Industrial equipment intelligent monitoring method and system based on multi-source data fusion

Abstract

The invention provides an industrial equipment intelligent monitoring method and system based on multi-source data fusion. The method comprises the following steps: acquiring multi-source heterogeneous data of an industrial site; performing real-time cleaning and anomaly detection processing on the multi-source heterogeneous data to obtain standardized processing data; based on a role access control model, in combination with the field attributes of the standardized processing data, generating a data access strategy for the user role; screening a target data field according to the data access strategy, analyzing the target data field through a visual adaptation algorithm, matching the analyzed target data field with a preset target chart template, and generating adaptive visual data; pushing the self-adaptive visual data to a Web display terminal through a remote transmission mechanism based on an edge cloud collaborative architecture; according to the invention, the problems of single function, rigid authority management, low visualization efficiency and delayed remote access in the traditional industrial monitoring process can be solved, so that the flexibility, real-time performance and safety of industrial equipment monitoring are improved.

Description

Technical Field

[0001] This invention relates to the field of industrial Internet of Things and data visualization technology, specifically to an intelligent monitoring method and system for industrial equipment based on multi-source data fusion. Background Technology

[0002] Currently, with the rapid development of industrial IoT and data visualization technology, intelligent monitoring of industrial equipment has become a core support for ensuring efficient and safe operation of industrial production, and is widely used in industrial automation, energy management and other scenarios.

[0003] Industrial sites require monitoring diverse parameters, covering multiple dimensions such as environment, safety, and equipment operation. This places stringent demands on monitoring systems, including multi-source data integration, dynamic permission adaptation, accurate visualization, and remote real-time response. However, existing industrial monitoring systems still suffer from numerous technical bottlenecks, failing to meet practical application needs: Existing monitoring systems generally suffer from functional limitations, mostly supporting only single-type parameter monitoring and lacking the ability to uniformly integrate and process multi-source heterogeneous data such as environmental, safety, and equipment operation parameters, resulting in incomplete monitoring dimensions and insufficient data value extraction; rigid permission management mechanisms and fixed user interfaces and data display content fail to dynamically adjust data visibility and operation permissions according to the actual needs of different user roles such as administrators, maintenance personnel, and visitors, posing data security risks and hindering adaptation to diverse usage scenarios; and low visualization efficiency, with chart generation relying heavily on... Relying on preset templates, existing industrial monitoring systems cannot automatically parse the attributes and types of data fields and match them with appropriate display formats. Furthermore, the fixed layout of multiple charts can easily lead to visual confusion due to data overload, affecting information delivery efficiency. Remote access suffers from significant latency, and the data transmission and rendering mechanisms have not been specifically optimized. High-frequency data access relies excessively on the central server, resulting in excessive server load and causing issues such as large-screen display lag and delayed data updates, failing to meet the core requirements of real-time monitoring. The system lacks scalability; adding new monitoring items requires redeveloping data acquisition, parsing interfaces, and display modules, hindering flexible configuration for rapid response to changes in business needs and increasing monitoring upgrade and maintenance costs. These problems, combined, result in insufficient flexibility, accuracy, real-time performance, and security in existing industrial monitoring systems, severely limiting their application effectiveness in complex industrial scenarios and becoming a core technical challenge that urgently needs to be addressed in the field of intelligent industrial equipment monitoring. Summary of the Invention

[0004] To address the technical problems of insufficient multi-source heterogeneous data fusion capabilities, rigid access control, poor visualization adaptability, and remote access latency in traditional industrial monitoring systems, this invention proposes an intelligent monitoring method for industrial equipment based on multi-source data fusion, comprising:

[0005] Acquire multi-source heterogeneous data from industrial sites;

[0006] The multi-source heterogeneous data is cleaned and anomaly detected in real time to obtain standardized processed data;

[0007] Based on the preset role access control model and combined with the field attributes of the standardized processed data, a data access strategy is generated for the pre-acquired user roles.

[0008] According to the data access strategy, target data fields are selected from the standardized processed data, and the target data fields are parsed through a visualization adaptation algorithm. The parsed target data fields are then matched with a preset target chart template to generate adaptive visualization data.

[0009] Based on the edge cloud collaborative architecture, the adaptive visualization data is pushed to the Web display terminal through a remote transmission mechanism;

[0010] The visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection.

[0011] Optionally, the multi-source heterogeneous data includes one or more of the following: environmental parameter data, safety parameter data, and equipment operating parameter data;

[0012] The environmental parameter data includes: temperature and humidity data and water immersion status data;

[0013] The safety parameter data includes oxygen concentration data;

[0014] The equipment operating parameter data includes: oxygen generator current data and oxygen generator vibration data;

[0015] The multi-source heterogeneous data is collected via MQTT / HTTP protocol and supports multiple network transmission modes including TCP / IP, 4G, 5G, and WiFi.

[0016] Optionally, the step of performing real-time cleaning and anomaly detection on the multi-source heterogeneous data to obtain standardized processed data includes:

[0017] The multi-source heterogeneous data is input into a preset streaming computing framework to filter out noise information in the data and obtain preliminary cleaned data.

[0018] The preliminary cleaned data is processed for cross-protocol compatibility based on the OPC-UA protocol to obtain preliminary cleaned data after unified modeling.

[0019] The preliminary cleaning data after unified modeling is input into a preset machine learning model to predict the fault status of equipment in the industrial site and obtain abnormal data.

[0020] The normal data in the preliminary cleaned data after unified modeling is standardized and transformed, and combined with the identification information of the abnormal data, standardized processed data containing normal standardized data and abnormal labeled data is generated.

[0021] The streaming computing framework is the Apache Flink framework, and the machine learning model is the LSTM model.

[0022] Optionally, the step of generating a data access strategy for pre-acquired user roles based on a preset role access control model and in conjunction with the field attributes of the standardized processed data includes:

[0023] User role hierarchy is defined based on a preset role-based access control model; the user role hierarchy includes administrator role, maintenance personnel role, and visitor role;

[0024] Configure the corresponding data visibility scope and operation permissions for each user role level, and generate a role permission configuration table;

[0025] Based on the pre-obtained currently logged-in user role, query the role permission configuration table of the user role, and combine it with the field attributes of the standardized processed data to determine the data visibility range and operation permissions corresponding to the currently logged-in user role;

[0026] A data access policy is generated based on the data visibility range and operation permissions.

[0027] Optionally, the step of parsing the target data field using a visualization adaptation algorithm, matching the parsed target data field with a preset target chart template, and generating adaptive visualization data includes:

[0028] The automatic field recognition function in the visualization adaptation algorithm is used to parse the name and data type of the target data field and establish a mapping relationship between the field and the data type.

[0029] Based on preset field chart type mapping rules, the fields and data type mapping relationships are matched with the corresponding target chart templates;

[0030] The force-guided layout optimization function in the visualization adaptation algorithm is used to dynamically arrange the matched target chart template.

[0031] The contents of the corresponding target data fields in the standardized data are filled into the corresponding target chart template to generate adaptive visualization data;

[0032] The target chart template includes one or more of the following: bar chart template, heat map template, line chart template, and dashboard template.

[0033] Optionally, the step of pushing the adaptive visualization data to the web display terminal via a remote transmission mechanism based on the edge cloud collaborative architecture includes:

[0034] Edge servers are deployed in the local data center at the industrial site to cache and store standardized processing data that is accessed frequently, generating edge cache data.

[0035] Determine whether the original data corresponding to the adaptive visualization data exists in the edge cache data:

[0036] If it exists, obtain the corresponding edge cache data from the edge server, and generate push data by combining the chart configuration information of the adaptive visualization data;

[0037] If it does not exist, retrieve the corresponding raw data from the cloud database and generate push data by combining the chart configuration information of the adaptive visualization data.

[0038] The push data is transmitted to the web display terminal via a WebSocket long connection in the remote transmission mechanism.

[0039] Optionally, the cloud database adopts a hybrid architecture of time-series database and relational database;

[0040] The time-series database is InfluxDB;

[0041] The relational database is MySQL;

[0042] The web display terminal renders the pushed data using a WebGL-based 3D large-screen rendering engine.

[0043] Based on the same inventive concept, this invention also provides an intelligent monitoring system for industrial equipment based on multi-source data fusion, comprising:

[0044] The data acquisition module is used to acquire multi-source heterogeneous data from industrial sites;

[0045] The data processing module is used to perform real-time cleaning and anomaly detection on the multi-source heterogeneous data to obtain standardized processed data.

[0046] The access control module is used to generate data access policies for pre-acquired user roles based on a preset role access control model and the field attributes of the standardized processed data.

[0047] The visualization adaptation module is used to filter target data fields from the standardized processed data according to the data access strategy, and parse the target data fields through the visualization adaptation algorithm, and match the parsed target data fields with the preset target chart template to generate adaptive visualization data.

[0048] The display push module is used to push the adaptive visualization data to the Web display terminal through a remote transmission mechanism based on the edge cloud collaborative architecture;

[0049] The visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection.

[0050] Optionally, the multi-source heterogeneous data includes one or more of the following: environmental parameter data, safety parameter data, and equipment operating parameter data;

[0051] The environmental parameter data includes: temperature and humidity data and water immersion status data;

[0052] The safety parameter data includes oxygen concentration data;

[0053] The equipment operating parameter data includes: oxygen generator current data and oxygen generator vibration data;

[0054] The multi-source heterogeneous data is collected via MQTT / HTTP protocol and supports multiple network transmission modes including TCP / IP, 4G, 5G, and WiFi.

[0055] Optionally, the data processing module includes:

[0056] The data cleaning submodule is used to input the multi-source heterogeneous data into a preset streaming computing framework, filter out noise information in the data, and obtain preliminary cleaned data.

[0057] The protocol compatibility submodule is used to perform cross-protocol compatibility processing on the preliminary cleaned data based on the OPC-UA protocol to obtain preliminary cleaned data after unified modeling.

[0058] The anomaly detection submodule is used to input the preliminary cleaning data after unified modeling into a preset machine learning model to predict the fault status of equipment in the industrial site and obtain anomaly data.

[0059] The standardization submodule is used to standardize the normal data in the preliminary cleaned data after unified modeling, and combine the identification information of the abnormal data to generate standardized processed data containing normal standardized data and abnormal labeled data.

[0060] The streaming computing framework is the Apache Flink framework, and the machine learning model is the LSTM model.

[0061] Optionally, the access control module includes:

[0062] The role definition submodule is used to define user role levels based on a preset role access control model; the user role levels include administrator roles, operation and maintenance personnel roles, and visitor roles;

[0063] The permission configuration submodule is used to configure the corresponding data visibility scope and operation permissions for each user role level and generate a role permission configuration table.

[0064] The role acquisition submodule is used to query the role permission configuration table of the currently logged-in user role based on the pre-acquired currently logged-in user role, and determine the data visibility range and operation permissions corresponding to the currently logged-in user role by combining the field attributes of the standardized processing data.

[0065] The policy generation submodule is used to generate data access policies based on the data visibility scope and operation permissions.

[0066] Optionally, the visualization adaptation module includes:

[0067] The field parsing submodule is used to parse the name and data type of the target data field through the automatic field recognition function in the visualization adaptation algorithm, and establish a mapping relationship between the field and the data type;

[0068] The chart matching submodule is used to match the field and data type mapping relationship with the corresponding target chart template based on the preset field chart type mapping rules;

[0069] The layout optimization submodule is used to dynamically arrange the matched target chart template through the force-oriented layout optimization function in the visualization adaptation algorithm.

[0070] The data population submodule is used to populate the content of the corresponding target data field in the standardized data into the corresponding target chart template to generate adaptive visualization data;

[0071] The target chart template includes one or more of the following: bar chart template, heat map template, line chart template, and dashboard template.

[0072] Optionally, the display push module includes:

[0073] The edge caching submodule is used to deploy edge servers in the local data center of the industrial site to cache and store standardized processing data that is accessed frequently, and generate edge cache data.

[0074] The data routing submodule is used to determine whether the original data corresponding to the adaptive visualization data exists in the edge cache data: if it exists, the corresponding edge cache data is obtained from the edge server, and push data is generated by combining the chart configuration information of the adaptive visualization data; if it does not exist, the corresponding original data is obtained from the cloud database, and push data is generated by combining the chart configuration information of the adaptive visualization data.

[0075] The transmission and display submodule is used to transmit the push data to the Web display terminal through the WebSocket long connection in the remote transmission mechanism.

[0076] Optionally, the cloud database adopts a hybrid architecture of time-series database and relational database;

[0077] The time-series database is InfluxDB;

[0078] The relational database is MySQL;

[0079] The web display terminal renders the pushed data using a WebGL-based 3D large-screen rendering engine.

[0080] In another aspect, the present invention also provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus;

[0081] The memory is used to store one or more programs;

[0082] When the one or more programs are executed by the at least one processor, an intelligent monitoring method for industrial equipment based on multi-source data fusion, as described above, is implemented.

[0083] In another aspect, the present invention also provides a computer device readable storage medium having an executable program stored thereon, wherein when the executable program is executed, it implements the aforementioned method for intelligent monitoring of industrial equipment based on multi-source data fusion.

[0084] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0085] This invention provides an intelligent monitoring method and system for industrial equipment based on multi-source data fusion, comprising: acquiring multi-source heterogeneous data from an industrial site; performing real-time cleaning and anomaly detection processing on the multi-source heterogeneous data to obtain standardized processed data; generating data access policies for pre-acquired user roles based on a preset role access control model and the field attributes of the standardized processed data; filtering target data fields from the standardized processed data according to the data access policies, parsing the target data fields using a visualization adaptation algorithm, matching the parsed target data fields with a preset target chart template to generate adaptive visualization data; and pushing the adaptive visualization data to a Web display terminal through a remote transmission mechanism based on an edge cloud collaborative architecture; wherein, the visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection collaboration; this invention acquires multi-source heterogeneous data from an industrial site. Multi-source heterogeneous data is cleaned and anomaly detected in real time. Combined with a preset role-based access control model and standardized data field attributes, targeted data access strategies are generated, which facilitates precise matching of user roles with data visibility and operation permissions. By parsing the filtered target data fields and matching them with preset chart templates through a visualization adaptation algorithm, the adaptability of data display and information transmission efficiency can be improved. Based on an edge cloud collaborative architecture and a remote transmission mechanism that uses edge node caching and WebSocket long connection collaboration, adaptive visualization data is pushed to the Web display terminal, which can effectively optimize the pain points of remote access latency and large-screen display lag in traditional systems, and ensure the real-time performance of data transmission and rendering. Therefore, the method of this invention does not rely on fixed monitoring parameter types and display modes, and can flexibly adapt to the diverse monitoring needs of industrial sites, which helps to solve the problem of insufficient scalability of traditional systems, thereby improving the flexibility and practicality of intelligent monitoring of industrial equipment. Attached Figure Description

[0086] Figure 1 A flowchart illustrating an intelligent monitoring method for industrial equipment based on multi-source data fusion provided by this invention;

[0087] Figure 2 A schematic diagram illustrating the structural composition of an intelligent monitoring system for industrial equipment based on multi-source data fusion, provided by this invention;

[0088] Figure 3 This is a schematic diagram of the structure of an electronic device provided by the present invention. Detailed Implementation

[0089] This invention proposes a method, system, device, and medium for intelligent monitoring of industrial equipment based on multi-source data fusion. The specific embodiments of this invention will be further described in detail below with reference to the accompanying drawings.

[0090] Example 1:

[0091] This invention provides an intelligent monitoring method for industrial equipment based on multi-source data fusion, the flowchart of which is shown below. Figure 1 As shown, it includes:

[0092] Step 1: Acquire multi-source heterogeneous data from the industrial site;

[0093] Step 2: Perform real-time cleaning and anomaly detection on the multi-source heterogeneous data to obtain standardized processed data;

[0094] Step 3: Based on the preset role access control model and combined with the field attributes of the standardized processed data, generate data access strategies for the pre-acquired user roles;

[0095] Step 4: Filter target data fields from the standardized processed data according to the data access strategy, and parse the target data fields through a visualization adaptation algorithm. Match the parsed target data fields with a preset target chart template to generate adaptive visualization data.

[0096] Step 5: Based on the edge cloud collaborative architecture, push the adaptive visualization data to the Web display terminal through a remote transmission mechanism;

[0097] The visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection.

[0098] Generally, industrial monitoring suffers from limited functionality, supporting only the monitoring of a single type of parameter. It is difficult to integrate multi-dimensional heterogeneous data such as environment, safety, and equipment operation. Furthermore, the data acquisition protocol is singular and the transmission mode is limited, failing to meet the diverse monitoring scenarios and data transmission needs of industrial sites. To solve this problem, this invention introduces multi-source heterogeneous data to achieve comprehensive coverage and flexible and efficient acquisition of multi-dimensional data.

[0099] For example, the multi-source heterogeneous data in step 1 above may include one or more of the following: environmental parameter data, safety parameter data, and equipment operating parameter data;

[0100] The environmental parameter data may include: temperature and humidity data and water immersion status data;

[0101] The safety parameter data may include oxygen concentration data;

[0102] The equipment operating parameter data may include: oxygen generator current data and oxygen generator vibration data;

[0103] The multi-source heterogeneous data can be collected via MQTT / HTTP protocol and supports multiple network transmission modes including TCP / IP, 4G, 5G, and WiFi.

[0104] In this example, by encompassing heterogeneous data types across multiple dimensions such as environment, security, and equipment operation, the limitations of traditional industrial monitoring, which only supports monitoring a single type of parameter, can be overcome. This enables comprehensive coverage and unified collection of various key data from industrial sites. Furthermore, leveraging MQTT / HTTP multi-protocol acquisition capabilities and multiple network transmission modes such as TCP / IP, 4G, 5G, and WiFi, it adapts to the differences in equipment interfaces and network environments across various industrial sites. This significantly improves the compatibility, flexibility, and efficiency of data collection, ensuring stable and rapid aggregation of multi-source data. It provides comprehensive and high-quality data source support for subsequent data cleaning, permission adaptation, and visualization, effectively meeting the diverse and complex monitoring data collection needs of industrial sites.

[0105] In one implementation, step 2 above, which involves real-time cleaning and anomaly detection of the multi-source heterogeneous data to obtain standardized processed data, may include:

[0106] The multi-source heterogeneous data is input into a preset streaming computing framework to filter out noise information in the data and obtain preliminary cleaned data.

[0107] The preliminary cleaned data is processed for cross-protocol compatibility based on the OPC-UA protocol to obtain preliminary cleaned data after unified modeling.

[0108] The preliminary cleaning data after unified modeling is input into a preset machine learning model to predict the fault status of equipment in the industrial site and obtain abnormal data.

[0109] The normal data in the preliminary cleaned data after unified modeling is standardized and transformed, and combined with the identification information of the abnormal data, standardized processed data containing normal standardized data and abnormal labeled data is generated.

[0110] The streaming computing framework can be the Apache Flink framework, and the machine learning model can be an LSTM model.

[0111] In this implementation, the specific steps of predicting equipment operating status in the industrial site and using a hybrid data training strategy when obtaining abnormal data include:

[0112] (1) Training data sources: equipment factory parameter calibration data (e.g., may include normal current / voltage range), historical fault records (e.g., may include full-dimensional data of oxygen generator shutdown events), and normal operating data (e.g., may include an average of 100,000 sampling points per day over the past 12 months).

[0113] (2) Threshold setting mechanism:

[0114] Static threshold: Set the initial alarm boundary based on the IEC 61508 standard (e.g., oxygen concentration <19.5%Vol).

[0115] Dynamic threshold: The concentration value is predicted for the next 3 minutes using an LSTM model. An early warning is triggered when the prediction error exceeds ±5% of the confidence interval.

[0116] Self-learning mechanism: The model is incrementally trained monthly using newly collected data to maintain a detection accuracy of ≥98%;

[0117] In the above implementation, the pre-set streaming computing framework can effectively filter noise information in multi-source heterogeneous data. Combined with the OPC-UA protocol, it can achieve cross-protocol compatibility and unified modeling of data from different sources, which is conducive to improving the purity and compatibility of the original data. Relying on multi-type training data to optimize the machine learning model, and with the judgment mechanism that combines static thresholds and dynamic thresholds, it can accurately capture abnormal states in the operation of industrial equipment and avoid the rigidity problem caused by single threshold judgment. At the same time, the self-learning mechanism enables the model to continuously adapt to new situations and changes in equipment operation, which is conducive to enhancing the robustness and timeliness of fault prediction. Finally, it generates standardized processed data containing normal standardized data and abnormal labeled data.

[0118] In one implementation, step 3 above, which involves generating a data access policy for pre-acquired user roles based on a preset role access control model and the field attributes of the standardized processed data, may include:

[0119] User role hierarchy is defined based on a preset role-based access control model; the user role hierarchy includes administrator role, maintenance personnel role, and visitor role;

[0120] Configure the corresponding data visibility scope and operation permissions for each user role level, and generate a role permission configuration table;

[0121] Based on the pre-obtained currently logged-in user role, query the role permission configuration table of the user role, and combine it with the field attributes of the standardized processed data to determine the data visibility range and operation permissions corresponding to the currently logged-in user role;

[0122] Based on the data visibility range and operation permissions, a data access policy is generated;

[0123] In this implementation, after generating the data access strategy, the corresponding UI data can be dynamically loaded or hidden through a front-end reactive framework (such as Vue.js framework) to achieve data permission isolation and UI update without page refresh.

[0124] For example, the process of implementing dynamic permission rendering using the Vue3 Composition API in the Vue.js framework can be as follows:

[0125] (1) Permission monitoring: Monitor changes in the userPermissions state in the store using watchEffect;

[0126] (2) Component rendering: Use v-if / v-show in combination with dynamic components;

[0127] (3) Data binding: Real-time calculation of visible fields using computed properties;

[0128] In the above implementation method, different user role levels are defined based on a preset role access control model. Corresponding data visibility range and operation permissions are configured for each level, and a role permission configuration table is generated. The permission range of the currently logged-in user is accurately determined by combining the field attributes of standardized data processing, and a targeted data access strategy is generated. Then, the interface data is dynamically loaded or hidden through a front-end responsive framework to achieve refresh-free updates and data permission isolation. This effectively breaks the limitations of rigid industrial monitoring permission management and fixed interface and data display. It is conducive to meeting the differentiated data access needs of different roles, and can ensure data security through precise permission control. It improves the flexibility and adaptability of permission management, and allows the interface data display to be dynamically adjusted according to user permissions, thereby optimizing the user experience.

[0129] In one implementation, step 4 above, which involves parsing the target data field using a visualization adaptation algorithm and matching the parsed target data field with a preset target chart template to generate adaptive visualization data, may include:

[0130] The automatic field recognition function in the visualization adaptation algorithm is used to parse the name and data type of the target data field and establish a mapping relationship between the field and the data type.

[0131] Based on preset field chart type mapping rules, the fields and data type mapping relationships are matched with the corresponding target chart templates;

[0132] The force-guided layout optimization function in the visualization adaptation algorithm is used to dynamically arrange the matched target chart template.

[0133] The contents of the corresponding target data fields in the standardized data are filled into the corresponding target chart template to generate adaptive visualization data;

[0134] The target chart template includes one or more of the following: bar chart template, heat map template, line chart template, and dashboard template;

[0135] In this implementation, the process of parsing the name and data type of the target data field and establishing the mapping relationship between the field and data type through the automatic field recognition function in the visualization adaptation algorithm may include:

[0136] (1) Syntax parsing: A domain-specific language (DSL) parser is built using ANTLR4 to identify field naming conventions. For example, the naming rules can be set as: {abbreviation of physical quantity}_{unit} (such as O2_Concentration, Temp_C);

[0137] (2) Contextual reasoning: Combined with device metadata (such as oxygen concentration detection corresponding to sensor model MET3000);

[0138] (3) Machine learning assistance: For example, a random forest classifier (with a training set containing 200+ field samples) can be introduced to predict data types; and finally, the best match is determined by the type priority matrix;

[0139] In this implementation, the parameter configuration for dynamically arranging the matched target chart template through the force-guided layout optimization function (e.g., D3.js) in the visualization adaptation algorithm can include:

[0140] const simulation = d3.forceSimulation(nodes)

[0141] .force("charge", d3.forceManyBody().strength(-200)) / / Node repulsion force

[0142] .force("link", d3.forceLink(links).distance(100)) / / Length of the link

[0143] .force("center", d3.forceCenter(width / 2, height / 2)) / / Center force

[0144] .force("collision", d3.forceCollide().radius(20)) / / Collision detection

[0145] .alphaDecay(0.02) / / Cooling rate

[0146] .on("tick", ticked); / / Callback for each iteration

[0147] Layout constraint rules can be set as follows:

[0148] (1) Key indicators (such as alarm items) should be placed in the center of the view;

[0149] (2) Data groups of the same type are kept horizontally aligned (achieved through a custom force field);

[0150] (3) The spacing between charts should be ≥50px to avoid visual obstruction;

[0151] (4) Supports elastic rebound effect after manual dragging (stiffness coefficient 0.8);

[0152] In the above implementation method, the automatic field recognition function in the visualization adaptation algorithm, combined with syntax parsing, contextual reasoning and machine learning assistance, accurately parses the name and type of the target data field and establishes a mapping relationship. Then, based on preset rules, it matches the corresponding target chart template. At the same time, with the help of the force-oriented layout optimization function and reasonable layout constraint rules, the matched chart template is dynamically arranged. This can effectively avoid the problems of traditional industrial monitoring visualization relying on fixed templates, being unable to automatically adapt to data types, and having messy and easily obscured chart layouts. It allows the data display to automatically adapt to the appropriate chart form according to the field attributes, and the chart arrangement is more regular and orderly, which is conducive to improving the adaptability and visual coherence of visualization, enabling data information to be conveyed more clearly and efficiently, thereby optimizing the user's viewing experience.

[0153] In one implementation, step 5 above, based on an edge cloud collaborative architecture, pushes the adaptive visualization data to a web display terminal via a remote transmission mechanism, including:

[0154] Edge servers are deployed in the local data center at the industrial site to cache and store standardized processing data that is accessed frequently, generating edge cache data.

[0155] Determine whether the original data corresponding to the adaptive visualization data exists in the edge cache data:

[0156] If it exists, obtain the corresponding edge cache data from the edge server, and generate push data by combining the chart configuration information of the adaptive visualization data;

[0157] If it does not exist, retrieve the corresponding raw data from the cloud database and generate push data by combining the chart configuration information of the adaptive visualization data.

[0158] The push data is transmitted to the web display terminal via a WebSocket long connection in the remote transmission mechanism.

[0159] For example, the cloud database can adopt a hybrid architecture of time-series database and relational database;

[0160] The time-series database can be InfluxDB;

[0161] The relational database can be MySQL;

[0162] The web display terminal can render the pushed data using a WebGL-based 3D large-screen rendering engine;

[0163] In the above implementation, by introducing an edge-cloud collaborative architecture, multi-dimensional data optimization and preliminary processing can be completed at the edge, effectively reducing cloud pressure and improving system stability. Specifically, the OpenVINO framework can be used to perform noise reduction on sensor data through mean filtering and wavelet transform to ensure data purity. Simultaneously, converting the ModbusRTU protocol to the MQTT protocol significantly reduces the data parsing burden on the cloud and supports the deployment of lightweight TensorFlow Lite models to achieve preliminary screening of device faults, improving the timeliness of fault detection. Regarding the collaborative mechanism, when the local CPU load reaches a set threshold, non-real-time data is automatically forwarded to the cloud for processing, achieving load balancing between the edge and the cloud through Azure IoT. Edge implements two-way authentication between edge nodes and the cloud to ensure data transmission and access security. A data buffer is also set up at the edge to ensure data integrity during network interruptions. By deploying edge servers in local data centers at industrial sites to cache frequently accessed standardized processing data, repeated access to the cloud database is reduced, significantly lowering the load on the central server. Furthermore, by determining whether the original data corresponding to the adaptive visualization data exists in the edge cache, data can be flexibly retrieved from either the edge server or the cloud database, optimizing the data retrieval path and avoiding unnecessary remote data transmission time. Finally, by combining WebSocket long connections to push data to the web display terminal, low-latency push and continuous transmission of data are achieved, effectively avoiding the time-consuming and stuttering issues caused by frequent connection establishment in traditional data transmission methods. This further improves the real-time performance and stability of adaptive visualization data push, allowing the web display terminal to quickly and smoothly present industrial equipment monitoring data, significantly improving the data display experience during remote monitoring.

[0164] Furthermore, existing intelligent monitoring processes suffer from the problem that adding new monitoring items requires redeveloping interfaces, making it difficult to quickly respond to changes in business requirements. This invention addresses this by standardizing interface design and configuring data parsing rules. Adding new monitoring items does not require modifying the core code; only adjustments to the relevant configurations are needed to quickly respond to changes in business requirements. This design enables the system to flexibly adapt to various scenarios such as industrial automation and energy management, significantly reducing monitoring upgrade and maintenance costs and improving the practicality and feasibility of technology implementation.

[0165] For example, the configuration content of the interface can be as follows:

[0166] (1) Data acquisition interface: It can support OPC-UA / Modbus protocol conversion and define a three-level nested structure containing device ID, timestamp, and numerical type;

[0167] {

[0168] "device_id": "string",

[0169] "timestamp": "ISO8601",

[0170] "metrics": [

[0171] {

[0172] "type": "temperature",

[0173] "value": 23.5,

[0174] "unit": "℃"

[0175] } ]

[0177] }

[0178] (2) Permission verification interface: Based on the OAuth2.0 protocol, it returns a JWT token containing role permissions.

[0179] {

[0180] "role": "admin",

[0181] "permissions": ["view_oxygen", "control_device"]

[0182] }

[0183] (3) Visual configuration interface: Provides dynamic template loading capability, defines chart type enumeration and parameter specifications.

[0184] {

[0185] "chart_type": "line|bar|heatmap",

[0186] "x_field": "timestamp",

[0187] "y_fields": ["temp", "humidity"]

[0188] }

[0189] Through the standardized interface design and data parsing rule configuration described above, a unified interface specification and structure are defined for core aspects such as data collection, permission verification, and visualization configuration. This allows new monitoring projects to be added without modifying the core system code; expansion can be completed simply by adjusting relevant configurations. This effectively solves the problem that existing intelligent monitoring systems need to redevelop interfaces for new projects and are unable to quickly respond to changes in business requirements. As a result, scalability and adaptability are greatly improved, enabling the system to flexibly adapt to the monitoring needs of different industrial scenarios. At the same time, it can significantly reduce upgrade and maintenance costs and enhance the practicality and feasibility of technology implementation.

[0190] In summary, this invention addresses the technical problems of insufficient multi-source heterogeneous data fusion capability, rigid access control, poor visualization adaptability, and remote access delays in traditional industrial monitoring systems. It proposes an intelligent monitoring method for industrial equipment based on multi-source data fusion. Through real-time multi-source data acquisition, intelligent access control, and adaptive visualization technology, it supports unified monitoring and dynamic display of multiple parameter types. Data access isolation and interface adaptation are achieved based on user roles. Intelligent algorithms optimize data rendering efficiency to ensure real-time remote access, and standardized interfaces facilitate rapid expansion of new monitoring items. This effectively solves the problems of poor flexibility and low response efficiency in traditional systems, thereby significantly improving the overall performance and practical application value of intelligent monitoring of industrial equipment.

[0191] Example 2:

[0192] Based on the same inventive concept, this invention also provides an intelligent monitoring system for industrial equipment based on multi-source data fusion, the structural composition of which is shown in the schematic diagram below. Figure 2 As shown, it includes:

[0193] The data acquisition module is used to acquire multi-source heterogeneous data from industrial sites;

[0194] The data processing module is used to perform real-time cleaning and anomaly detection on the multi-source heterogeneous data to obtain standardized processed data.

[0195] The access control module is used to generate data access policies for pre-acquired user roles based on a preset role access control model and the field attributes of the standardized processed data.

[0196] The visualization adaptation module is used to filter target data fields from the standardized processed data according to the data access strategy, and parse the target data fields through the visualization adaptation algorithm, and match the parsed target data fields with the preset target chart template to generate adaptive visualization data.

[0197] The display push module is used to push the adaptive visualization data to the Web display terminal through a remote transmission mechanism based on the edge cloud collaborative architecture;

[0198] The visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection.

[0199] For example, the multi-source heterogeneous data may include one or more of the following: environmental parameter data, safety parameter data, and equipment operating parameter data;

[0200] The environmental parameter data may include: temperature and humidity data and water immersion status data;

[0201] The safety parameter data may include oxygen concentration data;

[0202] The equipment operating parameter data may include: oxygen generator current data and oxygen generator vibration data;

[0203] The multi-source heterogeneous data can be collected via MQTT / HTTP protocol and supports multiple network transmission modes including TCP / IP, 4G, 5G, and WiFi.

[0204] In one implementation, the data processing module may include:

[0205] The data cleaning submodule is used to input the multi-source heterogeneous data into a preset streaming computing framework, filter out noise information in the data, and obtain preliminary cleaned data.

[0206] The protocol compatibility submodule is used to perform cross-protocol compatibility processing on the preliminary cleaned data based on the OPC-UA protocol to obtain preliminary cleaned data after unified modeling.

[0207] The anomaly detection submodule is used to input the preliminary cleaning data after unified modeling into a preset machine learning model to predict the fault status of equipment in the industrial site and obtain anomaly data.

[0208] The standardization submodule is used to standardize the normal data in the preliminary cleaned data after unified modeling, and combine the identification information of the abnormal data to generate standardized processed data containing normal standardized data and abnormal labeled data.

[0209] The streaming computing framework is the Apache Flink framework, and the machine learning model is the LSTM model.

[0210] In one implementation, the access control module may include:

[0211] The role definition submodule is used to define user role levels based on a preset role access control model; the user role levels include administrator roles, operation and maintenance personnel roles, and visitor roles;

[0212] The permission configuration submodule is used to configure the corresponding data visibility scope and operation permissions for each user role level and generate a role permission configuration table.

[0213] The role acquisition submodule is used to query the role permission configuration table of the currently logged-in user role based on the pre-acquired currently logged-in user role, and determine the data visibility range and operation permissions corresponding to the currently logged-in user role by combining the field attributes of the standardized processing data.

[0214] The policy generation submodule is used to generate data access policies based on the data visibility scope and operation permissions.

[0215] In one implementation, the visualization adaptation module may include:

[0216] The field parsing submodule is used to parse the name and data type of the target data field through the automatic field recognition function in the visualization adaptation algorithm, and establish a mapping relationship between the field and the data type;

[0217] The chart matching submodule is used to match the field and data type mapping relationship with the corresponding target chart template based on the preset field chart type mapping rules;

[0218] The layout optimization submodule is used to dynamically arrange the matched target chart template through the force-oriented layout optimization function in the visualization adaptation algorithm.

[0219] The data population submodule is used to populate the content of the corresponding target data field in the standardized data into the corresponding target chart template to generate adaptive visualization data;

[0220] The target chart template includes one or more of the following: bar chart template, heat map template, line chart template, and dashboard template.

[0221] In one implementation, the display push module may include:

[0222] The edge caching submodule is used to deploy edge servers in the local data center of the industrial site to cache and store standardized processing data that is accessed frequently, and generate edge cache data.

[0223] The data routing submodule is used to determine whether the original data corresponding to the adaptive visualization data exists in the edge cache data: if it exists, the corresponding edge cache data is obtained from the edge server, and push data is generated by combining the chart configuration information of the adaptive visualization data; if it does not exist, the corresponding original data is obtained from the cloud database, and push data is generated by combining the chart configuration information of the adaptive visualization data.

[0224] The transmission and display submodule is used to transmit the push data to the Web display terminal through the WebSocket long connection in the remote transmission mechanism.

[0225] For example, the cloud database can adopt a hybrid architecture of time-series database and relational database;

[0226] The time-series database can be InfluxDB;

[0227] The relational database can be MySQL;

[0228] The web display terminal can render the pushed data using a WebGL-based 3D large-screen rendering engine.

[0229] Example 3:

[0230] like Figure 3 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0231] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the intelligent monitoring method for industrial equipment based on multi-source data fusion in the above embodiments.

[0232] Example 4:

[0233] Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the intelligent monitoring method for industrial equipment based on multi-source data fusion in the above embodiments.

[0234] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0235] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0236] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0237] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0238] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the application, but these changes, modifications or equivalent substitutions are all within the scope of protection of the claims pending approval.

Claims

1. A method for intelligent monitoring of industrial equipment based on multi-source data fusion, characterized in that, include: Acquire multi-source heterogeneous data from industrial sites; The multi-source heterogeneous data is cleaned and anomaly detected in real time to obtain standardized processed data; Based on the preset role access control model and combined with the field attributes of the standardized processed data, a data access strategy is generated for the pre-acquired user roles. According to the data access strategy, target data fields are selected from the standardized processed data, and the target data fields are parsed through a visualization adaptation algorithm. The parsed target data fields are then matched with a preset target chart template to generate adaptive visualization data. Based on the edge cloud collaborative architecture, the adaptive visualization data is pushed to the Web display terminal through a remote transmission mechanism; The visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection.

2. The method as described in claim 1, characterized in that, The multi-source heterogeneous data includes one or more of the following: environmental parameter data, safety parameter data, and equipment operating parameter data; The environmental parameter data includes: temperature and humidity data and water immersion status data; The safety parameter data includes oxygen concentration data; The equipment operating parameter data includes: oxygen generator current data and oxygen generator vibration data; The multi-source heterogeneous data is collected via MQTT / HTTP protocol and supports multiple network transmission modes including TCP / IP, 4G, 5G, and WiFi.

3. The method as described in claim 2, characterized in that, The process of real-time cleaning and anomaly detection of the multi-source heterogeneous data to obtain standardized processed data includes: The multi-source heterogeneous data is input into a preset streaming computing framework to filter out noise information in the data and obtain preliminary cleaned data. The preliminary cleaned data is processed for cross-protocol compatibility based on the OPC-UA protocol to obtain preliminary cleaned data after unified modeling. The preliminary cleaning data after unified modeling is input into a preset machine learning model to predict the fault status of equipment in the industrial site and obtain abnormal data. The normal data in the preliminary cleaned data after unified modeling is standardized and transformed, and combined with the identification information of the abnormal data, standardized processed data containing normal standardized data and abnormal labeled data is generated. The streaming computing framework is the Apache Flink framework, and the machine learning model is the LSTM model.

4. The method as described in claim 1, characterized in that, The method, based on a preset role-based access control model and combined with the field attributes of the standardized processed data, generates data access policies for pre-acquired user roles, including: User role hierarchy is defined based on a preset role-based access control model; the user role hierarchy includes administrator role, maintenance personnel role, and visitor role; Configure the corresponding data visibility scope and operation permissions for each user role level, and generate a role permission configuration table; Based on the pre-obtained currently logged-in user role, query the role permission configuration table of the user role, and combine it with the field attributes of the standardized processed data to determine the data visibility range and operation permissions corresponding to the currently logged-in user role; A data access policy is generated based on the data visibility range and operation permissions.

5. The method as described in claim 1, characterized in that, The step of parsing the target data field using a visualization adaptation algorithm, matching the parsed target data field with a preset target chart template, and generating adaptive visualization data includes: The automatic field recognition function in the visualization adaptation algorithm is used to parse the name and data type of the target data field and establish a mapping relationship between the field and the data type. Based on preset field chart type mapping rules, the fields and data type mapping relationships are matched with the corresponding target chart templates; The force-guided layout optimization function in the visualization adaptation algorithm is used to dynamically arrange the matched target chart template. The contents of the corresponding target data fields in the standardized data are filled into the corresponding target chart template to generate adaptive visualization data; The target chart template includes one or more of the following: bar chart template, heat map template, line chart template, and dashboard template.

6. The method as described in claim 1, characterized in that, The edge cloud collaborative architecture pushes the adaptive visualization data to the web display terminal via a remote transmission mechanism, including: Edge servers are deployed in the local data center at the industrial site to cache and store standardized processing data that is accessed frequently, generating edge cache data. Determine whether the original data corresponding to the adaptive visualization data exists in the edge cache data: If it exists, obtain the corresponding edge cache data from the edge server, and generate push data by combining the chart configuration information of the adaptive visualization data; If it does not exist, retrieve the corresponding raw data from the cloud database and generate push data by combining the chart configuration information of the adaptive visualization data. The push data is transmitted to the web display terminal via a WebSocket long connection in the remote transmission mechanism.

7. The method as described in claim 6, characterized in that, The cloud database adopts a hybrid architecture of time-series database and relational database; The time-series database is InfluxDB; The relational database is MySQL; The web display terminal renders the pushed data using a WebGL-based 3D large-screen rendering engine.

8. An intelligent monitoring system for industrial equipment based on multi-source data fusion, characterized in that, include: The data acquisition module is used to acquire multi-source heterogeneous data from industrial sites; The data processing module is used to perform real-time cleaning and anomaly detection on the multi-source heterogeneous data to obtain standardized processed data. The access control module is used to generate data access policies for pre-acquired user roles based on a preset role access control model and the field attributes of the standardized processed data. The visualization adaptation module is used to filter target data fields from the standardized processed data according to the data access strategy, and parse the target data fields through the visualization adaptation algorithm, and match the parsed target data fields with the preset target chart template to generate adaptive visualization data. The display push module is used to push the adaptive visualization data to the Web display terminal through a remote transmission mechanism based on the edge cloud collaborative architecture; The visualization adaptation algorithm is constructed through automatic field recognition and force-oriented layout optimization, and the remote transmission mechanism is implemented through edge node caching and WebSocket long connection.

9. The system as described in claim 8, characterized in that, The multi-source heterogeneous data includes one or more of the following: environmental parameter data, safety parameter data, and equipment operating parameter data; The environmental parameter data includes: temperature and humidity data and water immersion status data; The safety parameter data includes oxygen concentration data; The equipment operating parameter data includes: oxygen generator current data and oxygen generator vibration data; The multi-source heterogeneous data is collected via MQTT / HTTP protocol and supports multiple network transmission modes including TCP / IP, 4G, 5G, and WiFi.

10. The system as described in claim 9, characterized in that, The data processing module includes: The data cleaning submodule is used to input the multi-source heterogeneous data into a preset streaming computing framework, filter out noise information in the data, and obtain preliminary cleaned data. The protocol compatibility submodule is used to perform cross-protocol compatibility processing on the preliminary cleaned data based on the OPC-UA protocol to obtain preliminary cleaned data after unified modeling. The anomaly detection submodule is used to input the preliminary cleaning data after unified modeling into a preset machine learning model to predict the fault status of equipment in the industrial site and obtain anomaly data. The standardization submodule is used to standardize the normal data in the preliminary cleaned data after unified modeling, and combine the identification information of the abnormal data to generate standardized processed data containing normal standardized data and abnormal labeled data. The streaming computing framework is the Apache Flink framework, and the machine learning model is the LSTM model.

Images