
[0066] Finally, the entire data analysis process closely revolves around the processing of standardized data streams, uncertainty quantification, and the optimization of adaptive models. In this way, the data analysis module 300 not only effectively identifies and quantifies the uncertainties in the data but also dynamically adjusts the analysis models based on this information, ensuring that the entire system can operate efficiently and stably in the complex and changing industrial big data environment. Based on this, the optimized model parameter set and uncertainty report generated by this module are transmitted to the visualization and warning module 400, providing the latest basis for real-time monitoring and fault warning. This enables the system to achieve precise predictive maintenance, improve production efficiency and safety, and support enterprise decision-making through intuitive information display and timely operation suggestions, completing the full-process automated management from data to decision support.
[0067] The visualization and warning module 400 is a key component in the entire industrial big data system that directly faces users and provides real-time monitoring and fault warning. As Figure 4 shown, this module receives the optimized model parameter set and uncertainty report from the data analysis module 300, and converts this information into intuitive and easy-to-understand charts, dashboards, and alarm information to help decision-makers quickly understand the status of production equipment and take corresponding measures.
[0068] First of all, based on the received optimized model parameter set, the visualization and warning module 400 can generate highly customized prediction result displays. For example, it shows the change trend of the equipment operation status through a dynamically updated trend chart, compares the changes in performance indicators in different time periods using bar charts or pie charts, or uses a heat map to display the spatial distribution of sensor data. These visualization means not only make complex data easy to understand but also provide a basis for operators to adjust production strategies in a timely manner.
[0069] Secondly, the uncertainty report also plays an important role in the visualization and warning module 400. It is used to enhance the reliability assessment of the prediction results. By integrating the uncertainty level into the visualization interface, the system can more transparently display the confidence interval of the prediction results. For example, adding a shaded area next to the trend graph to represent the possible fluctuation range of the predicted value, or using color coding on the dashboard to reflect the uncertainty level of the current data (green represents low uncertainty, and red represents high uncertainty). This enables the operators to make more cautious decisions based on the uncertainty level.
[0070] In addition, when potential problems are detected, the visualization and warning module 400 will automatically trigger a pre-set operation process. For example, once a key indicator exceeds the normal range or the predicted failure probability exceeds the threshold, the system will immediately issue an alarm and recommend specific countermeasures, such as dispatching a maintenance team for inspection or suggesting that the operator perform specific maintenance actions. This immediate feedback mechanism greatly reduces the downtime and maintenance costs, and improves the production efficiency and equipment reliability. By integrating and optimizing the model parameter set and uncertainty report, the visualization and warning module 400 not only enhances the user interactivity and decision-making support ability of the system, but also improves the overall system robustness and adaptability while ensuring efficient and accurate prediction. This design enables the system to provide stable and reliable services even in the face of complex and highly uncertain industrial environments, helping enterprises achieve intelligent transformation and continuous improvement.
[0071] In summary, by introducing an uncertainty quantification engine and optimizing the adaptive model training mechanism, the data analysis module 300 significantly improves the ability to identify and quantify uncertainties in industrial big data. By transmitting the updated optimized model parameter set and uncertainty report to the visualization and warning module 400, the present invention enables the operators to make decisions based on more detailed and accurate data, directly promoting the improvement of production efficiency and the effective control of maintenance costs. This improvement not only enhances the response speed and stability of the system, but also greatly improves the decision-making accuracy.
[0072] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. An industrial big data system for industrial data analysis, comprising a data acquisition module (100), wherein the data acquisition module (100) acquires data streams from a variety of sensors in real time, and performs preliminary cleaning and verification through edge computing nodes to generate original data packets, wherein the data acquisition module (100) transmits the original data packets to a data storage module (200), wherein the data storage module (200) receives the original data packets and adopts a hybrid storage strategy of a distributed file system combined with a relational database to organize and store the original data packets and convert them into standardized data streams, characterized in that: Also includes: A data analysis module (300), the data analysis module (300) comprising an uncertainty quantification engine, the uncertainty quantification engine receiving the standardized data stream and performing a detailed uncertainty assessment on the standardized data stream to generate an uncertainty report; The data analysis module (300) also includes an adaptive model training mechanism, which dynamically adjusts the analysis model parameters based on the uncertainty report and generates an optimized model parameter set to improve the accuracy and reliability of equipment status prediction. The data analysis module (300) transmits the optimized model parameter set and the uncertainty report to the visualization and early warning module (400) for real-time monitoring of the production equipment status.
2. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The data acquisition module (100) performs preliminary processing on the received sensor data, including removing noise, filling missing values and performing format standardization operations, and performs noise removal through the following formula: ； In the formula, Indicates The original measurement value at each time point; represents the clean data after denoising; It is an adjustment coefficient (usually between 0 and 1) used to control the denoising strength; Represents the window size, which determines the number of data points used to calculate the median; The function represents the process of calculating the median.
3. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The original data packet includes a timestamp, a sensor ID, a measurement value, and a preliminary processing status, which are used to ensure the accuracy and reliability of the data in the subsequent processing stage.
4. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The data storage module (200) converts unstructured sensor data into a format suitable for long-term storage and efficient query.
5. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The data storage module (200) uses a relational database management system to store sensor IDs and timestamps, and establishes an index mechanism to speed up querying.
6. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The uncertainty quantification engine evaluates data uncertainty by calculating the standard deviation of normally distributed data. The standard deviation formula is as follows: ； In the formula, It represents standard deviation, which is a measure of how far the values in a data set deviate from their mean; Represents the number of samples in the data set. In the formula, Indicates a total of measurements; Indicates the Measurements, which represent each observation in the data set; Represents the mean of the data set.
7. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The adaptive model training mechanism realizes parameter adjustment by applying the Bayesian optimization algorithm, and uses the Bayesian optimization algorithm to find the optimal parameter set. The conditional distribution probability formula is as follows: ; In the formula, Represents the conditional probability distribution, which means that given the data Under the condition of The best estimate of Indicates the maximum point of the solution function; Represents the conditional probability distribution, which means that given the data Under the condition of The probability distribution of .
8. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The data analysis module (300) outputs a more accurate and reliable equipment status prediction result through a multi-model fusion strategy and weight allocation based on historical performance.
9. The industrial big data system for industrial data analysis according to claim 1, characterized in that: The visualization and warning module (400) receives the optimization model parameter set and uncertainty report from the data analysis module (300), and converts them into intuitive and easy-to-understand charts and alarm information to support real-time monitoring and fault warning.
10. The industrial big data system for industrial data analysis according to claim 9, characterized in that: The visualization and early warning module (400) displays the uncertainty level of data through dynamically updated trend graphs, bar graphs and color coding, automatically triggers an alarm and recommends countermeasures when a potential problem is detected.