Life cycle management system for electrical equipment maintenance

Abstract

The application discloses a full life cycle management system for electrical equipment maintenance, belongs to the technical field of electrical equipment management and maintenance, and is based on a three-in-one technical architecture of multi-scene perception-cross-scene fusion-closed loop management, realizes accurate quantification and trend prediction of the health state of electrical equipment in the full life cycle, and specifically comprises the following steps: multi-modal data and multi-dimensional scene labels thereof are synchronously collected, are automatically classified by a scene classifier, and interference of working condition differences on evaluation is eliminated; single-point data is normalized by a scene sub-model, and a single-point health index is output; single-point health indexes across scenes are aligned and analyzed by a cross-scene fusion model, an equipment health evolution track is constructed, and future trends are predicted; grading early warning instructions are generated based on fusion analysis of current and future trends, maintenance tasks are automatically triggered and tracked in the whole process, a business closed loop is formed, a change from post-diagnosis to pre-prevention is realized, and evaluation accuracy is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of electrical equipment management and maintenance technology, and more specifically, to a full life-cycle management system for electrical equipment maintenance. Background Technology

[0002] Currently, many industrial enterprises (such as steel manufacturing enterprises) still use traditional offline methods for the maintenance and management of electrical equipment, such as paper-based authorization tickets, manual recording, and manual inspection, which result in problems such as low equipment operating efficiency, high maintenance costs, and shortened equipment lifespan.

[0003] To overcome the limitations of traditional offline methods, a site-wide equipment information management system disclosed in patent publication number CN119295056A focuses on how to improve the accuracy of fault diagnosis in complex environments (such as strong electromagnetic interference). This type of system typically collects real-time environmental parameters of the equipment through sensors and uses a predictive model (such as a positive / negative interference completion module) to correct or complete the interfered data. Then, the corrected data is matched with a fault parameter database to output a specific fault judgment, thereby assisting maintenance engineers in quickly locating the problem.

[0004] However, existing technologies of this kind mainly focus on fault diagnosis, that is, identifying the specific fault type through data correction and pattern matching after equipment malfunctions. In essence, they are still auxiliary tools for "post-incident maintenance" and have the following technical limitations: First, there is a lack of refined perception of equipment operating conditions: usually all collected data are evaluated under the same standard, ignoring the significant differences in the baseline values ​​of normal operating parameters when the equipment is running under different conditions (such as load level, ambient temperature, humidity, etc.). For example, the vibration value of the same motor under high temperature full load and low temperature no load conditions is different. If a uniform evaluation threshold is used, it is very easy to lead to misjudgment or omission.

[0005] Second, it is difficult to achieve data comparability across operating conditions: due to the lack of labeling and classification of data collection scenarios, existing technologies cannot normalize data collected at different times and under different conditions and place it in the same dimension for comparative analysis. Therefore, it is difficult to construct the long-term evolution trajectory of equipment health status and achieve true trend prediction.

[0006] Third, it cannot support preventative maintenance decisions. Current technologies output "what fault has occurred," rather than "how the equipment's health status changes and when a fault might occur." Maintenance personnel can only respond passively after a fault occurs, unable to plan maintenance in advance or optimize maintenance strategies, thus limiting the level of digitalization and intelligence in equipment management.

[0007] Therefore, there is an urgent need for an electrical equipment management system that can sense equipment operating scenarios, enable cross-condition data comparability, and support health trend prediction, so as to truly promote the transformation of equipment maintenance from "reactive maintenance" to "preventive maintenance". Summary of the Invention

[0008] The purpose of this invention is to address practical technical deficiencies. It provides a full lifecycle management system for electrical equipment maintenance, which uniquely encodes equipment, synchronously collects multimodal data and multidimensional scene tags, achieves differentiated assessment based on scene classification, and predicts health trends through cross-scene fusion analysis. Finally, it generates graded early warning instructions and drives the closed loop of the maintenance process, thereby upgrading equipment management from passive "post-event maintenance" to proactive "pre-event prediction", improving equipment operating efficiency and management level.

[0009] The objective of this invention can be achieved through the following technical solution: a full lifecycle management system for electrical equipment maintenance, comprising: The device unique identification coding module is used to assign and bind a unique identification code to each target device, and to establish an electronic file containing information on the entire process of the device from online, operation, maintenance to scrapping. The multimodal device status monitoring module is used to collect dynamic operating parameters of the target device in real time when it is running, and to receive offline detection parameters of the target device when it is not running. It synchronously collects and records the multi-dimensional scene label information corresponding to each set of parameters, forming multimodal data with multi-dimensional scene label information. The scene recognition and classification module has a built-in scene classifier, which is used to automatically classify multimodal data into one of a number of predefined scene categories based on scene label information; The differentiated intelligent diagnostic module integrates a large AI model containing multiple scene sub-models and a cross-scene fusion and trend prediction model. Each scene sub-model corresponds to a scene category. Based on the classified scene category, the corresponding scene sub-model is automatically called to perform single-scene evaluation of multimodal data. Based on the evaluation results of multiple single scenes, cross-scene fusion and trend prediction are performed through the cross-scene fusion and trend prediction model to generate warning instructions at different levels. The maintenance process management module is used to respond to the warning command, automatically generate maintenance tasks, and perform full-process visual tracking and recording of the online flow, execution progress and results of maintenance tasks. The maintenance records are automatically archived to the electronic archive.

[0010] Furthermore, the dynamic operating parameters include at least electrical quantity parameters, non-electrical quantity parameters, and environmental parameters, and the offline detection parameters include at least insulation characteristic parameters, mechanical characteristic parameters, and oil analysis parameters.

[0011] Furthermore, the multi-dimensional scene labels include at least equipment status labels, environmental condition labels, load level labels, and test condition labels. Equipment status labels include running, hot standby, cold standby, and shutdown for maintenance. Environmental condition labels include temperature range, humidity level, and weather conditions. Load level labels include no-load, light-load, rated load, and overload. Test condition labels include test voltage level, wiring method, and instrument model. Interference level labels include electromagnetic interference intensity and background noise level.

[0012] Furthermore, the specific process of single-scenario evaluation, cross-scenario fusion, and trend prediction of multimodal data includes: S1. When receiving multimodal data with multidimensional scene labels, automatically call the corresponding scene sub-model to perform single-scene device health status assessment according to the scene category, and output a normalized single-point health index. S2. When there is cross-scenario category data for the same target device, receive multiple single-point health indices and their corresponding scenario labels of the same target device in the historical time series, input them into the cross-scenario fusion and trend prediction model, construct the time-series evolution trajectory of the device's health status, and predict the health index of future time windows based on the evolution trajectory, and draw the future health index trend. S3. Based on the current single-point health index and the predicted future health index trend, generate early warning instructions of different levels according to the preset fusion decision rules.

[0013] Furthermore, the process of generating different levels of early warning instructions according to preset fusion decision rules includes: If the current single-point health index is less than the single-point health index threshold, an emergency repair warning command will be generated. Conversely, if the current single-point health index is higher than the single-point health index threshold, but the future health index will be lower than the cross-point health index threshold within a future time window, then a preventive maintenance early warning instruction will be generated. If the rate of decline of the health index exceeds the set lower limit, a status alert will be generated, prompting an increase in the monitoring frequency.

[0014] Furthermore, the scene sub-model and the cross-scene fusion and trend prediction model adopt a unified AI big model architecture, and train single-point evaluation and trend prediction capabilities simultaneously through multi-task learning.

[0015] Furthermore, the full lifecycle management system for electrical equipment maintenance also includes a unified data platform module, which is used to integrate and store the electronic records, multimodal data, multidimensional scene tags, single-point health indices, evolution trajectories and early warning instructions, and to provide data query, analysis and display services.

[0016] Compared with the prior art, the advantages of this invention are: This invention constructs multimodal data by synchronously collecting dynamic operating parameters, offline detection parameters, and their corresponding multidimensional scene label information. Based on a scene classifier, the multimodal data is automatically classified into multiple predefined scene categories, so that each type of data is analyzed under an evaluation standard adapted to its collection conditions. This enables scene-based and accurate evaluation of equipment operating status. Compared with the existing technology that uses a uniform threshold evaluation method, this invention can effectively eliminate the interference of operating condition differences on equipment status judgment, significantly improve the accuracy and reliability of health assessment, and avoid misjudgment or omission due to scene changes.

[0017] This invention also combines single-scenario assessment with cross-scenario fusion and trend prediction. It normalizes single-point data through multiple scenario sub-models to output comparable health indices. Then, through cross-scenario fusion and trend prediction models, it performs time-series alignment of health indices under different times and operating conditions, constructs the evolution trajectory of equipment health status, and predicts future trends. This solves the technical problem that existing technologies cannot achieve cross-operating condition data comparability analysis and trend prediction. Finally, based on the current health index and predicted trends, it generates hierarchical early warning instructions, automatically triggers maintenance tasks, and provides full-process visual tracking, realizing a fundamental shift from "post-event diagnosis" to "pre-event prevention," effectively extending equipment online lifespan and reducing unplanned downtime.

[0018] This invention also uses a unique equipment identification code to span the entire process of equipment from commissioning, operation, maintenance to scrapping, and unifies all information such as multimodal monitoring data, scene tags, evaluation results, early warning instructions, and maintenance records into electronic archives. Furthermore, the maintenance process management module enables the automatic generation, online transfer, and result feedback of maintenance tasks, forming a business closed loop of "monitoring-evaluation-early warning-maintenance-archiving". Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the principle of collaborative operation of the various modules of the present invention; Figure 2 This is a flowchart of the management method of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] Example 1: This invention discloses a full lifecycle management system for electrical equipment maintenance. Please refer to [link / reference]. Figure 1It includes a unique device identification coding module, a multimodal device status monitoring module, a scene recognition and classification module, a differentiated intelligent diagnosis module, a maintenance process management module, and a unified data platform module.

[0022] The device unique identification coding module is used to assign and bind a unique identification code to each target device, establish a full lifecycle electronic file containing information on the entire process from device commissioning, operation, maintenance to scrapping, and provide device identification services.

[0023] The multimodal device status monitoring module is used to collect dynamic operating parameters of the target device in real time when it is running, and to receive offline detection parameters of the target device when it is not running. It synchronously collects and records the multi-dimensional scene label information corresponding to each set of parameters, forming multimodal data with multi-dimensional scene label information. Among them, dynamic operating parameters include at least electrical quantity parameters (voltage, current, power, frequency, harmonics), non-electrical quantity parameters (temperature, vibration, partial discharge, sound) and environmental parameters (ambient temperature, ambient humidity). Offline testing parameters include at least insulation characteristic parameters (insulation resistance, polarization index, dielectric loss factor, capacitance), mechanical characteristic parameters (opening and closing time, speed, stroke), and oil analysis parameters (trace water content, acid value, breakdown voltage, dissolved gas). Offline testing data refers to the intrinsic state data of the equipment obtained through specialized testing methods when the equipment is not in operation or is out of the operating environment. Multidimensional scenario labels include at least equipment status labels, environmental condition labels, load level labels, and test condition labels. Equipment status labels include running, hot standby, cold standby, and shutdown for maintenance. Among them, environmental condition labels include temperature range, humidity level, and weather conditions. Load level labels include no load, light load, rated load, and overload. Test condition labels include test voltage level, wiring method, and instrument model. Interference level labels include electromagnetic interference intensity and background noise level. Synchronously collecting operational status data, offline detection data, and their multi-dimensional scene labels solves the shortcomings of traditional solutions that rely on only a single type of data or ignore collection conditions. By attaching scene labels to each set of data, subsequent evaluations can perceive the specific operating conditions that generated the data, providing a data prerequisite for eliminating the interference of operating condition differences on the evaluation.

[0024] The scene recognition and classification module has a built-in scene classifier, which automatically classifies multimodal data into one of several predefined scene categories based on scene label information. The scene category is a high-level abstraction of the combination of multidimensional scene labels. Each scene category corresponds to a set of working conditions with similar collection conditions and evaluation standards. For example, the running state scene category (rated load normal temperature operation scenario, rated load high temperature operation scenario, rated load high temperature and high humidity scenario, light load normal temperature operation scenario, overload operation scenario, special working condition operation scenario), the shutdown state scene category (shutdown insulation test scenario, shutdown insulation test low temperature scenario, shutdown mechanical characteristic test scenario, shutdown oiling sampling scenario, shutdown partial discharge test scenario), and the standby state scene category (hot standby status monitoring scenario, cold standby status monitoring scenario). Each scene category has an independently set judgment threshold, and there is a mapping relationship between the judgment thresholds of each category, supporting cross-scene comparison.

[0025] This module has a built-in scene classifier that automatically categorizes labeled multimodal data into predefined scene categories (such as rated load high-temperature operation scenario, shutdown insulation test scenario, etc.). Each category corresponds to an independent evaluation standard and threshold, and a mapping relationship is established between the thresholds of each category. This mechanism enables the system to accurately identify the current operating condition of the equipment, avoid misjudgments caused by changes in operating conditions, and significantly improve the pertinence and accuracy of the evaluation.

[0026] The differentiated intelligent diagnostic module integrates a large AI model containing multiple scene sub-models and a cross-scene fusion and trend prediction model. Each scene sub-model corresponds to a scene category. Based on the classified scene category, the corresponding scene sub-model is automatically called to perform single-scene evaluation on multimodal data. Based on the evaluation results of multiple single scenes, the cross-scene fusion and trend prediction model is used to perform cross-scene fusion and trend prediction to generate warning instructions at different levels. The scene sub-models and the cross-scene fusion and trend prediction model adopt a unified large AI model architecture, and the single-point evaluation and trend prediction capabilities are trained simultaneously through multi-task learning. The specific process of single-scenario evaluation, cross-scenario fusion, and trend prediction of multimodal data includes: S1. When receiving multimodal data with multidimensional scene labels, the corresponding scene sub-model is automatically called to evaluate the health status of single-scene devices according to the scene category, and a normalized single-point health index is output. After the normalization processing of the scene sub-model, the original data under different conditions are converted into a single-point health index of the same scale. At this time, trend analysis can be performed across time and conditions. S2. When there is cross-scenario category data for the same target device, receive multiple single-point health indices and their corresponding scenario labels of the same target device in the historical time series, input them into the cross-scenario fusion and trend prediction model, construct the time-series evolution trajectory of the device's health status, and predict the health index of future time windows based on the evolution trajectory, and draw the future health index trend. Cross-scenario refers to the fusion of evaluation results generated by the same device at different times, under different data collection conditions (i.e., different scenario categories). Its core purpose is not to compare different scenarios themselves, but to construct a unified "evolutionary trajectory" that reflects the intrinsic state of the device over time after eliminating the influence of scenario differences. S3. Based on the current single-point health index and the predicted future health index trend, generate early warning instructions of different levels according to the preset fusion decision rules. The process of generating different levels of early warning instructions according to preset fusion decision rules includes: If the current single-point health index is less than the single-point health index threshold, an emergency repair warning command will be generated. Conversely, if the current single-point health index is higher than the single-point health index threshold, but the future health index will be lower than the cross-point health index threshold within a future time window, then a preventive maintenance early warning instruction will be generated. If the rate of continuous decline of the health index exceeds the set lower limit, a status alert will be generated, prompting an increase in the monitoring frequency. This module integrates scenario sub-models and cross-scenario fusion and trend prediction models, forming a two-layer diagnostic architecture of "single-point assessment - trend prediction". First, the corresponding sub-model is called according to the scenario category to normalize the current data and output comparable single-point health indices, eliminating the interference of working condition differences on data interpretation. Second, the cross-scenario fusion model receives the health index sequence of the same device under different times and working conditions. Through time series alignment and trend analysis, it constructs the evolution trajectory of the device's health status and predicts future health change trends. Unlike traditional fault diagnosis systems, this invention does not directly pursue accurate identification of fault types. Instead, it achieves quantitative expression of equipment health status and early prediction of evolution trends through scenario-based health assessment and trend prediction. This mechanism elevates equipment management from a static judgment of "whether it is currently abnormal" to a dynamic prediction of "how the trend changes and when it may become abnormal," truly realizing a fundamental shift from "post-event diagnosis" to "pre-event prevention." This technical approach avoids the technical bottlenecks of scarce fault samples and complex fault mechanisms in existing systems, and has stronger versatility and practicality.

[0027] The maintenance process management module is used to respond to early warning commands, automatically generate maintenance tasks, and provide full-process visual tracking and recording of the online flow, execution progress and results of maintenance tasks. Maintenance records are automatically archived to electronic files. Based on preset fusion decision-making rules, tiered early warning instructions (emergency maintenance, preventative maintenance, and status monitoring) are generated, and the entire process of maintenance tasks is automatically tracked. Maintenance results are automatically archived to the equipment's electronic file, forming a closed-loop business process of "monitoring-evaluation-early warning-maintenance-archiving".

[0028] The unified data platform module integrates and stores electronic records, multimodal data, multidimensional scene tags, single-point health indices, evolutionary trajectories, and early warning commands. It also provides data query, analysis, and visualization services. This module integrates and stores the entire system's electronic records, multimodal data, scene tags, health indices, evolutionary trajectories, and early warning commands, offering multi-dimensional data query, analysis, and visualization services. The platform breaks down information silos between different business processes, supports scientific decision-making by management, and comprehensively improves the digitalization and intelligence of equipment management.

[0029] Example 2: Based on the management system of Example 1, a method for full lifecycle management of electrical equipment maintenance is proposed. Please refer to [link / reference]. Figure 2 This includes the following steps: S1. Equipment Identification Coding and Filing: Through the equipment unique identification coding module, a unique identification code is assigned and bound to each target electrical equipment to establish an electronic file containing information on the entire process of equipment from online, operation, maintenance to scrapping. S2. Multimodal data acquisition and scene labeling: Through the multimodal equipment status monitoring module, the dynamic operating parameters of the target electrical equipment in the running state are collected in real time, the offline detection parameters of the target electrical equipment in the non-operating state are received, and the multi-dimensional scene label information corresponding to each set of parameters is collected and recorded simultaneously to form multimodal data with multi-dimensional scene label information. S3. Scene Recognition and Classification: Using the scene classifier built into the scene recognition and classification module, multimodal data is automatically classified into one of the predefined scene categories based on scene label information; S4. Dual-layer intelligent diagnosis and early warning generation: Through the differentiated intelligent diagnosis module, the corresponding scenario sub-model is automatically called to perform single-point health status assessment on multimodal data according to the classified scenario category, and outputs a normalized health index. Meanwhile, through cross-scenario fusion and trend prediction models, multiple single-point health indices and their corresponding scenario labels of the same device in historical time series are received, the temporal evolution trajectory of the device's health status is constructed, and the future health index change trend is predicted based on the evolution trajectory. According to the single-point health index at the current moment and the predicted future health index trend, different levels of early warning instructions are generated according to the preset fusion decision rules. S5. Closed-loop management of maintenance tasks: Through the maintenance process management module, in response to early warning commands, maintenance tasks are automatically generated, and the online flow, execution progress and results of maintenance tasks are visualized and tracked and recorded throughout the entire process. The generated maintenance records are automatically archived to electronic files. S6. Data Platform Integration and Display: Through a unified data platform module, it integrates and stores electronic archives, multimodal data, multidimensional scene tags, single-point health indices, evolution trajectories and early warning instructions, and provides data query, analysis and display services.

[0030] In summary, the system includes: a unique device identification coding module to establish a full lifecycle electronic file for each device; a multimodal device status monitoring module to synchronously collect operational status data, offline detection data, and multidimensional scene tag information; a scene recognition and classification module with a built-in scene classifier to automatically classify multimodal data into multiple predefined scene categories; a differentiated intelligent diagnosis module that integrates multiple scene sub-models and cross-scene fusion and trend prediction models, calling the corresponding model according to the scene category to perform single-point health assessment and cross-scene trend prediction, and generating hierarchical early warning instructions; a maintenance process management module that automatically generates maintenance tasks in response to early warning instructions and tracks the entire process; and a unified data platform module to achieve data integration and visualization. This invention also achieves accurate assessment and trend prediction of equipment health status through scenario-based perception and cross-scenario fusion analysis, upgrading equipment management from passive response-based "post-event maintenance" to proactive prevention-based "pre-event prediction". While improving the accuracy of assessment, it also enables visualization of equipment health trends, precision of maintenance timing, and scientific management decisions, significantly extending the online life of equipment, reducing unplanned downtime, and promoting the digital and intelligent transformation of electrical equipment management.

[0031] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto; any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and its improved concept, should be covered within the scope of protection of the present invention.

Claims

1. A full lifecycle management system for electrical equipment maintenance, characterized by: include: The device unique identification coding module is used to assign and bind a unique identification code to each target device, and to establish an electronic file containing information on the entire process of the device from online, operation, maintenance to scrapping. The multimodal device status monitoring module is used to collect dynamic operating parameters of the target device in real time when it is running, and to receive offline detection parameters of the target device when it is not running. It synchronously collects and records the multi-dimensional scene label information corresponding to each set of parameters, forming multimodal data with multi-dimensional scene label information. The scene recognition and classification module has a built-in scene classifier, which is used to automatically classify multimodal data into one of a number of predefined scene categories based on scene label information; The differentiated intelligent diagnostic module integrates a large AI model containing multiple scene sub-models and a cross-scene fusion and trend prediction model. Each scene sub-model corresponds to a scene category. Based on the classified scene category, the corresponding scene sub-model is automatically called to perform single-scene evaluation of multimodal data. Based on the evaluation results of multiple single scenes, cross-scene fusion and trend prediction are performed through the cross-scene fusion and trend prediction model to generate warning instructions at different levels. The maintenance process management module is used to respond to the warning command, automatically generate maintenance tasks, and perform full-process visual tracking and recording of the online flow, execution progress and results of maintenance tasks. The maintenance records are automatically archived to the electronic archive.

2. The full lifecycle management system for electrical equipment maintenance according to claim 1, characterized in that: The dynamic operating parameters include at least electrical quantity parameters, non-electrical quantity parameters, and environmental parameters, and the offline detection parameters include at least insulation characteristic parameters, mechanical characteristic parameters, and oil analysis parameters.

3. The full lifecycle management system for electrical equipment maintenance according to claim 2, characterized in that: The multi-dimensional scene labels include at least equipment status labels, environmental condition labels, load level labels, and test condition labels. Equipment status labels include running, hot standby, cold standby, and shutdown for maintenance. Environmental condition labels include temperature range, humidity level, and weather conditions. Load level labels include no-load, light-load, rated load, and overload. Test condition labels include test voltage level, wiring method, and instrument model. Interference level labels include electromagnetic interference intensity and background noise level.

4. The full lifecycle management system for electrical equipment maintenance according to claim 3, characterized in that: The specific process of single-scenario evaluation, cross-scenario fusion, and trend prediction of multimodal data includes: S1. When receiving multimodal data with multidimensional scene labels, automatically call the corresponding scene sub-model to perform single-scene device health status assessment according to the scene category, and output a normalized single-point health index. S2. When there is cross-scenario category data for the same target device, receive multiple single-point health indices and their corresponding scenario labels of the same target device in the historical time series, input them into the cross-scenario fusion and trend prediction model, construct the time-series evolution trajectory of the device's health status, and predict the health index of future time windows based on the evolution trajectory, and draw the future health index trend. S3. Based on the current single-point health index and the predicted future health index trend, generate early warning instructions of different levels according to the preset fusion decision rules.

5. The full lifecycle management system for electrical equipment maintenance according to claim 4, characterized in that: The process of generating different levels of early warning instructions according to preset fusion decision rules includes: If the current single-point health index is less than the single-point health index threshold, an emergency repair warning command will be generated. Conversely, if the current single-point health index is higher than the single-point health index threshold, but the future health index will be lower than the cross-point health index threshold within a future time window, then a preventive maintenance early warning instruction will be generated. If the rate of decline of the health index exceeds the set lower limit, a status alert will be generated, prompting an increase in monitoring frequency.

6. The full lifecycle management system for electrical equipment maintenance according to claim 5, characterized in that: The scenario sub-model and the cross-scenario fusion and trend prediction model adopt a unified AI big model architecture, and train single-point evaluation and trend prediction capabilities simultaneously through multi-task learning.

7. The full lifecycle management system for electrical equipment maintenance according to claim 6, characterized in that: It also includes a unified data platform module, which is used to integrate and store the electronic archives, multimodal data, multidimensional scene tags, single-point health indices, evolution trajectories and early warning instructions, and to provide data query, analysis and display services.

Images