Multi-source data analysis and early warning method, system and medium based on intelligent agent graph

Abstract

The present application provides a multi-source data analysis and early warning method, system and medium based on an agent graph. The method includes: obtaining historical multi-source data by pre-processing collected data from different data sources within multiple historical preset time periods, and then obtaining agent information associated with the historical multi-source data, and extracting agent graph generation data, generating an agent graph, matching and analyzing the relationship data between agents in the agent graph, obtaining fusion data corresponding to agents with associated relationships, and extracting key feature data for training, obtaining an early warning model, and then obtaining real-time multi-source data within a preset time period through the agent graph and the early warning model to obtain corresponding early warning label data, and responding to the early warning; the present application realizes the analysis and early warning of real-time multi-source data by processing historical multi-source data, generating an agent graph and constructing an early warning model, and improves the intelligence and accuracy of the early warning.

Description

Technical Field

[0001] The present application relates to the field of data analysis and early warning technology, and more specifically, to a multi-source data analysis and early warning method, system and medium based on intelligent agent graphs. Background Art

[0002] In today's information age, data sources are increasingly extensive and complex, including sensor data, Internet data, business system data, etc. Traditional data analysis and early warning methods often find it difficult to handle the complex correlations and semantic differences between data, resulting in a significant reduction in the accuracy, timeliness and intelligence of early warnings. How to effectively integrate these multi-source data, mine valuable information, and issue early warnings in time to respond to various potential risks or opportunities has become a key issue faced by many fields; agent graphs, as an emerging technical means, can well describe the complex relationships between agents and between agents and the environment, providing a new solution for multi-source data analysis and early warning.

[0003] In view of the above problems, effective technical solutions are urgently needed. Summary of the invention

[0004] The purpose of this application is to provide a multi-source data analysis and early warning method, system and medium based on an intelligent agent graph, which can generate an intelligent agent graph and build an early warning model by processing historical multi-source data, thereby realizing analysis and early warning of real-time multi-source data and improving the intelligence and accuracy of the early warning.

[0005] The present application also provides a multi-source data analysis and early warning method based on an agent graph, comprising the following steps:

[0006] Acquire the collected data from different data sources within multiple historical preset time periods, and perform preprocessing to obtain historical multi-source data;

[0007] Processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data based on the agent information, and processing the graph generation data to obtain an agent graph;

[0008] Perform matching analysis according to the agent graph, obtain the data source of the to-be-fused data corresponding to the agents with associated relationships, and process the data source of the to-be-fused data to obtain fused data;

[0009] Extract key feature data according to the fused data, perform training according to the key feature data, and obtain an early warning model;

[0010] Real-time multi-source data within a preset time period is obtained for processing, and processed through the intelligent agent map and early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data.

[0011] Optionally, in the multi-source data analysis and early warning method based on the agent graph described in the present application, the acquisition of collected data from different data sources within multiple historical preset time periods and preprocessing to obtain historical multi-source data includes:

[0012] Acquire the collected data from different data sources within multiple historical preset time periods, and perform cleaning, denoising and format unification preprocessing to obtain preprocessed collected data;

[0013] Perform integrity check, accuracy check and availability check on the preprocessed collected data. The integrity check includes record integrity check and scope integrity check. The accuracy check includes historical mean deviation check and industry mean deviation check. The availability check includes data format check and data timeliness check.

[0014] If the integrity check, accuracy check and availability check are all passed, the preprocessed collected data is recorded as historical multi-source data;

[0015] If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected.

[0016] Optionally, in the multi-source data analysis and early warning method based on the agent graph described in the present application, the processing according to the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data according to the agent information, and processing according to the graph generation data to obtain the agent graph, including:

[0017] Processing the historical multi-source data to obtain agent information associated with the historical multi-source data;

[0018] Extracting graph generation data of intelligent agents according to the intelligent agent information, including identification data and attribute data of intelligent agents and relationship data between intelligent agents;

[0019] The relationship data between the agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship;

[0020] The identification data and attribute data of the intelligent agent are processed in combination with the relationship data between the intelligent agents to obtain an intelligent agent graph.

[0021] Optionally, in the multi-source data analysis and early warning method based on the agent graph described in the present application, performing matching analysis according to the agent graph, obtaining the data source of the data to be fused corresponding to the agents with associated relationships, and processing according to the data source of the data to be fused to obtain fused data include:

[0022] Perform matching analysis according to the agent graph to obtain the data source of the to-be-fused data corresponding to the agents with associated relationships;

[0023] According to the data source of the data to be fused, semantic conversion and unified processing are performed on the data to be fused to obtain data with the same semantics;

[0024] The data to be fused with the same semantics are fused through a preset data fusion algorithm to obtain fused data.

[0025] Optionally, in the multi-source data analysis and early warning method based on the agent graph described in the present application, extracting key feature data according to the fused data, training according to the key feature data, and obtaining an early warning model include:

[0026] Extracting key feature data according to the fused data, including a prediction feature data set and corresponding historical warning label data;

[0027] The prediction feature data set includes statistical feature data, time series feature data and associated feature data;

[0028] The initial prediction model is trained according to the statistical feature data, the time series feature data, the associated feature data and the corresponding historical warning label data to obtain a trained warning model.

[0029] Optionally, in the multi-source data analysis and early warning method based on the agent graph described in the present application, the real-time multi-source data within a preset time period is processed, and processed through the agent graph and the early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data, including:

[0030] Obtain real-time multi-source data within a preset time period;

[0031] Performing preprocessing on the real-time multi-source data to obtain real-time optimized multi-source data;

[0032] According to the real-time optimization multi-source data, the data is processed through the intelligent agent graph to obtain real-time fusion data, and a real-time prediction feature data set is extracted, including real-time statistical feature data, real-time time series feature data and real-time correlation feature data;

[0033] Input the real-time statistical feature data, real-time time series feature data and real-time correlation feature data into the early warning model for processing to obtain corresponding real-time early warning label data, including low risk, medium risk or high risk;

[0034] Respond to warnings based on real-time warning label data.

[0035] In a second aspect, the present application provides a multi-source data analysis and early warning system based on an agent graph, the system comprising: a memory and a processor, the memory comprising a program of a multi-source data analysis and early warning method based on an agent graph, and the program of the multi-source data analysis and early warning method based on an agent graph is executed by the processor to implement the following steps:

[0036] Acquire the collected data from different data sources within multiple historical preset time periods, and perform preprocessing to obtain historical multi-source data;

[0037] Processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data based on the agent information, and processing the graph generation data to obtain an agent graph;

[0038] Perform matching analysis according to the agent graph, obtain the data source of the to-be-fused data corresponding to the agents with associated relationships, and process the data source of the to-be-fused data to obtain fused data;

[0039] Extract key feature data according to the fused data, perform training according to the key feature data, and obtain an early warning model;

[0040] Real-time multi-source data within a preset time period is obtained for processing, and processed through the intelligent agent map and early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data.

[0041] Optionally, in the multi-source data analysis and early warning system based on the agent graph described in the present application, the acquisition of collected data from different data sources within multiple historical preset time periods and preprocessing to obtain historical multi-source data includes:

[0042] Acquire the collected data from different data sources within multiple historical preset time periods, and perform cleaning, denoising and format unification preprocessing to obtain preprocessed collected data;

[0043] Perform integrity check, accuracy check and availability check on the preprocessed collected data. The integrity check includes record integrity check and scope integrity check. The accuracy check includes historical mean deviation check and industry mean deviation check. The availability check includes data format check and data timeliness check.

[0044] If the integrity check, accuracy check and availability check are all passed, the preprocessed collected data is recorded as historical multi-source data;

[0045] If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected.

[0046] Optionally, in the multi-source data analysis and early warning system based on the agent graph described in the present application, the processing according to the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data according to the agent information, and processing according to the graph generation data to obtain the agent graph, includes:

[0047] Processing the historical multi-source data to obtain agent information associated with the historical multi-source data;

[0048] Extracting graph generation data of intelligent agents according to the intelligent agent information, including identification data and attribute data of intelligent agents and relationship data between intelligent agents;

[0049] The relationship data between the agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship;

[0050] The identification data and attribute data of the intelligent agent are processed in combination with the relationship data between the intelligent agents to obtain an intelligent agent graph.

[0051] In the third aspect, the present application also provides a computer-readable storage medium, which stores a multi-source data analysis and early warning method program based on an intelligent agent graph. When the multi-source data analysis and early warning method program based on an intelligent agent graph is executed by a processor, the steps of the multi-source data analysis and early warning method based on an intelligent agent graph as described in any one of the above items are implemented.

[0052] From the above, it can be seen that the multi-source data analysis and early warning method, system and medium based on the intelligent agent graph provided in this application, by processing historical multi-source data, generate an intelligent agent graph and construct a warning model, realize the analysis and early warning of real-time multi-source data, and improve the intelligence and accuracy of the warning.

[0053] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or understood by practicing the embodiments of the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 A flowchart of a multi-source data analysis and early warning method based on an agent graph provided in an embodiment of the present application;

[0056] Figure 2 A flowchart of obtaining an agent map according to a multi-source data analysis and early warning method based on an agent map provided in an embodiment of the present application;

[0057] Figure 3 A flowchart of obtaining a warning model for a multi-source data analysis and warning method based on an agent graph provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0059] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0060] Please refer to Figure 1 , Figure 1 1 is a flow chart of a multi-source data analysis and early warning method based on an agent graph in some embodiments of the present application. The multi-source data analysis and early warning method based on an agent graph is used in a terminal device, such as a computer, a mobile phone terminal, etc. The multi-source data analysis and early warning method based on an agent graph includes the following steps:

[0061] S11, acquiring collected data from different data sources within multiple historical preset time periods, and performing preprocessing to obtain historical multi-source data;

[0062] S12, processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data according to the agent information, and processing the graph generation data to obtain an agent graph;

[0063] S13, performing matching analysis according to the agent graph, obtaining the data source of the to-be-fused data corresponding to the agents with associated relationships, and processing according to the data source of the to-be-fused data to obtain fused data;

[0064] S14, extracting key feature data according to the fused data, and performing training according to the key feature data to obtain an early warning model;

[0065] S15. Acquire real-time multi-source data within a preset time period for processing, and process it through the intelligent agent map and early warning model to obtain corresponding early warning label data, and make an early warning response according to the early warning label data.

[0066] It should be noted that in order to improve the accuracy and timeliness of multi-source data analysis and early warning, we first obtain collected data from different data sources in multiple different historical time periods, and perform cleaning, noise reduction, formatting and verification to obtain historical multi-source data. We then analyze and process the historical multi-source data, build a corresponding intelligent agent graph, perform matching analysis based on the intelligent agent graph, perform semantic conversion and unified processing to obtain fused data, and then train to obtain the corresponding early warning model. Finally, we obtain real-time multi-source data, perform preprocessing and verification, and process it through the intelligent agent graph and early warning model to obtain the corresponding early warning label data, and then respond to the early warning based on the early warning label data.

[0067] According to an embodiment of the present invention, the acquisition of collected data from different data sources within multiple historical preset time periods and preprocessing to obtain historical multi-source data includes:

[0068] Acquire the collected data from different data sources within multiple historical preset time periods, and perform cleaning, denoising and format unification preprocessing to obtain preprocessed collected data;

[0069] Perform integrity check, accuracy check and availability check on the preprocessed collected data. The integrity check includes record integrity check and scope integrity check. The accuracy check includes historical mean deviation check and industry mean deviation check. The availability check includes data format check and data timeliness check.

[0070] If the integrity check, accuracy check and availability check are all passed, the preprocessed collected data is recorded as historical multi-source data;

[0071] If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected.

[0072] It should be noted that in order to accurately generate the intelligent agent map and build the early warning model, the collected data collected from multiple data sources are first cleaned, denoised and pre-processed in a unified format to obtain the pre-processed collected data, and integrity checks, accuracy checks and availability checks are performed to ensure the quality of the pre-processed collected data. If the integrity check, accuracy check and availability check are all passed, it means that the quality of the pre-processed collected data meets the requirements, and the pre-processed collected data is recorded as historical multi-source data. If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected, and then the corresponding check is performed. If it passes, it is recorded as historical multi-source data. If it fails, the corresponding pre-processed collected data is discarded.

[0073] Please refer to Figure 2 , Figure 2 The flowchart of the method for multi-source data analysis and early warning based on agent map in some embodiments of the present application is to obtain an agent map. According to the embodiment of the present invention, the method of processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent map generation data according to the agent information, and processing the map generation data to obtain the agent map includes:

[0074] S21, processing the historical multi-source data to obtain agent information associated with the historical multi-source data;

[0075] S22, extracting the graph of the agent based on the agent information to generate data, including identification data and attribute data of the agent and relationship data between agents;

[0076] S23, the relationship data between the agents includes data transmission relationship, dependency relationship, collaboration relationship or causal relationship;

[0077] S24. Process the identification data and attribute data of the intelligent agent in combination with the relationship data between the intelligent agents to obtain an intelligent agent graph.

[0078] It should be noted that in order to obtain the agent map, firstly, an in-depth analysis is performed based on the historical multi-source data obtained to determine the agent information associated with the data, and then the map generation data including the agent's identification data and attribute data and the relationship data between agents are extracted. The agent's identification data is used to represent the agent's role, and the attribute data refers to the agent's name, type, and field. The relationship data between agents is obtained by analyzing the data flow, business logic, and semantic information in the multi-source data. The relationship data between agents include data transmission relationships, dependency relationships, collaborative relationships, or causal relationships, such as the data transmission relationship between sensors transmitting data to data processing centers, the dependency relationship between production equipment and energy supply equipment, the collaborative relationship between doctors and nurses in the medical process, and the causal relationship between market demand changes leading to enterprise production adjustments; based on the determined agent's identification data and attribute data combined with the relationship data between the agents, processing is performed to obtain the agent map, and a graph database or a special graph construction tool is used to store the agent map.

[0079] According to an embodiment of the present invention, performing matching analysis according to the agent graph, obtaining a data source of the data to be fused corresponding to the agents having an associated relationship, and performing processing according to the data source of the data to be fused to obtain fused data include:

[0080] Perform matching analysis according to the agent graph to obtain the data source of the to-be-fused data corresponding to the agents with associated relationships;

[0081] According to the data source of the data to be fused, semantic conversion and unified processing are performed on the data to be fused to obtain data with the same semantics;

[0082] The data to be fused with the same semantics are fused through a preset data fusion algorithm to obtain fused data.

[0083] It should be noted that, according to the relationship between the agents in the agent map, the source of the data to be fused is determined, and the data corresponding to the agents with associated relationships are matched to prepare for subsequent fusion. Since multi-source data may come from different systems or devices, their data semantics are different. Using the pre-set domain ontology library, the data of each data source is mapped and converted according to the concepts and relationships in the ontology library, so that the data can understand and interact with each other at the semantic level, and the data with the same semantics to be fused is obtained. Finally, the data with the same semantics to be fused is fused through the preset data fusion algorithm to obtain fused data. For example, when fusing the same physical quantity data measured by multiple sensors, different weights are assigned to them according to factors such as the accuracy and reliability of the sensors, and then weighted average calculation is performed to obtain the fused temperature data. For non-numerical data, such as text data or classified data, fuzzy logic fusion method, evidence theory fusion method, etc. can be used. In the analysis of public opinion on social media, for text comments on an event from different sources, the fuzzy logic fusion method is used to comprehensively consider factors such as the emotional tendency and credibility of the comments, and a comprehensive public opinion evaluation result about the event is obtained.

[0084] Please refer to Figure 3 , Figure 3 The flowchart of the early warning model of the multi-source data analysis early warning method based on the agent map in some embodiments of the present application. According to the embodiment of the present invention, extracting key feature data according to the fusion data, training according to the key feature data, and obtaining the early warning model include:

[0085] S31, extracting key feature data according to the fused data, including a prediction feature data set and corresponding historical warning label data;

[0086] S32, the prediction feature data set includes statistical feature data, time series feature data and associated feature data;

[0087] S33: Train the initial prediction model according to the statistical feature data, time series feature data, associated feature data and corresponding historical warning label data to obtain a trained warning model.

[0088] It should be noted that in order to train and obtain the early warning model, key feature data is extracted from the fused data, including a prediction feature data set and corresponding historical warning label data. The prediction feature data set includes statistical feature data, time series feature data and associated feature data. The statistical feature data includes the mean and variance of the data. The time series feature data refers to the changing trend of the data, and the associated feature data refers to the correlation between two variables. The initial prediction model is trained based on the statistical feature data, time series feature data, associated feature data and the corresponding historical warning label data to obtain a trained early warning model.

[0089] According to an embodiment of the present invention, the real-time multi-source data within a preset time period is processed, and processed through the intelligent agent map and the early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data, including:

[0090] Obtain real-time multi-source data within a preset time period;

[0091] Performing preprocessing on the real-time multi-source data to obtain real-time optimized multi-source data;

[0092] According to the real-time optimization multi-source data, the data is processed through the intelligent agent graph to obtain real-time fusion data, and a real-time prediction feature data set is extracted, including real-time statistical feature data, real-time time series feature data and real-time correlation feature data;

[0093] Input the real-time statistical feature data, real-time time series feature data and real-time correlation feature data into the early warning model for processing to obtain corresponding real-time early warning label data, including low risk, medium risk or high risk;

[0094] Respond to warnings based on real-time warning label data.

[0095] It should be noted that in order to realize the analysis and early warning of real-time multi-source data, the real-time multi-source data within a preset time period is first obtained, and the real-time multi-source data is preprocessed and verified to obtain real-time optimized multi-source data, which is processed in combination with the generated intelligent body map to obtain real-time fusion data, and the real-time prediction feature data set including real-time statistical feature data, real-time time series feature data and real-time correlation feature data is extracted. Finally, the real-time statistical feature data, real-time time series feature data and real-time correlation feature data are input into the early warning model for processing to obtain the corresponding real-time warning label data, and a corresponding level of early warning response is performed according to the real-time warning label data. For example, in the intelligent transportation system, when it is monitored that the road traffic volume increases abnormally and the vehicle speed drops sharply, the early warning model issues a low-risk, medium-risk or high-risk congestion risk level warning, and the system sends a warning information to the traffic management department and displays the congestion prompt information on the road display screen.

[0096] It is worth mentioning that according to an embodiment of the present invention, it also includes:

[0097] Obtain the warning accuracy rate, false alarm rate, missed alarm rate and predicted response time data within the preset time period;

[0098] Processing is performed according to the warning accuracy rate, false alarm rate and missed alarm rate to obtain a warning accuracy evaluation index of the warning model;

[0099] The warning accuracy evaluation index is processed in combination with the predicted response time data to obtain a warning timeliness index of the warning model;

[0100] Comparing the warning timeliness index with a preset benchmark timeliness index to obtain a warning timeliness deviation rate;

[0101] Comparing the warning timeliness deviation rate with a preset warning timeliness deviation rate threshold;

[0102] If it is less than or equal to the preset warning timeliness deviation rate threshold, the warning of the warning model is determined to be valid;

[0103] If it is greater than the preset warning timeliness deviation rate threshold, it is determined that the warning timeliness of the warning model is insufficient, and the warning model is optimized.

[0104] It should be noted that due to the real-time variability of multi-source data, it is necessary to evaluate the accuracy of the prediction model and the early warning efficiency in a timely manner. First, the early warning accuracy rate, false alarm rate, missed alarm rate and prediction response time data within the preset time period are obtained, and the early warning accuracy evaluation index of the early warning model is obtained according to the early warning accuracy rate, false alarm rate and missed alarm rate.

[0105] The calculation formula of the early warning accuracy evaluation index is:

[0106] ;

[0107] in, is the early warning accuracy evaluation index, , , are the warning accuracy rate, false alarm rate and missed alarm rate, respectively. , is a preset characteristic coefficient (the characteristic coefficient is obtained by querying a preset multi-source data analysis early warning database);

[0108] Then the obtained warning accuracy evaluation index is combined with the predicted response time data to obtain the warning timeliness index of the warning model;

[0109] The calculation formula of the early warning timeliness index is:

[0110] ;

[0111] in, is the warning timeliness index, , They are the warning accuracy evaluation index and the predicted response time data, is a preset characteristic coefficient (the characteristic coefficient is obtained by querying a preset multi-source data analysis early warning database);

[0112] The obtained warning timeliness index is compared with the preset benchmark timeliness index to obtain the warning timeliness deviation rate, which refers to the ratio of the absolute value of the difference between the warning timeliness index and the preset benchmark timeliness index to the preset benchmark timeliness index;

[0113] Finally, the obtained warning time deviation rate is compared with the preset validity threshold. In this embodiment, the validity threshold is set to (0, 0.2], (0.2, 1], corresponding to effective warning and insufficient warning time. For example, if the obtained warning time deviation rate is 0.1, it means that the warning ability is good, and the warning of the warning model is determined to be effective. If the obtained warning time deviation rate is 0.3, it means that the warning deviation is large, then the warning time of the warning model is determined to be insufficient, and the warning model is optimized.

[0114] It is worth mentioning that according to an embodiment of the present invention, it also includes:

[0115] Obtaining information contribution rate data corresponding to the statistical feature data, time series feature data and associated feature data;

[0116] Comparing the information contribution rate data with a preset contribution benchmark threshold;

[0117] If it is less than the preset contribution reference threshold, the corresponding feature data is determined to be invalid and the corresponding feature data is discarded;

[0118] If it is greater than or equal to the preset contribution reference threshold, the corresponding feature data is determined to be valid.

[0119] It should be noted that after evaluating and monitoring the early warning model, if optimization is required, first determine whether the feature data of the training model matches. Technical personnel in this field can calculate the information contribution rate data corresponding to the statistical feature data, time series feature data and associated feature data through a method based on statistical indicators, and compare the information contribution rate data with the preset contribution benchmark threshold. In this embodiment, the preset contribution benchmark threshold is set to 0.1. If the obtained information contribution rate data is 0.05, which is less than the preset contribution benchmark threshold, the corresponding feature data is determined to be invalid and the corresponding feature data is eliminated. If the obtained information contribution rate data is 0.15, which is greater than the preset contribution benchmark threshold, the corresponding feature data is determined to be valid. The effectiveness of the early warning model is guaranteed by timely evaluation of the feature data.

[0120] The present invention also discloses a multi-source data analysis and early warning system based on an intelligent agent graph, comprising a memory and a processor, wherein the memory comprises a multi-source data analysis and early warning method program based on an intelligent agent graph, and when the multi-source data analysis and early warning method program based on an intelligent agent graph is executed by the processor, the following steps are implemented:

[0121] Acquire the collected data from different data sources within multiple historical preset time periods, and perform preprocessing to obtain historical multi-source data;

[0122] Processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data based on the agent information, and processing the graph generation data to obtain an agent graph;

[0123] Perform matching analysis according to the agent graph, obtain the data source of the to-be-fused data corresponding to the agents with associated relationships, and process the data source of the to-be-fused data to obtain fused data;

[0124] Extract key feature data according to the fused data, perform training according to the key feature data, and obtain an early warning model;

[0125] Real-time multi-source data within a preset time period is obtained for processing, and processed through the intelligent agent map and early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data.

[0126] It should be noted that in order to improve the accuracy and timeliness of multi-source data analysis and early warning, we first obtain collected data from different data sources in multiple different historical time periods, and perform cleaning, noise reduction, formatting and verification to obtain historical multi-source data. We then analyze and process the historical multi-source data, build a corresponding intelligent agent graph, perform matching analysis based on the intelligent agent graph, perform semantic conversion and unified processing to obtain fused data, and then train to obtain the corresponding early warning model. Finally, we obtain real-time multi-source data, perform preprocessing and verification, and process it through the intelligent agent graph and early warning model to obtain the corresponding early warning label data, and then respond to the early warning based on the early warning label data.

[0127] According to an embodiment of the present invention, the acquisition of collected data from different data sources within multiple historical preset time periods and preprocessing to obtain historical multi-source data includes:

[0128] Acquire the collected data from different data sources within multiple historical preset time periods, and perform cleaning, denoising and format unification preprocessing to obtain preprocessed collected data;

[0129] Perform integrity check, accuracy check and availability check on the preprocessed collected data. The integrity check includes record integrity check and scope integrity check. The accuracy check includes historical mean deviation check and industry mean deviation check. The availability check includes data format check and data timeliness check.

[0130] If the integrity check, accuracy check and availability check are all passed, the preprocessed collected data is recorded as historical multi-source data;

[0131] If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected.

[0132] It should be noted that in order to accurately generate the intelligent agent map and build the early warning model, the collected data collected from multiple data sources are first cleaned, denoised and pre-processed in a unified format to obtain the pre-processed collected data, and integrity checks, accuracy checks and availability checks are performed to ensure the quality of the pre-processed collected data. If the integrity check, accuracy check and availability check are all passed, it means that the quality of the pre-processed collected data meets the requirements, and the pre-processed collected data is recorded as historical multi-source data. If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected, and then the corresponding check is performed. If it passes, it is recorded as historical multi-source data. If it fails, the corresponding pre-processed collected data is discarded.

[0133] According to an embodiment of the present invention, the processing according to the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data according to the agent information, and processing according to the graph generation data to obtain an agent graph, includes:

[0134] Processing the historical multi-source data to obtain agent information associated with the historical multi-source data;

[0135] Extracting graph generation data of intelligent agents according to the intelligent agent information, including identification data and attribute data of intelligent agents and relationship data between intelligent agents;

[0136] The relationship data between the agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship;

[0137] The identification data and attribute data of the intelligent agent are processed in combination with the relationship data between the intelligent agents to obtain an intelligent agent graph.

[0138] It should be noted that in order to obtain the agent map, firstly, an in-depth analysis is performed based on the historical multi-source data obtained to determine the agent information associated with the data, and then the map generation data including the agent's identification data and attribute data and the relationship data between agents are extracted. The agent's identification data is used to represent the agent's role, and the attribute data refers to the agent's name, type, and field. The relationship data between agents is obtained by analyzing the data flow, business logic, and semantic information in the multi-source data. The relationship data between agents include data transmission relationships, dependency relationships, collaborative relationships, or causal relationships, such as the data transmission relationship between sensors transmitting data to data processing centers, the dependency relationship between production equipment and energy supply equipment, the collaborative relationship between doctors and nurses in the medical process, and the causal relationship between market demand changes leading to enterprise production adjustments; based on the determined agent's identification data and attribute data combined with the relationship data between the agents, processing is performed to obtain the agent map, and a graph database or a special graph construction tool is used to store the agent map.

[0139] According to an embodiment of the present invention, performing matching analysis according to the agent graph, obtaining a data source of the data to be fused corresponding to the agents having an associated relationship, and performing processing according to the data source of the data to be fused to obtain fused data include:

[0140] Perform matching analysis according to the agent graph to obtain the data source of the to-be-fused data corresponding to the agents with associated relationships;

[0141] According to the data source of the data to be fused, semantic conversion and unified processing are performed on the data to be fused to obtain data with the same semantics;

[0142] The data to be fused with the same semantics are fused through a preset data fusion algorithm to obtain fused data.

[0143] It should be noted that, according to the relationship between the agents in the agent map, the source of the data to be fused is determined, and the data corresponding to the agents with associated relationships are matched to prepare for subsequent fusion. Since multi-source data may come from different systems or devices, their data semantics are different. Using the pre-set domain ontology library, the data of each data source is mapped and converted according to the concepts and relationships in the ontology library, so that the data can understand and interact with each other at the semantic level, and the data with the same semantics to be fused is obtained. Finally, the data with the same semantics to be fused is fused through the preset data fusion algorithm to obtain fused data. For example, when fusing the same physical quantity data measured by multiple sensors, different weights are assigned to them according to factors such as the accuracy and reliability of the sensors, and then weighted average calculation is performed to obtain the fused temperature data. For non-numerical data, such as text data or classified data, fuzzy logic fusion method, evidence theory fusion method, etc. can be used. In the analysis of public opinion on social media, for text comments on an event from different sources, the fuzzy logic fusion method is used to comprehensively consider factors such as the emotional tendency and credibility of the comments, and a comprehensive public opinion evaluation result about the event is obtained.

[0144] According to an embodiment of the present invention, extracting key feature data according to the fused data, performing training according to the key feature data, and obtaining an early warning model includes:

[0145] Extracting key feature data according to the fused data, including a prediction feature data set and corresponding historical warning label data;

[0146] The prediction feature data set includes statistical feature data, time series feature data and associated feature data;

[0147] The initial prediction model is trained according to the statistical feature data, the time series feature data, the associated feature data and the corresponding historical warning label data to obtain a trained warning model.

[0148] It should be noted that in order to train and obtain the early warning model, key feature data is extracted from the fused data, including a prediction feature data set and corresponding historical warning label data. The prediction feature data set includes statistical feature data, time series feature data and associated feature data. The statistical feature data includes the mean and variance of the data. The time series feature data refers to the changing trend of the data, and the associated feature data refers to the correlation between two variables. The initial prediction model is trained based on the statistical feature data, time series feature data, associated feature data and the corresponding historical warning label data to obtain a trained early warning model.

[0149] According to an embodiment of the present invention, the real-time multi-source data within a preset time period is processed, and processed through the intelligent agent map and the early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data, including:

[0150] Obtain real-time multi-source data within a preset time period;

[0151] Performing preprocessing on the real-time multi-source data to obtain real-time optimized multi-source data;

[0152] According to the real-time optimization multi-source data, the data is processed through the intelligent agent graph to obtain real-time fusion data, and a real-time prediction feature data set is extracted, including real-time statistical feature data, real-time time series feature data and real-time correlation feature data;

[0153] Input the real-time statistical feature data, real-time time series feature data and real-time correlation feature data into the early warning model for processing to obtain corresponding real-time early warning label data, including low risk, medium risk or high risk;

[0154] Respond to warnings based on real-time warning label data.

[0155] It should be noted that in order to realize the analysis and early warning of real-time multi-source data, the real-time multi-source data within a preset time period is first obtained, and the real-time multi-source data is preprocessed and verified to obtain real-time optimized multi-source data, which is processed in combination with the generated intelligent body map to obtain real-time fusion data, and the real-time prediction feature data set including real-time statistical feature data, real-time time series feature data and real-time correlation feature data is extracted. Finally, the real-time statistical feature data, real-time time series feature data and real-time correlation feature data are input into the early warning model for processing to obtain the corresponding real-time warning label data, and a corresponding level of early warning response is performed according to the real-time warning label data. For example, in the intelligent transportation system, when it is monitored that the road traffic volume increases abnormally and the vehicle speed drops sharply, the early warning model issues a low-risk, medium-risk or high-risk congestion risk level warning, and the system sends a warning information to the traffic management department and displays the congestion prompt information on the road display screen.

[0156] It is worth mentioning that according to an embodiment of the present invention, it also includes:

[0157] Obtain the warning accuracy rate, false alarm rate, missed alarm rate and predicted response time data within the preset time period;

[0158] Processing is performed according to the warning accuracy rate, false alarm rate and missed alarm rate to obtain a warning accuracy evaluation index of the warning model;

[0159] The warning accuracy evaluation index is processed in combination with the predicted response time data to obtain a warning timeliness index of the warning model;

[0160] Comparing the warning timeliness index with a preset benchmark timeliness index to obtain a warning timeliness deviation rate;

[0161] Comparing the warning timeliness deviation rate with a preset warning timeliness deviation rate threshold;

[0162] If it is less than or equal to the preset warning timeliness deviation rate threshold, the warning of the warning model is determined to be valid;

[0163] If it is greater than the preset warning timeliness deviation rate threshold, it is determined that the warning timeliness of the warning model is insufficient, and the warning model is optimized.

[0164] It should be noted that due to the real-time variability of multi-source data, it is necessary to evaluate the accuracy of the prediction model and the early warning efficiency in a timely manner. First, the early warning accuracy rate, false alarm rate, missed alarm rate and prediction response time data within the preset time period are obtained, and the early warning accuracy evaluation index of the early warning model is obtained according to the early warning accuracy rate, false alarm rate and missed alarm rate.

[0165] The calculation formula of the early warning accuracy evaluation index is:

[0166] ;

[0167] in, is the early warning accuracy evaluation index, , , are the warning accuracy rate, false alarm rate and missed alarm rate, respectively. , is a preset characteristic coefficient (the characteristic coefficient is obtained by querying a preset multi-source data analysis early warning database);

[0168] Then the obtained warning accuracy evaluation index is combined with the predicted response time data to obtain the warning timeliness index of the warning model;

[0169] The calculation formula of the early warning timeliness index is:

[0170] ;

[0171] in, is the warning timeliness index, , They are the warning accuracy evaluation index and the predicted response time data, is a preset characteristic coefficient (the characteristic coefficient is obtained by querying a preset multi-source data analysis early warning database);

[0172] The obtained warning timeliness index is compared with the preset benchmark timeliness index to obtain the warning timeliness deviation rate, which refers to the ratio of the absolute value of the difference between the warning timeliness index and the preset benchmark timeliness index to the preset benchmark timeliness index;

[0173] Finally, the obtained warning time deviation rate is compared with the preset validity threshold. In this embodiment, the validity threshold is set to (0, 0.2], (0.2, 1], corresponding to effective warning and insufficient warning time. For example, if the obtained warning time deviation rate is 0.1, it means that the warning ability is good, and the warning of the warning model is determined to be effective. If the obtained warning time deviation rate is 0.3, it means that the warning deviation is large, then the warning time of the warning model is determined to be insufficient, and the warning model is optimized.

[0174] It is worth mentioning that according to an embodiment of the present invention, it also includes:

[0175] Obtaining information contribution rate data corresponding to the statistical feature data, time series feature data and associated feature data;

[0176] Comparing the information contribution rate data with a preset contribution benchmark threshold;

[0177] If it is less than the preset contribution reference threshold, the corresponding feature data is determined to be invalid and the corresponding feature data is discarded;

[0178] If it is greater than or equal to the preset contribution reference threshold, the corresponding feature data is determined to be valid.

[0179] It should be noted that after evaluating and monitoring the early warning model, if optimization is required, first determine whether the feature data of the training model matches. Technical personnel in this field can calculate the information contribution rate data corresponding to the statistical feature data, time series feature data and associated feature data through a method based on statistical indicators, and compare the information contribution rate data with the preset contribution benchmark threshold. In this embodiment, the preset contribution benchmark threshold is set to 0.1. If the obtained information contribution rate data is 0.05, which is less than the preset contribution benchmark threshold, the corresponding feature data is determined to be invalid and the corresponding feature data is eliminated. If the obtained information contribution rate data is 0.15, which is greater than the preset contribution benchmark threshold, the corresponding feature data is determined to be valid. The effectiveness of the early warning model is guaranteed by timely evaluation of the feature data.

[0180] The third aspect of the present invention provides a readable storage medium, which stores a multi-source data analysis and early warning method program based on an intelligent agent graph. When the multi-source data analysis and early warning method program based on an intelligent agent graph is executed by a processor, the steps of the multi-source data analysis and early warning method based on an intelligent agent graph as described in any one of the above items are implemented.

[0181] The multi-source data analysis and early warning method, system and medium based on the intelligent agent graph disclosed in the present invention process historical multi-source data, generate an intelligent agent graph and build an early warning model, thereby realizing analysis and early warning of real-time multi-source data and improving the intelligence and accuracy of the early warning.

[0182] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0183] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0184] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0185] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, the aforementioned program can be stored in a readable storage medium, and when the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0186] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

Claims

1. A multi-source data analysis and early warning method based on agent graph, characterized in that: The following steps are involved: Acquire the collected data from different data sources within multiple historical preset time periods, and perform preprocessing to obtain historical multi-source data; Processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data based on the agent information, and processing the graph generation data to obtain an agent graph; Perform matching analysis according to the agent graph, obtain the data source of the to-be-fused data corresponding to the agents with associated relationships, and process the data source of the to-be-fused data to obtain fused data; Extract key feature data according to the fused data, perform training according to the key feature data, and obtain an early warning model; Acquire real-time multi-source data within a preset time period for processing, and process it through the intelligent agent map and early warning model to obtain corresponding early warning label data, and make an early warning response according to the early warning label data; Extracting key feature data according to the fused data, and training according to the key feature data to obtain an early warning model, including: Extracting key feature data according to the fused data, including a prediction feature data set and corresponding historical warning label data; The prediction feature data set includes statistical feature data, time series feature data and associated feature data; The initial prediction model is trained according to the statistical feature data, the time series feature data, the associated feature data and the corresponding historical warning label data to obtain a trained warning model; Also includes: Obtain the warning accuracy rate, false alarm rate, missed alarm rate and predicted response time data within the preset time period; Processing is performed according to the warning accuracy rate, false alarm rate and missed alarm rate to obtain a warning accuracy evaluation index of the warning model; The warning accuracy evaluation index is processed in combination with the predicted response time data to obtain a warning timeliness index of the warning model; Comparing the warning timeliness index with a preset benchmark timeliness index to obtain a warning timeliness deviation rate; Comparing the warning timeliness deviation rate with a preset warning timeliness deviation rate threshold; If it is less than or equal to the preset warning timeliness deviation rate threshold, the warning of the warning model is determined to be valid; If it is greater than the preset warning timeliness deviation rate threshold, it is determined that the warning timeliness of the warning model is insufficient, and the warning model is optimized; The calculation formula of the early warning accuracy evaluation index is: ； in, is the early warning accuracy evaluation index, , , are the warning accuracy rate, false alarm rate and missed alarm rate, respectively. , is the preset characteristic coefficient; The calculation formula of the early warning timeliness index is: ； in, is the warning timeliness index, , They are the warning accuracy evaluation index and the predicted response time data, is the preset characteristic coefficient; Extract graph generation data including identification data and attribute data of intelligent agents and relationship data between intelligent agents. The identification data of intelligent agents is used to indicate the role of intelligent agents. The attribute data refers to the name, type and field of intelligent agents. The relationship data between intelligent agents is obtained by analyzing the data flow, business logic and semantic information in multi-source data. The relationship data between intelligent agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship, such as the data transmission relationship of sensors transmitting data to data processing centers, the dependency relationship of production equipment on energy supply equipment, the collaboration relationship of doctors and nurses in the medical process, and the causal relationship of enterprise production adjustment caused by changes in market demand; Based on the identification data and attribute data of the determined intelligent agent, combined with the relationship data between the intelligent agents, processing is performed to obtain an intelligent agent graph; Multi-source data comes from different systems or devices. Using the pre-set domain ontology library, the data from each data source is mapped and converted according to the concepts and relationships in the ontology library.

2. The multi-source data analysis and early warning method based on agent graph according to claim 1 is characterized in that: The acquisition of collected data from different data sources within multiple historical preset time periods and preprocessing to obtain historical multi-source data includes: Acquire the collected data from different data sources within multiple historical preset time periods, and perform cleaning, denoising and format unification preprocessing to obtain preprocessed collected data; Perform integrity check, accuracy check and availability check on the preprocessed collected data. The integrity check includes record integrity check and scope integrity check. The accuracy check includes historical mean deviation check and industry mean deviation check. The availability check includes data format check and data timeliness check. If the integrity check, accuracy check and availability check are all passed, the preprocessed collected data is recorded as historical multi-source data; If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected.

3. The multi-source data analysis and early warning method based on agent graph according to claim 2 is characterized in that: The processing according to the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data according to the agent information, and processing according to the graph generation data to obtain an agent graph, including: Processing the historical multi-source data to obtain agent information associated with the historical multi-source data; Extracting graph generation data of intelligent agents according to the intelligent agent information, including identification data and attribute data of intelligent agents and relationship data between intelligent agents; The relationship data between the agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship; The identification data and attribute data of the intelligent agent are processed in combination with the relationship data between the intelligent agents to obtain an intelligent agent graph.

4. The multi-source data analysis and early warning method based on agent graph according to claim 3 is characterized in that: The matching analysis is performed according to the agent graph to obtain the data source of the to-be-fused data corresponding to the agents with associated relationships, and the fused data is obtained by processing according to the data source of the to-be-fused data, including: Perform matching analysis according to the agent graph to obtain the data source of the to-be-fused data corresponding to the agents with associated relationships; According to the data source of the data to be fused, semantic conversion and unified processing are performed on the data to be fused to obtain data with the same semantics; The data to be fused with the same semantics are fused through a preset data fusion algorithm to obtain fused data.

5. The multi-source data analysis and early warning method based on agent graph according to claim 4 is characterized in that: The real-time multi-source data acquired within a preset time period is processed, and processed through the intelligent agent map and the early warning model to obtain corresponding early warning label data, and an early warning response is performed according to the early warning label data, including: Obtain real-time multi-source data within a preset time period; Performing preprocessing on the real-time multi-source data to obtain real-time optimized multi-source data; According to the real-time optimization multi-source data, the data is processed through the intelligent agent graph to obtain real-time fusion data, and a real-time prediction feature data set is extracted, including real-time statistical feature data, real-time time series feature data and real-time correlation feature data; Input the real-time statistical feature data, real-time time series feature data and real-time correlation feature data into the early warning model for processing to obtain corresponding real-time early warning label data, including low risk, medium risk or high risk; Respond to warnings based on real-time warning label data.

6. The multi-source data analysis and early warning system based on the agent graph is characterized by: The method comprises a memory and a processor, wherein the memory comprises a program of a multi-source data analysis and early warning method based on an intelligent agent graph, and the program of the multi-source data analysis and early warning method based on an intelligent agent graph is executed by the processor to implement the following steps: Acquire the collected data from different data sources within multiple historical preset time periods, and perform preprocessing to obtain historical multi-source data; Processing the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data based on the agent information, and processing the graph generation data to obtain an agent graph; Perform matching analysis according to the agent graph, obtain the data source of the to-be-fused data corresponding to the agents with associated relationships, and process the data source of the to-be-fused data to obtain fused data; Extract key feature data according to the fused data, perform training according to the key feature data, and obtain an early warning model; Acquire real-time multi-source data within a preset time period for processing, and process it through the intelligent agent map and early warning model to obtain corresponding early warning label data, and make an early warning response according to the early warning label data; Extracting key feature data according to the fused data, and training according to the key feature data to obtain an early warning model, including: Extracting key feature data according to the fused data, including a prediction feature data set and corresponding historical warning label data; The prediction feature data set includes statistical feature data, time series feature data and associated feature data; The initial prediction model is trained according to the statistical feature data, the time series feature data, the associated feature data and the corresponding historical warning label data to obtain a trained warning model; Also includes: Obtain the warning accuracy rate, false alarm rate, missed alarm rate and predicted response time data within the preset time period; Processing is performed according to the warning accuracy rate, false alarm rate and missed alarm rate to obtain a warning accuracy evaluation index of the warning model; The warning accuracy evaluation index is processed in combination with the predicted response time data to obtain a warning timeliness index of the warning model; Comparing the warning timeliness index with a preset benchmark timeliness index to obtain a warning timeliness deviation rate; Comparing the warning timeliness deviation rate with a preset warning timeliness deviation rate threshold; If it is less than or equal to the preset warning timeliness deviation rate threshold, the warning of the warning model is determined to be valid; If it is greater than the preset warning timeliness deviation rate threshold, it is determined that the warning timeliness of the warning model is insufficient, and the warning model is optimized; The calculation formula of the early warning accuracy evaluation index is: ； in, is the early warning accuracy evaluation index, , , are the warning accuracy rate, false alarm rate and missed alarm rate, respectively. , is the preset characteristic coefficient; The calculation formula of the early warning timeliness index is: ； in, is the warning timeliness index, , They are the warning accuracy evaluation index and the predicted response time data, is the preset characteristic coefficient; Extract graph generation data including identification data and attribute data of intelligent agents and relationship data between intelligent agents. The identification data of intelligent agents is used to indicate the role of intelligent agents. The attribute data refers to the name, type and field of intelligent agents. The relationship data between intelligent agents is obtained by analyzing the data flow, business logic and semantic information in multi-source data. The relationship data between intelligent agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship, such as the data transmission relationship of sensors transmitting data to data processing centers, the dependency relationship of production equipment on energy supply equipment, the collaboration relationship of doctors and nurses in the medical process, and the causal relationship of enterprise production adjustment caused by changes in market demand; Based on the identification data and attribute data of the determined intelligent agent, combined with the relationship data between the intelligent agents, processing is performed to obtain an intelligent agent graph; Multi-source data comes from different systems or devices. Using the pre-set domain ontology library, the data from each data source is mapped and converted according to the concepts and relationships in the ontology library.

7. The multi-source data analysis and early warning system based on agent graph according to claim 6 is characterized in that: The acquisition of collected data from different data sources within multiple historical preset time periods and preprocessing to obtain historical multi-source data includes: Acquire the collected data from different data sources within multiple historical preset time periods, and perform cleaning, denoising and format unification preprocessing to obtain preprocessed collected data; Perform integrity check, accuracy check and availability check on the preprocessed collected data. The integrity check includes record integrity check and scope integrity check. The accuracy check includes historical mean deviation check and industry mean deviation check. The availability check includes data format check and data timeliness check. If the integrity check, accuracy check and availability check are all passed, the preprocessed collected data is recorded as historical multi-source data; If one or more of the integrity check, accuracy check and availability check fails, the corresponding pre-processed collected data is marked and corrected.

8. The multi-source data analysis and early warning system based on agent graph according to claim 7 is characterized in that: The processing according to the historical multi-source data to obtain agent information associated with the historical multi-source data, extracting agent graph generation data according to the agent information, and processing according to the graph generation data to obtain an agent graph, including: Processing the historical multi-source data to obtain agent information associated with the historical multi-source data; Extracting graph generation data of intelligent agents according to the intelligent agent information, including identification data and attribute data of intelligent agents and relationship data between intelligent agents; The relationship data between the agents include data transmission relationship, dependency relationship, collaboration relationship or causal relationship; The identification data and attribute data of the intelligent agent are processed in combination with the relationship data between the intelligent agents to obtain an intelligent agent graph.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a multi-source data analysis and early warning method program based on the intelligent agent graph. When the multi-source data analysis and early warning method program based on the intelligent agent graph is executed by the processor, the steps of the multi-source data analysis and early warning method based on the intelligent agent graph as described in any one of claims 1 to 5 are implemented.

Images