Data mining method for intelligent management and control of informatization system and informatization system

Abstract

The invention discloses a data mining method for intelligent management and control of an informatization system, which overcomes the defects in the prior art, and comprises the following steps of: 1, acquiring historical data from each node and component of the informatization system, and integrating the historical data to form structured data; 2, clustering the structured data through a clustering algorithm to form a plurality of clusters; 3, for all data in the same cluster, extracting a plurality of typical features according to the similarity of the feature values; for the typical features in the different clusters, judging whether the typical feature values have relevance or not, and if yes, establishing a data association relationship according to the relevance between the different clusters; and step 4, monitoring the data of the informatization system, if the numerical value of the data changes, estimating the change condition of the numerical values of other data having the association relationship, analyzing the change condition of the numerical values, and if the variable quantity exceeds a set threshold value, sending out early warning information to remind related personnel.

Description

Technical Field

[0001] The present invention relates to the field of intelligent data information technology, and in particular to a data mining method and an information system for intelligent management and control of an information system. Background Art

[0002] In the power grid, the information system of the power system, in short, is to use modern information technology to transform and upgrade the power system into a digital, networked and intelligent one. This process covers all key links such as power generation, transmission, distribution, power consumption and power dispatching, aiming to improve the operating efficiency, safety and reliability of the power system, while optimizing resource allocation and reducing operating costs.

[0003] However, existing information systems often rely on manually set rules and thresholds when conducting management and control, lack intelligent analysis and prediction capabilities, and have difficulty discovering hidden correlations between data. The monitoring methods are relatively simple and cannot accurately determine the causes of emergencies or predict possible failures in a timely manner. The management and control efficiency of information systems is low. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the information systems in the prior art that they often rely on manually set rules and thresholds when performing management and control, lack intelligent analysis and prediction capabilities, and have difficulty in discovering hidden correlations between data, and to provide a data mining method and information system for intelligent management and control of information systems.

[0005] The purpose of the present invention is achieved through the following technical solutions: The data mining method for intelligent management and control of information systems includes the following steps: Step 1: Collect historical data from each node and component of the information system and integrate the historical data to form structured data; Step 2: cluster the structured data using a clustering algorithm to form several clusters; Step 3: for all data in the same cluster, extract several typical features based on the similarity of feature values; for typical features in different clusters, determine whether the typical feature values ​​are correlated. If so, establish the correlation relationship of data based on the correlation between different clusters; Step 4: Monitor the data in the information system. If the value of the data changes, estimate the change in the values ​​of other related data and analyze the change in the values. If the change exceeds the set threshold, issue an early warning to alert relevant personnel.

[0006] The design of this solution uses clustering algorithms to classify a large amount of data in the information system into several clusters. The data in each cluster has a high similarity, so it is only necessary to analyze the typical features within a cluster. Therefore, the clustering algorithm simplifies the amount of data analysis, and can also find the typical features of similar data, which helps to discover the relationship between different clusters. By establishing the relationship between data, it is possible to more accurately predict how the values ​​of other related data points will change when the value of a certain data point changes. This design can better detect faults and more effectively allocate the resources of the information system.

[0007] Preferably, in step 3, for all structured data in the same cluster, several typical features are extracted according to the similarity of feature values, specifically: All eigenvalues ​​of all structured data in the same cluster are obtained, and eigenvalues ​​contained in all structured data are selected as eigenvalues ​​for similarity judgment. Among the eigenvalues ​​for similarity judgment, the similarity is judged according to the fluctuation range or discrete degree of the eigenvalue. If the fluctuation range is within the set range or the discrete degree is less than the set discrete threshold, the eigenvalue is judged to have similarity and is used as a typical feature.

[0008] Preferably, if the number of typical features is greater than a set value N, the feature values ​​are sorted according to the fluctuation range or the discreteness, and the first N feature values ​​with the smallest fluctuation range or the first N feature values ​​with the smallest discreteness are selected as typical features.

[0009] Preferably, in step 3, for typical features in different clusters, it is determined whether typical feature values ​​are correlated, specifically: For the typical features in different clusters, different typical feature data sets are constructed, and the data sets are preprocessed to unify the format and dimension; Select an association rule mining algorithm. The association rule mining algorithm takes the typical feature data set as input and the association rules as output. The association rules are used to determine whether different typical feature data sets are associated with each other. If association rules are generated between different typical feature data sets, then it is determined that different typical feature data sets are associated with each other through the association rules.

[0010] Preferably, the association rule algorithm is specifically an Apriori algorithm, and the mining method of the association rules output by the Apriori algorithm is: Through iterative calculation, all frequent item sets that meet the minimum support threshold are found. For each frequent item set, all possible non-empty subsets are generated, and each subset is used as the antecedent of the rule, and the remaining part is used as the consequent of the rule to form association rules. The confidence of each rule is calculated, and the association rules with confidence not less than the minimum confidence are screened out. These rules are strong association rules. If there are strong association rules, it is judged that different typical feature data sets are associated through the association rule.

[0011] Preferably, in step 1, the clustering algorithm is a K-means algorithm or a DBSCAN algorithm.

[0012] Information systems, including: The data acquisition unit collects historical data from various nodes and components of the information system and integrates the data to form structured data; Data monitoring unit, which monitors the data of the information system; The data analysis unit is used to find the correlation between data. When the data monitored by the data monitoring unit changes, it estimates the change of the values ​​of other data with correlation and analyzes the change of the values. If the change exceeds the set threshold, it issues an early warning message to remind relevant personnel. The system executes data mining methods for intelligent management and control of information systems during operation.

[0013] The beneficial effects of the present invention are as follows: this solution integrates the historical data of the information system, applies clustering algorithms for intelligent classification, extracts typical features and establishes correlations between data. When changes in the monitored data values ​​are detected, it can predict the changes in the values ​​of other data with correlations, and issue early warning information when the change exceeds the set threshold, thereby significantly improving the timeliness and accuracy of management and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0015] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.

[0016] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0017] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0018] Example: Data mining methods for intelligent management and control of information systems, such as Figure 1 As shown, the following steps are included: Step 1: Collect historical data from each node and component of the information system and integrate the historical data to form structured data; Step 2: cluster the structured data using a clustering algorithm to form several clusters; Step 3: for all data in the same cluster, extract several typical features based on the similarity of feature values; for typical features in different clusters, determine whether the typical feature values ​​are correlated. If so, establish the correlation relationship of data based on the correlation between different clusters; Step 4: Monitor the data in the information system. If the value of the data changes, estimate the change in the values ​​of other related data and analyze the change in the values. If the change exceeds the set threshold, issue an early warning to alert relevant personnel.

[0019] In this embodiment, a clustering algorithm can be used to classify a large amount of data in an information system into several clusters. The data in each cluster has a high similarity, so it is only necessary to analyze the typical features in one cluster. Therefore, the clustering algorithm simplifies the amount of data to be analyzed, and can also find the typical features of similar data, which helps to discover the association between different clusters. By establishing the association between data, it is possible to more accurately predict how the values ​​of other related data points will change when the value of a certain data point changes. Such a design can better detect faults and more effectively allocate the resources of the information system.

[0020] Specifically, for example, in the power system, parameters such as power generation, load demand, grid voltage and current are interrelated. Through this solution, historical data can be collected from individual nodes (such as power stations, substations, distribution networks, etc.) and components (such as generators, transformers, transmission lines, etc.), and clusters representing different load levels, power generation status or grid operation conditions are formed after cluster analysis. Further analysis of the correlation of typical characteristics within and between clusters may reveal that peak load demand is often accompanied by a decrease in grid voltage and an increase in power generation. Based on this correlation, when the load demand begins to rise, the system can more accurately predict the trend of grid voltage changes and the required increase in power generation. This predictive capability helps the power dispatching department to adjust the power generation plan in advance, ensure the stable operation of the power grid, and avoid voltage instability or insufficient power supply.

[0021] For example, in the operation of the power system, equipment failures or abnormal conditions often trigger a series of chain reactions. By using data mining methods and establishing associations between equipment or parameters, the system can accurately identify which equipment or parameter changes are interrelated. Taking the transformer in the substation as an example, when its temperature rises abnormally, the system can quickly check the grid parameters associated with the transformer, such as voltage and current, while analyzing other equipment states, such as the load conditions of adjacent lines, and retrieving historical fault records for comparison. Through this comprehensive comparative analysis, the system can quickly and accurately locate the possible causes of the fault, such as overload operation, cooling system failure, or insulation aging. In addition, the system can also predict the possible spread and impact of the fault based on the established associations, so as to take corresponding measures in time to isolate the fault point and effectively prevent the fault from further expanding. This fast and accurate fault detection capability not only helps to significantly reduce power outage time, but also greatly improves the reliability and safety of the power system.

[0022] In step 3, for all structured data in the same cluster, several typical features are extracted according to the similarity of feature values, specifically: All eigenvalues ​​of all structured data in the same cluster are obtained, and eigenvalues ​​contained in all structured data are selected as eigenvalues ​​for similarity judgment. Among the eigenvalues ​​for similarity judgment, the similarity is judged according to the fluctuation range or discrete degree of the eigenvalue. If the fluctuation range is within the set range or the discrete degree is less than the set discrete threshold, the eigenvalue is judged to have similarity and is used as a typical feature.

[0023] In this scheme, it is considered that small fluctuation range or small dispersion degree indicates that the characteristic values ​​of the data in this cluster are relatively stable and can represent the typical characteristics of this cluster. If the fluctuation range is large or the dispersion degree is large, the error will be large when correlating the characteristics of the data in other clusters. The dispersion degree can be calculated by variance, standard deviation or interquartile range, and can be flexibly adjusted according to different data in actual application.

[0024] If the number of typical features is greater than the set value N, the eigenvalues ​​are sorted according to the fluctuation range or the degree of dispersion, and the first N eigenvalues ​​with the smallest fluctuation range or the first N eigenvalues ​​with the smallest degree of dispersion are selected as typical features.

[0025] In step 3, for typical features in different clusters, it is determined whether typical feature values ​​are correlated, specifically: For the typical features in different clusters, different typical feature data sets are constructed, and the data sets are preprocessed to unify the format and dimension; Select an association rule mining algorithm. The association rule mining algorithm takes the typical feature data set as input and the association rules as output. The association rules are used to determine whether different typical feature data sets are associated with each other. If association rules are generated between different typical feature data sets, then it is determined that different typical feature data sets are associated with each other through the association rules.

[0026] The association rule algorithm is specifically the Apriori algorithm. The Apriori algorithm outputs the mining method of the association rules as follows: through iterative calculation, all frequent item sets that meet the minimum support threshold are found, and for each frequent item set, all possible non-empty subsets are generated, and each subset is used as the antecedent of the rule, and the remaining part is used as the consequent of the rule to form association rules, and the confidence of each rule is calculated, and the association rules whose confidence is not less than the minimum confidence are screened out. These rules are strong association rules. If there are strong association rules, it is determined that different typical feature data sets are associated through the association rule.

[0027] The Apriori algorithm is an association rule mining algorithm, first proposed by Agrawal, Srikant and others in 1993. It is mainly used to discover frequent patterns between item sets in a database and the association rules between them. The Apriori algorithm is based on a priori property: all non-empty subsets of frequent item sets must also be frequent. The algorithm uses an iterative method of layer-by-layer search, first finding the frequent 1-item set, then using the frequent 1-item set to find the frequent 2-item set, and so on, until no more frequent item sets can be found. The frequent item sets of each layer are generated based on the results of the previous layer. The specific application of the Apriori algorithm in this embodiment can effectively determine whether the typical eigenvalues ​​in different clusters are correlated, and provide strong support for the intelligent management and control of information systems.

[0028] In the step 1, the clustering algorithm is a K-means algorithm or a DBSCAN algorithm.

[0029] Information systems, including: The data acquisition unit collects historical data from various nodes and components of the information system and integrates the data to form structured data; Data monitoring unit, which monitors the data of the information system; The data analysis unit is used to find the correlation between data. When the data monitored by the data monitoring unit changes, it estimates the change of the values ​​of other data with correlation and analyzes the change of the values. If the change exceeds the set threshold, it issues an early warning message to remind relevant personnel. The system executes the data mining method for intelligent management and control of information systems when running.

[0030] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.

[0031] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A data mining method for intelligent management and control of information systems, characterized in that: The following steps are involved: Step 1: Collect historical data from each node and component of the information system, and integrate the historical data to form structured data; Step 2: cluster the structured data using a clustering algorithm to form several clusters; Step 3: for all data in the same cluster, extract several typical features based on the similarity of feature values; for typical features in different clusters, determine whether the typical feature values ​​are correlated. If so, establish the correlation relationship of data based on the correlation between different clusters; Step 4: Monitor the data in the information system. If the value of the data changes, estimate the change in the values ​​of other related data and analyze the change in the values. If the change exceeds the set threshold, issue an early warning to alert relevant personnel.

2. The data mining method for intelligent management and control of information systems according to claim 1 is characterized in that: In the step 3, for all structured data in the same cluster, a number of typical features are extracted according to the similarity of the feature values, specifically: all feature values ​​of all structured data in the same cluster are obtained, and the feature values ​​contained in all structured data are selected as feature values ​​for similarity judgment. Among the feature values ​​for similarity judgment, the similarity is judged according to the fluctuation range or discreteness of the feature values. If the fluctuation range is within the set range or the discreteness is less than the set discrete threshold, the feature value is judged to have similarity, and the feature value is used as a typical feature.

3. The data mining method for intelligent management and control of information systems according to claim 2 is characterized in that: If the number of typical features is greater than the set value N, the eigenvalues ​​are sorted according to the fluctuation range or the degree of dispersion, and the first N eigenvalues ​​with the smallest fluctuation range or the first N eigenvalues ​​with the smallest degree of dispersion are selected as typical features.

4. The data mining method for intelligent management and control of an information system according to any one of claims 1 to 3, characterized in that: In step 3, for typical features in different clusters, it is determined whether typical feature values ​​are correlated, specifically: For the typical features in different clusters, different typical feature data sets are constructed, and the data sets are preprocessed to unify the format and dimension; Select an association rule mining algorithm. The association rule mining algorithm takes the typical feature data set as input and the association rules as output. The association rules are used to determine whether different typical feature data sets are associated with each other. If association rules are generated between different typical feature data sets, then it is determined that different typical feature data sets are associated with each other through the association rules.

5. The data mining method for intelligent management and control of information systems according to claim 4 is characterized in that: The association rule algorithm is specifically the Apriori algorithm, and the mining method of the association rules output by the Apriori algorithm is: Through iterative calculation, all frequent item sets that meet the minimum support threshold are found. For each frequent item set, all possible non-empty subsets are generated, and each subset is used as the antecedent of the rule, and the remaining part is used as the consequent of the rule to form association rules. The confidence of each rule is calculated, and the association rules with confidence not less than the minimum confidence are screened out. These rules are strong association rules. If there are strong association rules, it is judged that different typical feature data sets are associated through the association rule.

6. The data mining method for intelligent management and control of information systems according to claim 1 is characterized in that: In the step 1, the clustering algorithm is a K-means algorithm or a DBSCAN algorithm.

7. An information system, characterized by including: The data acquisition unit collects historical data from various nodes and components of the information system and integrates the data to form structured data; Data monitoring unit, which monitors the data of the information system; The data analysis unit is used to find the correlation between data. When the data monitored by the data monitoring unit changes, it estimates the change of the values ​​of other data with correlation and analyzes the change of the values. If the change exceeds the set threshold, it issues an early warning message to remind relevant personnel. The system executes the data mining method for intelligent management and control of information systems as described in claim 1 when running.

Images