Artificial intelligence monitoring and diagnosing device for gas holder and use method of artificial intelligence monitoring and diagnosing device

Abstract

The invention relates to the field of control and adjustment, in particular to an artificial intelligence monitoring and diagnosing device for a gas cabinet and a using method of the artificial intelligence monitoring and diagnosing device. The artificial intelligence monitoring and diagnosis device for the gas cabinet comprises a data acquisition module (1), and is characterized by further comprising a data processing and analysis module (2), a fault early warning and diagnosis module (3) and an application display and management module (4), the data acquisition module (1), the data processing and analysis module (2), the fault early warning and diagnosis module (3) and the application display and management module (4) are sequentially connected through signal lines. The use method of the artificial intelligence monitoring and diagnosis device for the gas holder is characterized by comprising the following steps of S1, data acquisition; s2, data processing and analysis; s3, fault early warning and diagnosis; and S4, application display and management. The method is timely in prediction and high in diagnosis precision.

Description

Technical Field

[0001] The present invention relates to the field of control and regulation, and in particular to an artificial intelligence monitoring and diagnostic device for a gas cabinet and a method of using the same. Background Art

[0002] Gasholders, a crucial component of steel companies' gas pipeline networks, are widely used to recover and store flammable, explosive, and toxic gases. They regulate imbalances in gas production and usage across the plant, stabilize pipeline pressure, and reduce gas emissions. Given their crucial role in the gas pipeline network and their potential as a major hazard source, effective monitoring of their operating status is essential for promptly identifying and eliminating potential safety hazards and ensuring their safe operation.

[0003] Currently, steel companies primarily ensure safe gasholder operation by monitoring operating parameters such as the gasholder position, piston lift speed, and piston drift and tilt, and then setting alarms and interlock controls in the PLC. This approach ensures safe gasholder operation, but suffers from three drawbacks: 1. A passive mode, where the safety monitoring system only issues an alarm when the gasholder's operating parameters exceed the safe range, failing to proactively analyze and issue early warnings; 2. A low level of intelligence, where the control system is unable to intelligently analyze instrument data, preventing the data from the instrument from effectively guiding gasholder operation and effectively enabling the gasholder to fully function within the gas network; 3. A low level of integration, where the lack of comprehensive, highly integrated product solutions means many gasholder inspection items are incomplete, making it difficult to fully and accurately understand the actual status of the gasholder. Summary of the Invention

[0004] In order to overcome the defects of the prior art and provide a control system with timely prediction and high diagnostic accuracy, the present invention discloses an artificial intelligence monitoring and diagnostic device for a gas cabinet and a method of using the same.

[0005] The present invention achieves the purpose of the invention through the following technical solutions: An artificial intelligence monitoring and diagnostic device for a gas cabinet includes a data acquisition module, which is characterized by further including a data processing and analysis module, a fault warning and diagnosis module, and an application display and management module. The data acquisition module, data processing and analysis module, fault warning and diagnosis module and application display and management module are connected in sequence through signal lines; The data acquisition module includes a sensor, an edge computing unit and a data transmission module. The sensors are respectively arranged on the gas cabinet. The signal output end of the sensor is connected to the signal input end of the edge computing unit through a signal line. The signal output end of the edge computing unit is connected to the signal input end of the data transmission module through a signal line. The signal output end of the data transmission module is connected through a signal line; The data acquisition module integrates front-end sensors, industrial PLCs (such as gas tank PLCs) serving as edge computing units, and the data transmission module (including wireless acquisition gateways and data gateways) for data access. The data processing and analysis module uses message queues (such as Kafka) and diversified databases (such as MySQL, Redis, and MongoDB) for efficient data processing and storage. The fault warning and diagnosis module integrates expert fault knowledge bases and artificial intelligence prediction models (such as warning rules, fault diagnosis algorithms (including machine learning, deep neural network, and other artificial intelligence models)) to implement intelligent analysis and diagnosis. The application display and management module provides comprehensive visual monitoring and management functions such as equipment status monitoring, fault warning, and user management through a Web graphical interface.

[0006] The artificial intelligence monitoring and diagnostic device for a gas cabinet is characterized in that the data transmission module uses a data gateway.

[0007] The method for using the artificial intelligence monitoring and diagnostic device for a gas cabinet is characterized by being carried out in sequence according to the following steps: S1 data acquisition: the data acquisition module of the sensor real-time measurement of the raw data of the gas tank, calculated by the edge computing unit through the data transmission module input data processing and analysis module; S2. Data Processing and Analysis: The data processing and analysis module first preprocesses the raw data to remove noise and outliers, ensuring data accuracy and reliability. It then uses database systems such as MySQL, Redis, and MongoDB to efficiently store the cleaned data. Finally, it uses intelligent learning algorithms such as multivariate linear regression and XGBoost to train a health score prediction model. S3. Fault Warning and Diagnosis: The fault warning and diagnosis module draws on years of experience gained by process experts in gasholder construction and operation. It constructs an expert fault database that includes fault types, causes, and solutions. Combined with a health score prediction model, the module analyzes real-time gasholder operating data, combines expert-provided parameter thresholds and fault characteristics, and automatically identifies potential problems in equipment operation. It then provides corresponding fault analysis and solution recommendations. S4. Application Display and Management: The application display and management module uses a graphical interface to display the gas tank's operating status, fault warning information, and device reports on the web, allowing users to intuitively understand the system status. It also implements user permission allocation, role management, and device management functions in the cloud.

[0008] The method for using the artificial intelligence monitoring and diagnostic device for a gas cabinet is characterized by: In step S1, the sensor measures various raw data of the gas tank in real time, including the position, tilt, and drift of the gas tank. An industrial PLC is set as the edge computing unit to control, preliminarily process, and filter the collected raw data. In step S2, at least one database system selected from MySQL, Redis, and MongoDB is used to efficiently store the cleaned data, wherein MySQL is used to store structured data, Redis is used to cache frequently accessed data, and MongoDB is used to store unstructured or semi-structured data; a health score prediction model is obtained by training a multivariate linear regression model or an XGBoost model intelligent learning algorithm; In step S4, the application display and management module displays information including device overview, trend analysis, device warning, warning device ranking list and device monitoring on the web side through a graphical interface.

[0009] The present invention discloses a monitoring and diagnosis system for gas cabinet application scenarios, which is trained using artificial intelligence algorithms. The system includes the collection, uploading, analysis, and diagnosis of gas cabinet operating parameters, and ultimately provides a visual score for the current operating conditions of the gas cabinet. It can also analyze various fault conditions of the gas cabinet and provide treatment suggestions.

[0010] The present invention realizes comprehensive monitoring and intelligent management of the operating status of the gas tank through the collaborative work of multiple levels such as data collection, processing and analysis, fault warning and diagnosis, application display and management.

[0011] Through multiple experiments and field tests, this system demonstrated excellent overall accuracy. The scoring model, driven by data, fully leverages the integration of historical and real-time data, ensuring highly accurate predictions of device health status. In the training set, the mean squared error (MSE) of the multivariate linear regression model was 1.29, the MSE of the XGBoost model was 1.28, the MSE of the BP neural network was 1.327, and the MSE of the LSTM model was 2.66, all maintaining a low MSE. On the test set, the MSE of the multivariate linear regression model was 1.51, the MSE of the XGBoost model was 1.88, the MSE of the BP neural network was 2.67, and the MSE of the LSTM model significantly increased to 12.84. This demonstrates that the multivariate linear regression model demonstrated the strongest generalization ability, the lowest prediction error, and good stability and robustness on the test set. Overall, the multivariate linear regression model achieved balanced performance on both the training and test sets, achieving the best prediction results. Therefore, the multivariate linear regression model was selected as the primary algorithm training model. In the fault identification model, a threshold-based judgment mechanism ensures rapid response and accurate diagnosis of known fault types. By combining parameter ranges and fault types from an expert experience database, the threshold model can identify common faults with high accuracy and provide corresponding treatment recommendations.

[0012] Compared with the existing gas cabinet monitoring mode, the present invention has the following beneficial effects: 1. Active prediction: The artificial intelligence algorithm model of this invention can proactively predict the future health changes of gas tanks and provide early warning of potential equipment risks.

[0013] 2. Intelligent Guidance: This invention intelligently provides fault cause analysis and treatment suggestions for equipment failures in gas cabinets.

[0014] 3. Real-time Monitoring: The web-based graphical monitoring interface of this invention enables remote, wireless, and real-time monitoring of gas tanks, a major hazard source. 4. High-Precision Diagnosis: By combining big data analysis and sensor data, our system can reduce false alarms and improve diagnostic accuracy.

[0015] 4. The AI ​​monitoring and diagnostic system of this invention uses AI algorithms for training, and implements active prediction, intelligent guidance, real-time monitoring, and high-precision diagnosis in the field of gas tank monitoring, contributing to smart manufacturing in metallurgical enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic flow diagram of the present invention. DETAILED DESCRIPTION

[0017] The present invention is further illustrated below by means of specific examples. Example

[0018] An artificial intelligence monitoring and diagnostic device for a gas cabinet, comprising a data acquisition module 1, a data processing and analysis module 2, a fault warning and diagnosis module 3, and an application display and management module 4. Figure 1 As shown, the specific structure is: The data acquisition module 1, the data processing and analysis module 2, the fault warning and diagnosis module 3 and the application display and management module 4 are connected in sequence through signal lines; The data acquisition module 1 includes a sensor, an edge computing unit, and a data transmission module. The sensors are respectively arranged on the gas cabinet. The data transmission module uses a data gateway. The signal output end of the sensor is connected to the signal input end of the edge computing unit through a signal line. The signal output end of the edge computing unit is connected to the signal input end of the data transmission module through a signal line. The signal output end of the data transmission module is connected through a signal line; The data acquisition module 1 integrates front-end sensors, an industrial PLC (such as a gas tank PLC) serving as the edge computing unit, and the data transmission module (including a wireless acquisition gateway and a data gateway) for data access. The data processing and analysis module 2 uses message queues (such as Kafka) and diversified databases (such as MySQL, Redis, and MongoDB) for efficient data processing and storage. The fault warning and diagnosis module 3 integrates expert fault knowledge bases with artificial intelligence prediction models (such as warning rules and fault diagnosis algorithms (including machine learning, deep neural network, and other artificial intelligence models)) to implement intelligent analysis and diagnosis. The application display and management module 4 provides comprehensive visual monitoring and management functions such as equipment status monitoring, fault warning, and user management through a Web graphical interface.

[0019] When using this embodiment, the following steps are implemented in sequence: S1 data acquisition: the data acquisition module 1 of the sensor real-time measurement of the raw data of the gas tank, calculated by the edge computing unit through the data transmission module input data processing and analysis module 2; The sensor measures various raw data of the gas tank in real time, including the position, tilt, and drift of the gas tank. An industrial PLC is set as the edge computing unit to control, preliminarily process, and filter the collected raw data. S2. Data Processing and Analysis: Data Processing and Analysis Module 2 first preprocesses the raw data to remove noise and outliers, ensuring data accuracy and reliability. It then uses database systems such as MySQL, Redis, and MongoDB to efficiently store the cleaned data. MySQL is used to store structured data, Redis is used to cache frequently accessed data, and MongoDB is used to store unstructured or semi-structured data. Finally, a health score prediction model is trained using intelligent learning algorithms such as multivariate linear regression and XGBoost. S3. Fault Warning and Diagnosis: Module 3 leverages the years of experience of process experts in gasholder construction and operation to construct an expert fault database that includes fault types, causes, and solutions. Combined with a health score prediction model, this module analyzes real-time gasholder operating data, combined with expert-provided parameter thresholds and fault characteristics, to automatically identify and warn of potential equipment problems and provide appropriate fault analysis and resolution recommendations. S4. Application Display and Management: Application Display and Management Module 4 uses a graphical interface to display the operating status of gas cabinets, fault warning information, and device reports on the web. This information includes device overviews, trend analysis, device warnings, warning device rankings, and device monitoring, allowing users to intuitively understand system status. It also implements user permission allocation, role management, and device management in the cloud.

Claims

1. An artificial intelligence monitoring and diagnostic device for a gas cabinet, comprising a data acquisition module (1), characterized in that: It also includes a data processing and analysis module (2), a fault warning and diagnosis module (3) and an application display and management module (4). The data acquisition module (1), the data processing and analysis module (2), the fault warning and diagnosis module (3) and the application display and management module (4) are connected in sequence through signal lines; The data acquisition module (1) includes a sensor, an edge computing unit and a data transmission module, wherein the sensors are respectively arranged on the gas cabinet, the signal output end of the sensor is connected to the signal input end of the edge computing unit via a signal line, the signal output end of the edge computing unit is connected to the signal input end of the data transmission module via a signal line, and the signal output end of the data transmission module is connected via a signal line; The data acquisition module (1) integrates the front-end sensor, the edge computing unit and the data transmission module for realizing data access; the data processing and analysis module (2) uses a message queue and a diversified database to process and store data efficiently; the fault warning and diagnosis module (3) integrates the expert fault knowledge base and the artificial intelligence prediction model to implement intelligent analysis and diagnosis; The application display and management module (4) provides comprehensive visual monitoring and management functions of equipment status monitoring, fault warning and user management through a Web graphical interface.

2. The artificial intelligence monitoring and diagnostic device for a gas cabinet according to claim 1, characterized in that: The data transmission module uses a data gateway.

3. The method for using the artificial intelligence monitoring and diagnostic device for a gas cabinet according to claim 1 or 2, characterized in that: Follow the steps below: S1. Data acquisition: The data acquisition module (1) measures the raw data of the gas tank in real time, and the data is input into the data processing and analysis module (2) after being calculated by the edge computing unit. S2. Data processing and analysis: The data processing and analysis module (2) first pre-processes the raw data to remove noise and outliers to ensure the accuracy and reliability of the data. The cleaned data is then stored efficiently in the database system. Finally, a health score prediction model is obtained through model intelligent learning algorithm training. S3. Fault Warning and Diagnosis: The Fault Warning and Diagnosis module (3) builds an expert fault database that includes fault types, causes, and treatment methods. It combines the health score prediction model and analyzes the real-time data of gas tank operation. It combines the parameter thresholds and fault characteristics provided by experts to automatically identify potential problems that may occur in the operation of the warning equipment and provide corresponding fault analysis and treatment suggestions. S4. Application Display and Management: The application display and management module (4) displays the operating status, fault warning information and equipment reports of the gas cabinet on the web through a graphical interface, allowing users to intuitively understand the system status; and realizes user permission allocation, role management and equipment management functions in the cloud.

4. The method for using the artificial intelligence monitoring and diagnostic device for a gas cabinet according to claim 3, wherein: In step S1, the sensor measures various raw data of the gas tank in real time, including the position, tilt, and drift of the gas tank. An industrial PLC is set as the edge computing unit to control, preliminarily process, and filter the collected raw data. In step S2, at least one database system selected from MySQL, Redis, and MongoDB is used to efficiently store the cleaned data, wherein MySQL is used to store structured data, Redis is used to cache frequently accessed data, and MongoDB is used to store unstructured or semi-structured data; a health score prediction model is obtained by training a multivariate linear regression model or an XGBoost model intelligent learning algorithm; In step S4, the application display and management module (4) displays information including device overview, trend analysis, device warning, warning device ranking list and device monitoring on the web side through a graphical interface.

Images