Operation scene behavior pattern recognition method, system and storage medium based on big data analysis

Abstract

The application relates to the technical field of operation behavior analysis, and discloses an operation scene behavior mode recognition method and system based on big data analysis and a storage medium. Multidimensional data of a user in an operation scene, such as login behavior data, page access data, user preference data and the like, is collected, and is divided into types in combination with preset behavior mode division types; behavior mode recognition of respective dimensions is carried out based on a recognition model matched by each behavior mode division type; finally, the final behavior mode of the user in the operation scene is recognized in combination with behavior mode recognition results corresponding to each behavior mode division type, the accuracy of behavior mode recognition is improved, full-link tracking of user behavior is realized, and a fine decision basis is provided for operation decision-making, such as customized scene pushing for different behavior mode users; and the resource allocation accuracy of resource configuration is improved by dynamically adjusting system resources based on the behavior mode recognition result.

Description

Technical Field

[0001] This invention relates to the field of operational behavior analysis technology, and in particular to a method, system, and storage medium for identifying operational scenario behavior patterns based on big data analysis. Background Technology

[0002] With the rapid development of internet technology, user behavior data in operational scenarios has experienced explosive growth. Therefore, accurate identification of user behavior data is key to precise and efficient operations.

[0003] However, practice has shown that traditional user behavior analysis methods rely heavily on manual statistics, making it difficult to process massive amounts of user behavior data, resulting in low identification efficiency and accuracy. While existing data analysis tools have improved identification efficiency to some extent compared to manual methods, they are limited by single-dimensional analysis and struggle to accurately identify dynamic changes and abnormal features in user behavior patterns, leading to a lack of in-depth data support for operational decisions. For example, in core operational scenarios such as user login behavior and scenario access preferences, existing methods cannot capture key behaviors such as high-frequency access and abnormal logins in real time, making it difficult to identify user behavior patterns and resulting in a lack of effective data support for operational decisions.

[0004] Therefore, it is particularly important to propose a new method for identifying behavioral patterns in operational scenarios to improve the accuracy of identifying these patterns and thus provide a precise basis for operational decision-making. Summary of the Invention

[0005] This invention provides a method, system, and storage medium for identifying operational scenario behavior patterns based on big data analysis, which can improve the accuracy of identifying operational scenario behavior patterns, thereby providing accurate operational decision-making basis.

[0006] The first aspect of this invention discloses a method for recognizing operational scenario behavior patterns based on big data analysis, the method comprising: Collect operational datasets generated by target users performing operational operations in the operational scenario within a target time period; Based on multiple predetermined behavioral pattern classification types, the operational dataset is divided to obtain multidimensional operational data corresponding to each behavioral pattern classification type. Based on each behavior pattern classification type, obtain the behavior pattern recognition model that matches the behavior pattern classification type; Based on the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to the behavior pattern classification type is identified to obtain the behavior pattern recognition result of the behavior pattern classification type. Based on the behavior pattern recognition results of all the aforementioned behavior pattern classifications, the target user's behavior pattern is identified to obtain the target user's behavior pattern recognition results in the operational scenario.

[0007] As an optional implementation, in a first aspect of the present invention, the method further includes: Based on the classification of all the aforementioned behavioral patterns, determine whether there is a correlation between the behavioral pattern recognition results obtained by analyzing the multi-dimensional operational data for each of them; When a correlation is found between the results, all the behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all the behavior pattern classification types. The step of identifying the multidimensional operational data corresponding to each behavior pattern classification type using a behavior pattern recognition model to obtain the behavior pattern recognition result for that behavior pattern classification type includes: Based on the ranking analysis results, and using the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type.

[0008] As an optional implementation, in the first aspect of the present invention, the step of identifying the multidimensional operational data corresponding to each behavior pattern classification type based on the ranking analysis results and the behavior pattern recognition model corresponding to each behavior pattern classification type to obtain the behavior pattern recognition result of that behavior pattern classification type includes: Based on the ranking analysis results, the target behavior pattern classification type required for this identification is selected from all the behavior pattern classification types. Then, based on the behavior pattern recognition model corresponding to the target behavior pattern classification type, the multidimensional operational data corresponding to the target behavior pattern classification type is identified to obtain the behavior pattern recognition result for that target behavior pattern classification type. The process of selecting the target behavior pattern classification type from all the behavior pattern classification types based on the ranking analysis results and identifying the multidimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result for that target behavior pattern classification type is repeated. Specifically, the behavior pattern recognition result of the previous target behavior pattern classification type is updated to the multidimensional operational data corresponding to the current target behavior pattern classification type.

[0009] As an optional implementation, in the first aspect of the present invention, before the step of re-executing the step of selecting the target behavior pattern classification type required for this identification from all the behavior pattern classification types based on the sorting analysis results, and identifying the multidimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result of the target behavior pattern classification type, the method further includes: Determine whether the behavior pattern recognition result of the target behavior pattern classification type is a preset abnormal behavior pattern recognition result; When it is determined that the result is not the preset abnormal behavior pattern identification result, the operation of filtering out the target behavior pattern classification type required for this identification from all the behavior pattern classification types according to the sorting analysis result is re-executed, and the multi-dimensional operational data corresponding to the target behavior pattern classification type is identified based on the behavior pattern identification model corresponding to the target behavior pattern classification type to obtain the behavior pattern identification result of the target behavior pattern classification type is obtained. The method further includes: When the result is determined to be the preset abnormal behavior pattern identification result, for any behavior pattern classification type for which no identification operation has been performed, the multidimensional operational data corresponding to the behavior pattern classification type is identified based on the behavior pattern identification model corresponding to the behavior pattern classification type, and the behavior pattern identification result of the behavior pattern classification type is obtained.

[0010] As an optional implementation, in the first aspect of the present invention, the step of determining whether there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data of each behavior pattern based on the classification of all the aforementioned behavior patterns includes: Determine whether there are corresponding analysis order labels for all the aforementioned behavior pattern classification types; When it is determined that all the behavior pattern classification types have corresponding analysis order labels, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing their respective multi-dimensional operational data; The step of analyzing and ranking all the behavior pattern classification types to obtain the ranking analysis results corresponding to all the behavior pattern classification types includes: Based on the analysis order label corresponding to each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

[0011] As an optional implementation, in a first aspect of the present invention, the method further includes: When it is determined that there is no corresponding analysis order label for any of the behavior pattern classification types, the historical behavior pattern recognition results and historical model input parameters corresponding to each behavior pattern classification type are obtained. Based on the historical behavior pattern recognition results and historical model input parameters corresponding to all the behavior pattern classification types, determine whether there is a behavior pattern classification type among all the behavior pattern classification types whose historical behavior pattern recognition results are used as historical model input parameters for other behavior pattern classification types. When the result is determined to be yes, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data of each of them; The step of analyzing and ranking all the behavior pattern classification types to obtain the ranking analysis results corresponding to all the behavior pattern classification types includes: Based on the historical model input parameters of each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

[0012] As an optional implementation, in the first aspect of the present invention, all the behavior pattern classification types include user login behavior pattern type, page access behavior pattern type, and user preference behavior pattern type; When there is a correlation between the behavior pattern recognition results corresponding to the user login behavior pattern type, the page access behavior pattern type, and the user preference behavior pattern type, the ranking analysis result is used to indicate that the user login behavior pattern type, the page access behavior pattern type, and the user preference behavior pattern type are analyzed sequentially.

[0013] A second aspect of this invention discloses an operational scenario behavior pattern recognition device based on big data analysis, the device comprising: The data collection module is used to collect operational datasets generated by target users performing operational operations in the operational scenario within a target time period. The segmentation module is used to perform a segmentation operation on the operational dataset according to multiple pre-determined behavioral patterns to obtain multi-dimensional operational data corresponding to each behavioral pattern segmentation type. The acquisition module is used to classify each behavior pattern into a type and acquire a behavior pattern recognition model that matches the behavior pattern classification type. The identification module is used to identify the multi-dimensional operational data corresponding to each behavior pattern classification type based on the behavior pattern identification model corresponding to each behavior pattern classification type, and obtain the behavior pattern identification result of the behavior pattern classification type. The identification module is further configured to perform behavior pattern identification on the target user based on the behavior pattern identification results of all the behavior pattern classifications, and obtain the behavior pattern identification results of the target user in the operation scenario.

[0014] As an optional implementation, in a second aspect of the invention, the apparatus further includes: The judgment module is used to classify all the behavior patterns into types and determine whether there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data of each of them. The sorting module is used to analyze and sort all the behavior pattern classification types when it is determined that there is a correlation between the results, and to obtain the sorting analysis results corresponding to all the behavior pattern classification types. The specific method by which the recognition module identifies the multidimensional operational data corresponding to each behavior pattern classification type based on the behavior pattern recognition model for that behavior pattern classification type, and obtains the behavior pattern recognition result for that behavior pattern classification type, includes: Based on the ranking analysis results, and using the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type.

[0015] As an optional implementation, in a second aspect of the present invention, the specific method by which the identification module identifies the multidimensional operational data corresponding to each behavior pattern classification type based on the ranking analysis result and the behavior pattern identification model corresponding to each behavior pattern classification type, to obtain the behavior pattern identification result of that behavior pattern classification type, includes: Based on the ranking analysis results, the target behavior pattern classification type required for this identification is selected from all the behavior pattern classification types. Then, based on the behavior pattern recognition model corresponding to the target behavior pattern classification type, the multidimensional operational data corresponding to the target behavior pattern classification type is identified to obtain the behavior pattern recognition result for that target behavior pattern classification type. The process of selecting the target behavior pattern classification type from all the behavior pattern classification types based on the ranking analysis results and identifying the multidimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result for that target behavior pattern classification type is repeated. Specifically, the behavior pattern recognition result of the previous target behavior pattern classification type is updated to the multidimensional operational data corresponding to the current target behavior pattern classification type.

[0016] As an optional implementation, in a second aspect of the present invention, the judgment module is further configured to, before the identification module re-executes the operation of selecting the target behavior pattern classification type required for this identification from all the behavior pattern classification types according to the sorting analysis results, and identifying the multi-dimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern identification model corresponding to the target behavior pattern classification type to obtain the behavior pattern identification result of the target behavior pattern classification type, determine whether the behavior pattern identification result of the target behavior pattern classification type is a preset abnormal behavior pattern identification result; when it is determined that it is not the preset abnormal behavior pattern identification result, trigger the identification module to re-execute the operation of selecting the target behavior pattern classification type required for this identification from all the behavior pattern classification types according to the sorting analysis results, and identifying the multi-dimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern identification model corresponding to the target behavior pattern classification type to obtain the behavior pattern identification result of the target behavior pattern classification type; The identification module is further configured to, when it is determined that the result is the preset abnormal behavior pattern identification result, identify the multi-dimensional operational data corresponding to any behavior pattern classification type for which no identification operation has been performed, based on the behavior pattern identification model corresponding to the behavior pattern classification type, and obtain the behavior pattern identification result of the behavior pattern classification type.

[0017] As an optional implementation, in a second aspect of the invention, the specific method by which the judging module determines whether there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data of all the behavior patterns according to the classification of all the behavior patterns includes: Determine whether there are corresponding analysis order labels for all the aforementioned behavior pattern classification types; When it is determined that all the behavior pattern classification types have corresponding analysis order labels, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing their respective multi-dimensional operational data; The specific method by which the sorting module analyzes and sorts all the behavior pattern classification types to obtain the sorting analysis results corresponding to all the behavior pattern classification types includes: Based on the analysis order label corresponding to each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

[0018] As an optional implementation, in a second aspect of the invention, the specific method by which the judging module classifies all the behavioral patterns and judges whether there is a correlation between the behavioral pattern recognition results obtained by analyzing the multi-dimensional operational data for each of them further includes: When it is determined that there is no corresponding analysis order label for any of the behavior pattern classification types, the historical behavior pattern recognition results and historical model input parameters corresponding to each behavior pattern classification type are obtained. Based on the historical behavior pattern recognition results and historical model input parameters corresponding to all the behavior pattern classification types, determine whether there is a behavior pattern classification type among all the behavior pattern classification types whose historical behavior pattern recognition results are used as historical model input parameters for other behavior pattern classification types. When the result is determined to be yes, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data of each of them; The specific method by which the sorting module analyzes and sorts all the behavior pattern classification types to obtain the sorting analysis results corresponding to all the behavior pattern classification types includes: Based on the historical model input parameters of each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

[0019] As an optional implementation, in the second aspect of the present invention, all the behavior pattern classification types include user login behavior pattern type, page access behavior pattern type, and user preference behavior pattern type; When there is a correlation between the behavior pattern recognition results corresponding to the user login behavior pattern type, the page access behavior pattern type, and the user preference behavior pattern type, the ranking analysis result is used to indicate that the user login behavior pattern type, the page access behavior pattern type, and the user preference behavior pattern type are analyzed sequentially.

[0020] A third aspect of this invention discloses an operational scenario behavior pattern recognition device based on big data analysis, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in any of the methods described in the first aspect of the present invention.

[0021] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in any of the methods described in the first aspect of the present invention.

[0022] Compared with the prior art, the present invention has the following beneficial effects: In this embodiment of the invention, an operational dataset is collected, generated by target users performing operational operations within a target time period in an operational scenario. Based on multiple predetermined behavioral pattern classifications, the operational dataset is divided to obtain multi-dimensional operational data corresponding to each behavioral pattern classification. For each behavioral pattern classification, a behavioral pattern recognition model matching that classification is obtained. Based on the behavioral pattern recognition model corresponding to each classification, the multi-dimensional operational data corresponding to that classification is recognized to obtain the behavioral pattern recognition result for that classification. Based on the behavioral pattern recognition results for all classifications, behavioral pattern recognition is performed on the target user to obtain the target user's behavioral pattern recognition result in the operational scenario. As can be seen, implementing this invention involves collecting multi-dimensional user data in operational scenarios, such as login behavior data, page access data, and user preference data. This data is then categorized based on preset behavior pattern classifications. Each behavior pattern classification is matched with a corresponding recognition model to identify its respective dimension. Finally, the recognition results from each behavior pattern classification are combined to identify the user's final behavior pattern in the operational scenario. This improves the accuracy of behavior pattern recognition and enables full-link tracking of user behavior, providing refined decision-making support for operational decisions. For example, customized scenarios can be pushed to users with different behavior patterns, such as scheduling monitoring and defect work orders. Furthermore, it facilitates dynamic adjustment of system resources based on behavior pattern recognition results, such as server load and bandwidth allocation, improving the accuracy of resource allocation. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a method for recognizing operational scenario behavior patterns based on big data analysis, as disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of an operational scenario behavior pattern recognition device based on big data analysis disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of another operational scenario behavior pattern recognition device based on big data analysis disclosed in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of another operational scenario behavior pattern recognition device based on big data analysis disclosed in an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will understand, explicitly and implicitly, that the embodiments described herein can be combined with other embodiments.

[0028] This invention discloses a method, system, and storage medium for identifying operational behavior patterns based on big data analysis. It collects multi-dimensional user data in operational scenarios, such as login behavior data, page access data, and user preference data, and categorizes this data according to preset behavior pattern classifications. Then, it identifies behavior patterns in each dimension based on a matching recognition model for each behavior pattern classification. Finally, it combines the behavior pattern recognition results corresponding to each behavior pattern classification to identify the user's final behavior pattern in the operational scenario. This improves the accuracy of behavior pattern recognition and enables full-link tracking of user behavior, providing refined decision-making basis for operational decisions. For example, it allows for the delivery of customized scenarios based on different user behavior patterns, such as scheduling monitoring and defect work orders. It also facilitates the dynamic adjustment of system resources based on behavior pattern recognition results, such as server load and bandwidth allocation, improving the accuracy of resource allocation. These are described in detail below.

[0029] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for recognizing operational scenario behavior patterns based on big data analysis, as disclosed in an embodiment of the present invention. Wherein, Figure 1 The described method can be applied to scenarios requiring behavioral pattern recognition, such as the operation and maintenance of power grid digital platforms. Figure 1 As shown, the method may include the following operations: 101. Collect operational datasets generated by target users performing operational operations in operational scenarios within a target time period.

[0030] In this embodiment of the invention, the target duration may optionally be one week, one month, or other durations. The operational dataset corresponding to the target user includes login medium data, login behavior data, operation behavior data, and operation result data of the target user within the target duration. Login medium data includes, but is not limited to, login method (e.g., account password login, third-party login), login time, login location, login device, and login network IP. Login behavior data includes, but is not limited to, accessed page type, access duration, and access time period. Operation behavior data includes, but is not limited to, clicks, inputs, submissions, and modifications. Operation result data includes, but is not limited to, order volume and transaction amount.

[0031] 102. Based on the predetermined multiple behavioral pattern classification types, perform a partitioning operation on the operational dataset to obtain multidimensional operational data corresponding to each behavioral pattern classification type.

[0032] In this embodiment of the invention, optionally, all behavior pattern classification types include, but are not limited to, user login behavior pattern type, page access behavior pattern type, and user preference behavior pattern type. Optionally, the multi-dimensional operational data corresponding to different behavior pattern classification types may overlap; for example, for access time periods, they may simultaneously appear in the multi-dimensional operational data corresponding to both the page access behavior pattern type and the user preference behavior pattern type.

[0033] 103. Classify each behavior pattern into categories and obtain the behavior pattern recognition model that matches the classification category of that behavior pattern.

[0034] In this embodiment of the invention, optionally, the behavior pattern recognition models corresponding to each behavior pattern classification type are different. Specifically, the behavior pattern recognition model corresponding to the user login behavior pattern type is a login behavior pattern recognition model, which is trained using the K-Means algorithm on multi-dimensional sample operational data corresponding to the user login behavior pattern type; the behavior pattern recognition model corresponding to the page access behavior pattern type is a page access behavior pattern recognition model, which is trained using the Apriori algorithm and the Hidden Markov Chain algorithm on multi-dimensional sample operational data corresponding to the page access behavior pattern type; the behavior pattern recognition model corresponding to the user preference behavior pattern type is a user preference behavior pattern recognition model, which is trained using collaborative filtering recommendation algorithms and label propagation algorithms (such as Label Propagation) on multi-dimensional sample operational data corresponding to the user preference behavior pattern type. The multi-dimensional sample operational data corresponding to each behavior pattern classification type includes multi-dimensional sample operational data from multiple sample users, as detailed in the foregoing description of the multi-dimensional operational data corresponding to each behavior pattern type.

[0035] 104. Based on the behavior pattern recognition model corresponding to each behavior pattern classification type, identify the multidimensional operational data corresponding to that behavior pattern classification type to obtain the behavior pattern recognition result for that behavior pattern classification type.

[0036] In this embodiment of the invention, the behavior pattern recognition result corresponding to each behavior pattern classification type includes the normal behavior pattern recognition result and normal situation, or the abnormal behavior pattern recognition result and abnormal situation. For example, for a user with a preference for handling work orders, if the behavior pattern recognition result is normal, then the normal situation is submitting defect work orders 10-20 times per month, downloading maintenance reports 3-5 times, and having no record of not submitting work orders for 3 consecutive days; if the behavior pattern recognition result is abnormal, then the abnormal situation is submitting defect work orders more than 50 times in a single month, far exceeding the historical average, or not submitting any work orders for 7 consecutive days.

[0037] In this embodiment of the invention, for user login behavior pattern types, multi-dimensional operational data of user login behavior patterns are identified based on a login behavior pattern recognition model to obtain corresponding login behavior recognition results, including: Based on the login behavior pattern recognition model, cluster analysis is performed on the multi-dimensional operational data corresponding to the user login behavior pattern types to obtain the number of clusters of the target user, such as login time, login location, login device, login method, etc., such as K=4; Calculate the Euclidean login distance between each login data point and its corresponding login cluster center from the target user's multi-dimensional operational data; compare the Euclidean login distance for each login data point with the corresponding preset maximum Euclidean login distance to obtain the corresponding login matching degree. Based on the login matching degree corresponding to each login data, the login behavior pattern recognition result of the target user is determined. If the login matching degree corresponding to the login location, login time, login device and login method is greater than 85%, it is considered a normal login behavior pattern recognition result; otherwise, it is considered an abnormal login behavior pattern recognition result.

[0038] In this embodiment of the invention, optionally, the login behavior pattern recognition model is trained using the K-Means algorithm based on corresponding multi-dimensional sample operational data, i.e., multi-dimensional sample login data of multiple sample users. Specifically: Collect multi-dimensional login data from multiple sample users corresponding to different user login behavior patterns, such as login time, login location, login device, and login method. Extract login features from each sample user's login data, such as mapping IP addresses to office / home / remote locations, dividing time periods into work / non-work periods, and classifying devices as PC / mobile devices. Generate corresponding login features such as login frequency and location stability based on the multi-dimensional login data of each sample user. Iteratively train the K-Means algorithm based on the login features of each sample user obtained above, and determine the optimal number of clusters using the SSE curve (Sum of Squared Errors). The number of iterations can be 100, or the model can be considered complete when the model's loss value is less than or equal to a preset loss value, thus obtaining the required login behavior pattern recognition model.

[0039] In this embodiment of the invention, for page access behavior pattern types, multi-dimensional operational data of page access behavior patterns are identified based on a page access behavior pattern recognition model to obtain corresponding page access behavior recognition results, including: Based on the page access recognition pattern recognition model, we analyze the multi-dimensional operational data corresponding to the page access behavior pattern type to obtain the page access path corresponding to the target user; we analyze the page access path corresponding to the target user to obtain the corresponding path analysis results. If the path analysis results indicate that the page access path is normal, then the support of the page access path is analyzed based on the page access path; the confidence of the page access path is analyzed based on the support and the page access path; the confidence of the page access path is compared with the preset confidence; if the confidence of the page access path is greater than the preset confidence, it indicates that the page is normally associated; the page jump situation is analyzed to obtain the page transfer rate; the page transfer rate is compared with the preset transfer rate and the page visit dwell time is compared with the preset dwell time to obtain the transfer rate comparison result and dwell time comparison result; based on the transfer rate comparison result, dwell time comparison result, and path analysis results, the page access behavior pattern recognition result of the target user in the operation scenario is determined. If the path analysis results indicate abnormal page access paths, then the support of the page access paths is analyzed based on the page access paths; the confidence of page access is analyzed based on the support and the page access paths; the confidence of page access is compared with a preset confidence level, and if the confidence level is greater than the preset confidence level, it indicates an abnormal association; the page jump situation is analyzed to obtain the page transfer rate; the page transfer rate is compared with the preset transfer rate and the number of page jumps is compared with the preset number of jumps to obtain the transfer rate comparison result and the jump number comparison result; based on the transfer rate comparison result, the jump number comparison result, and the path analysis results, the page access behavior pattern recognition result of the target user in the operational scenario is determined.

[0040] For example, analyzing user A's multi-dimensional operational data, we obtained the page access path: Homepage → Device Type → Device Monitoring → Defective Order View → Submit. Based on the Apriori algorithm, we performed correlation analysis between pages, obtaining a path support of 15% (i.e., 150 out of 1000 users adopted this path) and a confidence level of 85% for viewing defective orders details → submitting (i.e., user A submitted 85% of the time on the details page), which are high-frequency normal paths. Based on the Hidden Markov Chain algorithm, we analyzed that the transition rate from viewing defective orders details to submitting was 80% (greater than the preset transition rate of 75%, consistent with the platform's normal distribution), and the dwell time on the details page was 35 seconds (greater than the preset dwell time of 30 seconds). This indicates that user A is a goal-oriented normal user, meaning that user A's normal page access was identified.

[0041] For example, analyzing user B's multi-dimensional operational data reveals the following page access path: Homepage → Device Type → Device Monitoring → Exit → Re-login → Device Type → Defect Order View → Exit (repeated 3 times). Based on the Apriori algorithm, correlation analysis between pages shows a path support of only 0.5% (extremely low) and ≥3 category jumps → 90% confidence (greater than the preset confidence of 70%, indicating an abnormal page access path). Based on the Hidden Markov Chain algorithm, the device type to device monitoring transfer rate is 100% (greater than the preset transfer rate of 80%), the number of page jumps exceeds 5 (greater than the preset number of jumps of 2), and the single-page dwell time is 5 seconds (less than the preset dwell time of 30 seconds). Therefore, user B is identified as an abnormal user, meaning abnormal page access by user B has been detected.

[0042] In this embodiment of the invention, optionally, the page access behavior pattern recognition model is trained on the Apriori algorithm and the Hidden Markov Chain algorithm based on corresponding multi-dimensional sample operational data, that is, multi-dimensional sample page access data of multiple sample users. Specifically: Collect multi-dimensional sample page access data from multiple sample users corresponding to page access behavior pattern types, such as page ID, access time, dwell time, and operation sequence. Extract features of each sample user's operational data, such as page access frequency, dwell time percentage, and operation path length, and construct a page access matrix for each sample user. Train the Apriori algorithm based on the page access matrices of all sample users, and train the Hidden Markov Chain algorithm based on the page access matrices of all sample users. The number of iterations can be 100, or the model can be considered complete when the model's loss value is less than or equal to a preset loss value, thus obtaining the required page access behavior pattern recognition model.

[0043] In this embodiment of the invention, for user preference behavior pattern types, multi-dimensional operational data of user preference behavior patterns are identified based on a user preference behavior pattern recognition model to obtain corresponding user preference behavior recognition results, including: Based on the collaborative filtering recommendation algorithm, this study analyzes multi-dimensional operational data corresponding to user preference behavior patterns to obtain the user-item interaction matrix of the target user, such as the frequency of defect ticket submissions and maintenance report downloads. Based on the user-item interaction matrix, the cosine similarity of the target user's preferences is calculated. This cosine similarity is then compared with the pre-determined cosine similarities of other users' preferences to obtain the similarity comparison result. Based on this result, corresponding preference behavior patterns are matched to the target user. Finally, based on the tag propagation algorithm, this study analyzes multi-dimensional operational data corresponding to user preference behavior patterns to obtain the user-tag matrix of the target user. The target user's user-label bipartite graph is divided into sub-graphs, such as labels for abnormal equipment temperature and voltage fluctuations. Based on the target user's user-label bipartite graph, similar users with similar user-label bipartite graphs are identified, and their user-label bipartite graphs are obtained. The target user's user-label bipartite graph and the similar users' user-label bipartite graphs are analyzed to obtain label analysis results. Based on the label analysis results, the missing labels of the target user are propagated from the similar users' user-label bipartite graphs to obtain the target user-label bipartite graph of the target user. Based on the target user-label bipartite graph and the target user's corresponding preference behavior patterns, the target user's user preference behavior patterns are determined.

[0044] For example, Xiaoming submits defect work orders 18 times a month on the platform, downloads maintenance reports only 2 times, and never uses the device alarm subscription function. Based on a collaborative filtering recommendation algorithm, this data is analyzed to construct Xiaoming's user-item interaction matrix, where rows represent Xiaoming, columns represent functions, and values ​​represent usage frequency. Based on Xiaoming's user-item interaction matrix, Xiaoming's preference cosine similarity is calculated and compared with the preference cosine similarities of Xiaohong and Xiaowang, respectively. The analysis reveals that Xiaowang and Xiaoming have a higher similarity (0.9), both frequently submitting work orders. Xiaowang uses the device alarm subscription function 15 times a month, thus matching Xiaoming with the device alarm subscription function. The platform has device anomaly tags such as voltage fluctuation and abnormal device temperature. Xiaoming only clicked on the voltage fluctuation tag 5 times and did not interact with the temperature anomaly tag. Similarly, based on Xiaoming's data, a user-label bipartite graph is constructed, where the left side represents users and the right side represents labels (the lines connecting them represent click behavior). It is found that Xiaowang, who is similar to Xiaoming, clicked on the temperature anomaly 12 times, which is greater than or equal to 10 times and belongs to the high confidence label. Xiaowang's temperature monitoring label is then propagated to Xiaoming. Finally, Xiaoming's user preference behavior pattern recognition result is the device monitoring preference type.

[0045] In this embodiment of the invention, optionally, the user preference behavior pattern recognition model is trained on the collaborative filtering recommendation algorithm and the tag propagation algorithm based on the corresponding multi-dimensional sample operational data, that is, the multi-dimensional sample user preference data of multiple sample users. Specifically: Collect multi-dimensional sample user preference data from multiple sample users corresponding to user preference behavior pattern types, such as function ID, usage frequency, and dwell time; extract user features from the multi-dimensional sample operation data of each sample user, and construct a user-function rating matrix for each sample user feature, where usage frequency is standardized to 0-10 points, and extract label weights, such as label click count / total click count; train the latent factor matrix of the collaborative filtering algorithm using alternating least squares method based on the user-function rating matrix and label weights of each sample user, where the number of latent factors = 50 and the regularization parameter = 0.01; train the propagation probability of the label propagation algorithm (such as Label Propagation) based on the user similarity map between each sample user in the multi-dimensional operation data (where the similarity threshold = 0.6), the number of iterations can be 100, or the model is considered complete when the model's loss value is less than or equal to a preset loss value, thus obtaining the required user preference behavior recognition model.

[0046] 105. Based on the behavior pattern recognition results of all behavior patterns, perform behavior pattern recognition on the target user to obtain the behavior pattern recognition results of the target user in the operation scenario.

[0047] In this embodiment of the invention, the behavior pattern recognition results for each behavior pattern classification are summarized to obtain the behavior pattern recognition results of the target user in the operational scenario. Further, based on the target user's behavior pattern recognition results, operational decisions are generated for the target user, and the behavior pattern recognition results and operational decisions for the target user are output. The operational decision corresponding to the target user indicates the operational operations to be performed on the target user using that decision. Further, for abnormal behavior pattern recognition results, specific abnormal nodes, such as login location steps and page access steps, are analyzed based on the abnormal behavior pattern recognition results and corresponding multi-dimensional operational data. Based on the abnormal nodes and the target user's identifier, the abnormal behavior pattern recognition results of the target user in the operational scenario are corrected so that the abnormal behavior pattern recognition results include the abnormal nodes and the target user's identifier.

[0048] It is evident that implementation Figure 1The described method collects multi-dimensional user data in operational scenarios, such as login behavior data, page access data, and user preference data. It then categorizes this data using pre-defined behavior pattern classifications. Next, it identifies behavior patterns in each dimension based on a matching recognition model for each behavior pattern classification. Finally, it combines the behavior pattern recognition results for each behavior pattern classification to identify the user's final behavior pattern in the operational scenario. This improves the accuracy of behavior pattern recognition and enables full-link tracking of user behavior, providing refined decision-making support for operational decisions. For example, it allows for the delivery of customized scenarios based on different behavior patterns, such as scheduling monitoring and defect work orders. Furthermore, it facilitates the dynamic adjustment of system resources based on behavior pattern recognition results, such as server load and bandwidth allocation, improving the accuracy of resource allocation.

[0049] In an optional embodiment, the method may further include the following operations: Based on the classification of all behavioral patterns, determine whether there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data; When a correlation is found between the results, all behavior pattern classifications are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classifications. Specifically, based on the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type, including: Based on the ranking analysis results, and using the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type.

[0050] In this optional embodiment, the ranking analysis results are used to indicate that each behavior pattern classification type needs to be identified by the corresponding behavior pattern recognition model in the corresponding order to obtain the corresponding behavior pattern recognition results.

[0051] In this optional embodiment, when it is determined that there is no correlation between the results, the above-mentioned operation of identifying the multi-dimensional operational data corresponding to each behavior pattern classification type based on the behavior pattern recognition model is performed to obtain the behavior pattern recognition result of the behavior pattern classification type. That is, the multi-dimensional operational data corresponding to each behavior pattern classification type is analyzed in parallel according to their respective behavior pattern recognition models to obtain the corresponding behavior pattern recognition result.

[0052] Taking the recognition of three types of behavioral patterns on the YunJing platform—login behavior, page access, and user preferences—as examples, this optional embodiment is illustrated below: Based on operational data such as login time, location, and device corresponding to login behavior patterns, stable login types (e.g., logins during weekdays from 9:00 AM to 6:00 PM) and abnormal login types (e.g., logins from different locations outside of working hours) are identified. Based on the operation path and dwell time corresponding to page access behavior patterns, operation-driven types are identified (e.g., high-frequency access to device monitoring → work orders → reports) and browsing types (without fixed paths). Based on the frequency of function usage corresponding to user preference behavior patterns, work order processing preferences and report download preferences are identified. Furthermore, through association rule mining model analysis, it is found that 85% of stable login users are operation-driven, and 70% have work order processing preferences, indicating a strong correlation among the three. Based on the strength of this correlation, the execution order is determined to be login behavior, page access, and user preference, and behavioral pattern identification is performed on the multi-dimensional operational data corresponding to login behavior patterns, page access behavior patterns, and user preference behavior patterns according to this ranking analysis results.

[0053] As can be seen, before implementing this optional embodiment to identify the corresponding multi-dimensional operational data using their respective behavior pattern recognition models, it first analyzes whether there is a correlation between the behavior pattern recognition results corresponding to each behavior pattern classification type. If there is no correlation, each behavior pattern classification type performs behavior pattern recognition in parallel based on its own behavior pattern recognition model. If there is a correlation, the behavior pattern classification types are first sorted, and then behavior pattern recognition is performed sequentially based on the sorting results. That is, a behavior pattern recognition method matching the situation is adopted according to different situations, which improves the flexibility of the behavior pattern recognition method, thereby helping to further improve the accuracy and efficiency of behavior pattern recognition, and thus helping to further improve the accuracy of operational decision-making.

[0054] In this optional embodiment, optionally, based on the ranking analysis results and the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type, including: Based on the ranking analysis results, the target behavior pattern classification type required for this identification is selected from all behavior pattern classification types. Then, based on the behavior pattern recognition model corresponding to the target behavior pattern classification type, the multidimensional operational data corresponding to that target behavior pattern classification type is identified to obtain the behavior pattern recognition result for that target behavior pattern classification type. This process is repeated, involving selecting the target behavior pattern classification type from all behavior pattern classification types based on the ranking analysis results, and identifying the multidimensional operational data corresponding to that target behavior pattern classification type based on the behavior pattern recognition model corresponding to that target behavior pattern classification type to obtain the behavior pattern recognition result for that target behavior pattern classification type. Specifically, the behavior pattern recognition result of the previous target behavior pattern classification type is updated to the multidimensional operational data corresponding to the current target behavior pattern classification type.

[0055] Taking the recognition of three types of behavioral patterns on the YunJing platform—login behavior, page access, and user preferences—as examples, this optional embodiment is illustrated below: Assuming the priority order is login behavior, page visits, and user preferences, we first analyze login behavior patterns. A login behavior pattern recognition model is used to identify multi-dimensional operational data such as login time, login location, and login device, yielding a stable login pattern (login between 9:00 AM and 6:00 PM on weekdays). Based on the ranking analysis results, we next analyze page visit behavior patterns. A page visit behavior pattern recognition model is used to identify stable login patterns along with multi-dimensional operational data such as operation path and dwell time, yielding an operations-driven pattern (high-frequency access path: device monitoring → work order → report). Finally, we analyze user preference behavior patterns. A user preference behavior pattern recognition model is used to identify stable login operations-driven patterns along with multi-dimensional operational data such as work order processing frequency and report download volume, yielding a work order processing preference pattern. The final comprehensive behavior pattern is: stable login + operations-driven + work order processing preference.

[0056] As can be seen, implementing this optional embodiment can also classify behavioral patterns that are related to each other, and based on the ranking analysis results, identify behavioral patterns in the multi-dimensional operational data corresponding to each behavioral pattern classification. In the identification process, the behavioral pattern identification results corresponding to the previous related behavioral pattern classification are used as input parameters for the behavioral pattern identification model corresponding to the current behavioral pattern classification, which strengthens the correlation of behavioral pattern identification results of each behavioral pattern classification and further improves the accuracy of behavioral pattern identification, thereby further improving the accuracy and reliability of matching corresponding operational decisions for users.

[0057] In another optional embodiment, before re-executing the process of selecting the target behavior pattern classification type required for this identification from all behavior pattern classification types based on the ranking analysis results, and identifying the multidimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result of the target behavior pattern classification type, the method may further include the following operations: Determine whether the behavior pattern recognition result of this target behavior pattern classification is the preset abnormal behavior pattern recognition result; When it is determined that the result is not the preset abnormal behavior pattern identification result, the operation of re-executing the sorting analysis result, selecting the target behavior pattern classification type required for this identification from all behavior pattern classification types, and identifying the multi-dimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern identification model corresponding to the target behavior pattern classification type, and obtaining the behavior pattern identification result of the target behavior pattern classification type. Optionally, the method may also include the following operations: When the result is determined to be a preset abnormal behavior pattern, for any behavior pattern classification type for which no recognition operation has been performed, the multidimensional operational data corresponding to the behavior pattern classification type is identified based on the behavior pattern recognition model corresponding to that behavior pattern classification type, and the behavior pattern recognition result of that behavior pattern classification type is obtained.

[0058] Taking the recognition of three types of behavioral patterns on the YunJing platform—login behavior, page access, and user preferences—as examples, this optional embodiment is illustrated below: Assuming the priority order is login behavior, page visits, and user preferences, we first analyze login behavior patterns. A login behavior pattern recognition model is used to identify login time, location, and device across multiple operational dimensions. This yields a stable login pattern, i.e., logins between 9:00 AM and 6:00 PM on weekdays, which is considered a non-abnormal behavior pattern. Based on the ranking analysis, we next analyze page visit behavior patterns. A page visit behavior pattern recognition model is used to identify stable login patterns along with operation paths, dwell time, and other operational data. If an operation-driven pattern is obtained, i.e., a high-frequency access path of device monitoring → work orders → reports, this is also considered a non-abnormal behavior pattern. Next, we analyze user preference behavior patterns. A user preference behavior pattern recognition model is used to identify stable login operation-driven patterns along with work order processing frequency, report download volume, and other operational data. If a work order processing preference is obtained, or if repeated access and exit patterns indicate termination, this is considered an abnormal behavior pattern. At this point, the stable login operation and maintenance driven identification results are no longer used as input parameters for the collaborative filtering model. Instead, the collaborative filtering model is used to identify multi-dimensional operational data such as work order processing frequency and report download volume to obtain the corresponding user preference behavior pattern identification results.

[0059] As can be seen, implementing this optional embodiment for behavior pattern recognition of related behavior pattern classification types involves judging abnormal behavior patterns for the behavior pattern recognition result of each behavior pattern classification type. Only when it is a normal behavior pattern recognition result is its behavior pattern recognition result used as the input parameter for the behavior pattern recognition model of the next related behavior pattern classification type to perform corresponding behavior pattern recognition. Otherwise, the next related behavior pattern classification type still performs behavior pattern analysis based on its own multi-dimensional operational data. This reduces the possibility of using abnormal behavior pattern recognition results as input parameters for the behavior pattern recognition model of the next behavior pattern classification type, which could lead to recognition errors in the next behavior pattern and thus operational decision-making errors.

[0060] In this optional embodiment, optionally, based on classifying all behavioral patterns into types, determining whether there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data includes: Determine whether there are corresponding analysis order labels for all behavior pattern classification types; When it is determined that there are corresponding analysis order labels for all behavior pattern classification types, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing their respective multi-dimensional operational data; This involves analyzing and ranking all behavior pattern classifications to obtain the ranking analysis results for each behavior pattern classification type, including: Based on the analysis order label corresponding to each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

[0061] In this optional embodiment, for behavior patterns with related relationships, corresponding analysis order labels have been pre-set for them. These analysis order labels can be numeric or non-numeric, such as a page access behavior pattern carrying the message "analysis order follows login behavior pattern".

[0062] As can be seen, by determining whether there are corresponding analysis order labels for behavior pattern classification types, the correlation between the pattern recognition results is analyzed, which improves the efficiency and accuracy of the analysis of correlation between pattern recognition results. Furthermore, based on the analysis order labels, the ranking analysis results corresponding to all behavior pattern classification types are determined, which improves the accuracy and efficiency of the ranking analysis results. In this way, the recognition accuracy and reliability of the behavior patterns corresponding to each behavior pattern classification type are improved.

[0063] In another alternative embodiment, the method may further include the following operations: When it is determined that there is no corresponding analysis order label for all behavior pattern classification types, obtain the historical behavior pattern recognition results and historical model input parameters for each behavior pattern classification type. Based on the historical behavior pattern recognition results and historical model input parameters corresponding to all behavior pattern classification types, determine whether there are any behavior pattern classification types whose historical behavior pattern recognition results serve as historical model input parameters for other behavior pattern classification types. When the result is determined to be yes, it is determined that there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data; This involves analyzing and ranking all behavior pattern classifications to obtain the ranking analysis results for each behavior pattern classification type, including: Based on the historical model input parameters for each behavior pattern classification type, all behavior pattern classification types are analyzed and ranked to obtain the ranking analysis results corresponding to all behavior pattern classification types.

[0064] In this optional embodiment, when the result is determined to be negative, it is determined that there is no correlation between the behavior pattern recognition results obtained by analyzing their respective multi-dimensional operational data.

[0065] Taking the recognition of three types of behavioral patterns on the YunJing platform—login behavior, page access, and user preferences—as examples, this optional embodiment is illustrated below: The system retrieves historical behavior pattern recognition results and historical model input parameters for three types of behavior patterns: login behavior pattern, page access pattern, and user preference pattern. Specifically, for login behavior pattern: the historical behavior pattern recognition result is a stable login type, i.e., logins occurring between 9:00 and 18:00 on weekdays, accounting for 60%; for page access pattern: the historical behavior pattern recognition result is an operation and maintenance driven type, i.e., frequent access to device monitoring pages, accounting for 45%, and its historical model input parameters include a login behavior recognition result field, indicating a correlation between login behavior pattern and page access pattern; for user preference pattern: its historical behavior pattern recognition result is a work order processing preference recognition result, and its historical model input parameters include a frequent access to device monitoring pages field, indicating a correlation between page access pattern and user preference pattern. In short, all three are correlated.

[0066] As can be seen, after analyzing that there are no corresponding analysis order labels for each behavior pattern classification type, this optional embodiment further performs correlation analysis based on the historical behavior pattern recognition results and historical model input parameters of each behavior pattern classification type, which improves the comprehensiveness of the correlation analysis of each behavior pattern classification type, thereby further improving the accuracy of the correlation analysis of each behavior pattern classification type; and based on the historical model input parameters of each behavior pattern classification type, it performs ranking analysis results for each behavior pattern classification type, which improves the accuracy of the ranking analysis results, and thus helps to further improve the recognition accuracy and reliability of the behavior patterns corresponding to each behavior pattern classification type.

[0067] Example 2 Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an operational scenario behavior pattern recognition device based on big data analysis disclosed in an embodiment of the present invention. Figure 2 The described device can be applied to scenarios requiring behavioral pattern recognition, such as the operation and maintenance of power grid digital platforms. Figure 2 As shown, the device may include: The data collection module 201 is used to collect the operational dataset generated by the target user performing operational operations in the operational scenario within the target time period; The segmentation module 202 is used to perform segmentation operations on the operational dataset according to multiple pre-determined behavioral patterns to obtain multi-dimensional operational data corresponding to each behavioral pattern segmentation type. The acquisition module 203 is used to classify each behavior pattern into a type and acquire a behavior pattern recognition model that matches the behavior pattern classification type. The identification module 204 is used to identify the multi-dimensional operational data corresponding to each behavior pattern classification type based on the behavior pattern identification model corresponding to each behavior pattern classification type, and obtain the behavior pattern identification result of the behavior pattern classification type. The recognition module 204 is also used to perform behavior pattern recognition on the target user based on the behavior pattern recognition results of all behavior patterns classified into categories, so as to obtain the behavior pattern recognition results of the target user in the operation scenario.

[0068] It is evident that implementation Figure 2 The described device collects multi-dimensional user data in operational scenarios, such as login behavior data, page access data, and user preference data. It then categorizes this data based on preset behavior pattern classifications, and performs behavior pattern recognition for each category using a matching recognition model. Finally, by combining the behavior pattern recognition results for each category, the device identifies the user's final behavior pattern in the operational scenario. This improves the accuracy of behavior pattern recognition and enables full-link tracking of user behavior, providing refined decision-making support for operational decisions. For example, it allows for the delivery of customized scenarios based on different behavior patterns, such as scheduling monitoring and defect work orders. Furthermore, it facilitates the dynamic adjustment of system resources based on behavior pattern recognition results, such as server load and bandwidth allocation, improving the accuracy of resource allocation.

[0069] In an optional embodiment, such as Figure 3 As shown, the device may further include: The judgment module 205 is used to classify all behavioral patterns into types and determine whether there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data. The sorting module 206 is used to analyze and sort all behavior pattern classification types when it is determined that there is a correlation between the results, and to obtain the sorting analysis results corresponding to all behavior pattern classification types. The identification module 204, based on the behavior pattern identification model corresponding to each behavior pattern classification type, identifies the multidimensional operational data corresponding to that behavior pattern classification type, and obtains the behavior pattern identification result for that behavior pattern classification type in the following specific ways: Based on the ranking analysis results, and using the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type.

[0070] It is evident that implementation Figure 3Before using their respective behavior pattern recognition models to identify corresponding multi-dimensional operational data, the described device first analyzes whether there is a correlation between the behavior pattern recognition results corresponding to each behavior pattern classification type. If there is no correlation, each behavior pattern classification type performs behavior pattern recognition in parallel based on its own behavior pattern recognition model. If there is a correlation, the behavior pattern classification types are first sorted, and then behavior pattern recognition is performed sequentially based on the sorting results. That is, a behavior pattern recognition method matching the situation is adopted according to different situations, which improves the flexibility of the behavior pattern recognition method, thereby helping to further improve the accuracy and efficiency of behavior pattern recognition, and thus helping to further improve the accuracy of operational decision-making.

[0071] In this optional embodiment, such as Figure 3 As shown, the identification module 204, based on the ranking analysis results and the behavior pattern identification model corresponding to each behavior pattern classification type, identifies the multidimensional operational data corresponding to that behavior pattern classification type, and obtains the behavior pattern identification result for that behavior pattern classification type in the following specific ways: Based on the ranking analysis results, the target behavior pattern classification type required for this identification is selected from all behavior pattern classification types. Then, based on the behavior pattern recognition model corresponding to the target behavior pattern classification type, the multidimensional operational data corresponding to that target behavior pattern classification type is identified to obtain the behavior pattern recognition result for that target behavior pattern classification type. This process is repeated, involving selecting the target behavior pattern classification type from all behavior pattern classification types based on the ranking analysis results, and identifying the multidimensional operational data corresponding to that target behavior pattern classification type based on the behavior pattern recognition model corresponding to that target behavior pattern classification type to obtain the behavior pattern recognition result for that target behavior pattern classification type. Specifically, the behavior pattern recognition result of the previous target behavior pattern classification type is updated to the multidimensional operational data corresponding to the current target behavior pattern classification type.

[0072] It is evident that implementation Figure 3 The described device categorizes interrelated behavioral patterns and, based on ranking analysis results, identifies behavioral patterns from multi-dimensional operational data corresponding to each behavioral pattern category. During the identification process, the behavioral pattern identification results from the previous related behavioral pattern category are used as input parameters for the behavioral pattern identification model corresponding to the current behavioral pattern category. This strengthens the correlation between the behavioral pattern identification results of each behavioral pattern category, further improves the accuracy of behavioral pattern identification, and thus further enhances the accuracy and reliability of matching users with corresponding operational decisions.

[0073] In an optional embodiment, such as Figure 3As shown, the judgment module 205 is also used to determine whether the behavior pattern recognition result of the current target behavior pattern classification is a preset abnormal behavior pattern recognition result before the recognition module 204 re-executes the operation of filtering the target behavior pattern classification type from all behavior pattern classification types according to the sorting analysis results, and identifying the multi-dimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result of the target behavior pattern classification type. When it is determined that it is not a preset abnormal behavior pattern recognition result, the recognition module 204 is triggered to re-execute the operation of filtering the target behavior pattern classification type from all behavior pattern classification types according to the sorting analysis results, and identifying the multi-dimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result of the target behavior pattern classification type. The identification module 204 is also used to identify the multi-dimensional operational data corresponding to any behavior pattern classification type that has not been identified when it is determined to be a preset abnormal behavior pattern identification result, based on the behavior pattern identification model corresponding to the behavior pattern classification type, and to obtain the behavior pattern identification result of the behavior pattern classification type.

[0074] It is evident that implementation Figure 3 The described device identifies behavioral patterns in related behavioral pattern categories. For each behavioral pattern identification result, it performs an abnormal behavioral pattern judgment. Only when the result is a normal behavioral pattern is it used as the input parameter for the behavioral pattern identification model of the next related behavioral pattern category. Otherwise, the next related behavioral pattern category still performs behavioral pattern analysis based on its own multi-dimensional operational data. This reduces the possibility of using abnormal behavioral pattern identification results as input parameters for the behavioral pattern identification model of the next behavioral pattern category, which could lead to identification errors in the next behavioral pattern and thus operational decision-making errors.

[0075] In this optional embodiment, such as Figure 3 As shown, the judgment module 205 classifies all behavioral patterns and determines whether there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data in specific ways, including: Determine whether there are corresponding analysis order labels for all behavior pattern classification types; When it is determined that there are corresponding analysis order labels for all behavior pattern classification types, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing their respective multi-dimensional operational data; The sorting module 206 analyzes and sorts all behavior pattern classification types, and obtains the specific methods for the sorting analysis results corresponding to all behavior pattern classification types, including: Based on the analysis order label corresponding to each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

[0076] It is evident that implementation Figure 3 The described device improves the efficiency and accuracy of analyzing the correlation between pattern recognition results by judging whether there is a corresponding analysis sequence label for the behavior pattern classification type. It also improves the accuracy and efficiency of determining the ranking analysis results corresponding to all behavior pattern classification types based on the analysis sequence label, thereby improving the accuracy and efficiency of the ranking analysis results and thus improving the recognition accuracy and reliability of the behavior patterns corresponding to each behavior pattern classification type.

[0077] In this optional embodiment, such as Figure 3 As shown, the specific method by which the judgment module 205 classifies all behavioral patterns and determines whether there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data also includes: When it is determined that there is no corresponding analysis order label for all behavior pattern classification types, obtain the historical behavior pattern recognition results and historical model input parameters for each behavior pattern classification type. Based on the historical behavior pattern recognition results and historical model input parameters corresponding to all behavior pattern classification types, determine whether there are any behavior pattern classification types whose historical behavior pattern recognition results serve as historical model input parameters for other behavior pattern classification types. When the result is determined to be yes, it is determined that there is a correlation between the behavioral pattern recognition results obtained by analyzing their respective multi-dimensional operational data; The sorting module 206 analyzes and sorts all behavior pattern classification types, and obtains the specific methods for the sorting analysis results corresponding to all behavior pattern classification types, including: Based on the historical model input parameters for each behavior pattern classification type, all behavior pattern classification types are analyzed and ranked to obtain the ranking analysis results corresponding to all behavior pattern classification types.

[0078] It is evident that implementation Figure 3The described device, after analyzing that there are no corresponding analysis sequence labels for each behavior pattern classification type, further performs correlation analysis based on the historical behavior pattern recognition results and historical model input parameters of each behavior pattern classification type. This improves the comprehensiveness of the correlation analysis for each behavior pattern classification type, thereby further improving the accuracy of the correlation analysis for each behavior pattern classification type. Furthermore, based on the historical model input parameters of each behavior pattern classification type, it performs ranking analysis results for each behavior pattern classification type, improving the accuracy of the ranking analysis results. This, in turn, helps to further improve the recognition accuracy and reliability of the behavior patterns corresponding to each behavior pattern classification type.

[0079] Example 3 Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of another operational scenario behavior pattern recognition device based on big data analysis disclosed in an embodiment of the present invention. Figure 4 The described device can be applied to scenarios requiring behavioral pattern recognition, such as the operation and maintenance of power grid digital platforms. Figure 4 As shown, the device may include: Memory 301 storing executable program code; Processor 302 coupled to memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute some or all of the steps in any of the operation scenario behavior pattern recognition methods based on big data analysis disclosed in Embodiment 1 of the present invention.

[0080] Example 5 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in any of the big data analysis-based operational scenario behavior pattern recognition methods disclosed in Embodiment 1 of this invention.

[0081] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0082] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0083] Finally, it should be noted that the above embodiments are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying operational scenario behavior patterns based on big data analysis, characterized in that, The method includes: Collect operational datasets generated by target users performing operational operations in the operational scenario within a target time period; Based on multiple predetermined behavioral pattern classification types, the operational dataset is divided to obtain multidimensional operational data corresponding to each behavioral pattern classification type. Based on each behavior pattern classification type, obtain the behavior pattern recognition model that matches the behavior pattern classification type; Based on the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to the behavior pattern classification type is identified to obtain the behavior pattern recognition result of the behavior pattern classification type. Based on the behavior pattern recognition results of all the aforementioned behavior pattern classifications, the target user's behavior pattern is identified to obtain the target user's behavior pattern recognition results in the operational scenario.

2. The method according to claim 1, characterized in that, The method further includes: Based on the classification of all the aforementioned behavioral patterns, determine whether there is a correlation between the behavioral pattern recognition results obtained by analyzing the multi-dimensional operational data for each of them; When a correlation is found between the results, all the behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all the behavior pattern classification types. The step of identifying the multidimensional operational data corresponding to each behavior pattern classification type using a behavior pattern recognition model to obtain the behavior pattern recognition result for that behavior pattern classification type includes: Based on the ranking analysis results, and using the behavior pattern recognition model corresponding to each behavior pattern classification type, the multidimensional operational data corresponding to that behavior pattern classification type is identified to obtain the behavior pattern recognition result for that behavior pattern classification type.

3. The method according to claim 2, characterized in that, The process of identifying the multidimensional operational data corresponding to each behavior pattern classification type based on the ranking analysis results and the behavior pattern recognition model corresponding to that behavior pattern classification type, to obtain the behavior pattern recognition result for that behavior pattern classification type, includes: Based on the ranking analysis results, the target behavior pattern classification type required for this identification is selected from all the behavior pattern classification types. Then, based on the behavior pattern recognition model corresponding to the target behavior pattern classification type, the multidimensional operational data corresponding to the target behavior pattern classification type is identified to obtain the behavior pattern recognition result for that target behavior pattern classification type. The process of selecting the target behavior pattern classification type from all the behavior pattern classification types based on the ranking analysis results and identifying the multidimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result for that target behavior pattern classification type is repeated. Specifically, the behavior pattern recognition result of the previous target behavior pattern classification type is updated to the multidimensional operational data corresponding to the current target behavior pattern classification type.

4. The method according to claim 3, characterized in that, Before the step of re-executing the step of selecting the target behavior pattern classification type required for this identification from all the behavior pattern classification types based on the sorting analysis results, and identifying the multidimensional operational data corresponding to the target behavior pattern classification type based on the behavior pattern recognition model corresponding to the target behavior pattern classification type to obtain the behavior pattern recognition result of the target behavior pattern classification type, the method further includes: Determine whether the behavior pattern recognition result of the target behavior pattern classification type is a preset abnormal behavior pattern recognition result; When it is determined that the result is not the preset abnormal behavior pattern identification result, the operation of filtering out the target behavior pattern classification type required for this identification from all the behavior pattern classification types according to the sorting analysis result is re-executed, and the multi-dimensional operational data corresponding to the target behavior pattern classification type is identified based on the behavior pattern identification model corresponding to the target behavior pattern classification type to obtain the behavior pattern identification result of the target behavior pattern classification type is obtained. The method further includes: When the result is determined to be the preset abnormal behavior pattern identification result, for any behavior pattern classification type for which no identification operation has been performed, the multidimensional operational data corresponding to the behavior pattern classification type is identified based on the behavior pattern identification model corresponding to the behavior pattern classification type, and the behavior pattern identification result of the behavior pattern classification type is obtained.

5. The method according to claim 3 or 4, characterized in that, The process of classifying behaviors into types based on all the aforementioned behaviors and determining whether there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data for each type includes: Determine whether there are corresponding analysis order labels for all the aforementioned behavior pattern classification types; When it is determined that all the behavior pattern classification types have corresponding analysis order labels, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing their respective multi-dimensional operational data; The step of analyzing and ranking all the behavior pattern classification types to obtain the ranking analysis results corresponding to all the behavior pattern classification types includes: Based on the analysis order label corresponding to each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

6. The method according to claim 5, characterized in that, The method further includes: When it is determined that there is no corresponding analysis order label for any of the behavior pattern classification types, the historical behavior pattern recognition results and historical model input parameters corresponding to each behavior pattern classification type are obtained. Based on the historical behavior pattern recognition results and historical model input parameters corresponding to all the behavior pattern classification types, determine whether there is a behavior pattern classification type among all the behavior pattern classification types whose historical behavior pattern recognition results are used as historical model input parameters for other behavior pattern classification types. When the result is determined to be yes, it is determined that there is a correlation between the behavior pattern recognition results obtained by analyzing the multi-dimensional operational data of each of them; The step of analyzing and ranking all the behavior pattern classification types to obtain the ranking analysis results corresponding to all the behavior pattern classification types includes: Based on the historical model input parameters of each behavior pattern classification type, all behavior pattern classification types are analyzed and sorted to obtain the sorting analysis results corresponding to all behavior pattern classification types.

7. The method according to claim 6, characterized in that, All the aforementioned behavior pattern classifications include user login behavior pattern types, page access behavior pattern types, and user preference behavior pattern types; When there is a correlation between the behavior pattern recognition results corresponding to the user login behavior pattern type, the page access behavior pattern type, and the user preference behavior pattern type, the ranking analysis result is used to indicate that the user login behavior pattern type, the page access behavior pattern type, and the user preference behavior pattern type are analyzed sequentially.

8. A device for recognizing operational scenario behavior patterns based on big data analysis, characterized in that, The device includes: The data collection module is used to collect operational datasets generated by target users performing operational operations in the operational scenario within a target time period. The segmentation module is used to perform a segmentation operation on the operational dataset according to multiple pre-determined behavioral patterns to obtain multi-dimensional operational data corresponding to each behavioral pattern segmentation type. The acquisition module is used to classify each behavior pattern into a type and acquire a behavior pattern recognition model that matches the behavior pattern classification type. The identification module is used to identify the multidimensional operational data corresponding to each behavior pattern classification type based on the behavior pattern identification model corresponding to each behavior pattern classification type, and obtain the behavior pattern identification result of the behavior pattern classification type. The identification module is further configured to perform behavior pattern identification on the target user based on the behavior pattern identification results of all the behavior pattern classifications, and obtain the behavior pattern identification results of the target user in the operation scenario.

9. A device for recognizing operational scenario behavior patterns based on big data analysis, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the operational scenario behavior pattern recognition method based on big data analysis as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, execute the operational scenario behavior pattern recognition method based on big data analysis as described in any one of claims 1-7.

Images