An intelligent analysis system and method based on Android system upgrade anomaly

Abstract

The application provides an intelligent analysis system and method based on Android system upgrade anomaly, comprising: a log collection module for collecting corresponding system logs and program running logs when the Android system fails; a packing module for packing the system logs and program running logs to generate a compressed file; an uploading module for obtaining a target address and authentication information for uploading the compressed file, and uploading the compressed file to a server; a decompression module for decompressing the compressed file based on the server, and extracting key information from the logs based on a predefined keyword; an analysis module for inputting the key information into a preset AI analysis model for analysis to obtain an analysis result and a processing suggestion of the upgrade failure event; and an associated storage module for associating and storing the analysis result, the processing suggestion and storage information of the compressed file, and querying and downloading logs based on a device identifier. The efficiency and accuracy of upgrade anomaly positioning are improved.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to an intelligent analysis system and method for Android system upgrade anomalies. Background Technology

[0002] During OTA (Over-The-Air) upgrades for the Android operating system, the upgrade package is typically sent to the device over the network. Upon receiving the package, the device installs it and completes the system update. However, due to various reasons, such as network fluctuations, corrupted upgrade packages, device compatibility issues, or system file conflicts, anomalies may occur during the OTA upgrade process, leading to upgrade failure. In such cases, obtaining detailed log information is crucial for developers to locate and fix the problem. Current methods for collecting Android system upgrade logs have some difficulties and shortcomings, specifically in the following aspects: 1. Poor timeliness of log capture: Traditional log capture methods usually require developers to prepare log tools in advance or capture logs in real time through debugging tools. This often delays the capture of abnormal information and makes it easy to miss important error details. 2. Large and disordered log volume: Android system log files are large and complex. Logs during the upgrade process are often intertwined with logs from other device operations, which requires developers to manually filter relevant logs when locating problems, increasing workload and the risk of errors. 3. Inconvenient log capture: Traditional log capture methods usually require connecting devices, manually entering commands, or using special software to extract logs. This is not only time-consuming and laborious, but also difficult for some non-professional users, which can easily lead to operational errors or omission of key logs. Therefore, in order to overcome the above-mentioned technical problems, the present invention provides an intelligent analysis system and method based on Android system upgrade anomalies. Summary of the Invention

[0003] This invention provides an intelligent analysis system and method for Android system upgrade anomalies, effectively solving problems such as poor timeliness of log capture, large log volume and difficulty in filtering, and inconvenient operation during traditional OTA upgrades. This technical solution incorporates a trigger mechanism within the upgrade program to automatically collect system logs and program execution logs when an upgrade failure is detected. These logs are packaged and sent to a cloud server via segmented upload. The server decompresses the log package, extracts key information based on predefined keywords, and imports it into an AI model for analysis, generating failure causes and solution suggestions. Finally, the analysis results and suggestions are associated with the device identifier and stored, supporting querying details or downloading the original logs by device number. This effectively automates and intelligently integrates log collection, uploading, analysis, and querying, improving the efficiency and accuracy of upgrade anomaly location and providing data support for subsequent system optimization.

[0004] This invention provides an intelligent analysis system based on Android system upgrade anomalies, comprising: The log collection module is used to collect system logs and program execution logs corresponding to the upgrade process when a failure event occurs during the OTA upgrade of the Android system. The packaging module is used to package the collected system logs and program runtime logs into a compressed file; The upload module is used to request and obtain the target address and authentication information for uploading the compressed file from the server, and upload the compressed file to the server according to the target address and authentication information; The decompression module is used to decompress compressed files based on the server and extract key information from the decompression logs based on predefined keywords; The analysis module is used to input key information into a preset AI analysis model for analysis, and obtain analysis results and handling suggestions for upgrade failure events; The associated storage module is used to associate and save the analysis results, processing suggestions, and storage information of the compressed file, and to query and download logs based on the device identifier.

[0005] Preferably, an intelligent analysis system based on Android system upgrade anomalies includes a log collection module, comprising: The log collection unit is used to collect system logs and program execution logs corresponding to the upgrade process when a failure event is detected. The collected system logs include: The first system log is obtained from the system runtime cache by simulating log capture commands; When the OTA upgrade process is in recovery upgrade mode, the system log also includes a second system log obtained from the recovery partition that records details of the system installation process; The program execution log is obtained from a specified log file written during the runtime of the upgrade program.

[0006] Preferably, an intelligent analysis system based on Android system upgrade anomalies includes an upload module comprising: The authentication information acquisition unit is used to acquire the authentication information from the server, including: cloud service provider identifier, storage node information, and temporary key; The upload connection building unit is used to build an upload connection with the target cloud storage service based on the cloud service provider identifier, storage node information, and temporary key. An upload unit is used to upload the compressed file to the server based on the target address and the upload connection.

[0007] Preferably, in an intelligent analysis system based on Android system upgrade anomalies, the decompression module is used for: The decompression unit is used to decompress the received compressed files on the server side; The retrieval unit is used to retrieve the decompressed log content based on a predefined set of keywords; Key information determination unit, used for: Extract the log line containing any keyword from the keyword set and the context information corresponding to the log line from the log content; Log lines and their corresponding context information are considered key information.

[0008] Preferably, an intelligent analysis system based on Android system upgrade anomalies, with an associated storage module, includes: The association unit is used to associate the analysis results and processing suggestions with the device identifier where the upgrade failure event occurred, and at the same time record the storage path of the compressed file on the server. A storage unit is used to store the associated result data and the storage path in the database; The download access unit is used to receive query requests based on the device identifier stored, retrieve and return the corresponding analysis results and processing suggestions from the database according to the device identifier, and provide download access to compressed files according to the associated storage path.

[0009] Preferably, an intelligent analysis system based on Android system upgrade anomalies includes an upload unit comprising: The communication channel construction subunit is used to construct a communication channel with the target cloud storage service based on the target address and upload connection. The piecewise parameter constraints determine the sub-units, used for: Information such as the size of compressed files, the network type of the device that experienced the upgrade failure, and real-time bandwidth information are obtained based on the communication channel. Based on the storage node information corresponding to the target address and the cloud service provider identifier, determine the sharding parameter constraints supported by the target cloud storage service. The matching subunit is used to match the network type with the preset policy library to determine the baseline fragment size and baseline concurrency corresponding to the network type; The target fragment size determination subunit is used for: The upload duration of a single fragment is determined based on real-time bandwidth information, and the minimum number of fragments required to meet the target total upload duration is determined based on the size information of the compressed file and the upload duration of a single fragment. The minimum number of fragments is compared with the baseline concurrency, and the maximum result is taken as the final number of concurrent fragment uploads. The target fragment size is determined based on the fragmentation parameter constraints and the final number of concurrent fragment uploads.

[0010] Preferably, in an intelligent analysis system based on Android system upgrade anomalies, the upload unit further includes: The upload priority determination subunit is used to split the compressed file according to the target fragment size to obtain multiple data fragments. At the same time, it assigns upload priority to multiple data fragments according to the position of the data fragments in the compressed file. The data fragment upload subunit is used to upload multiple data fragments to the server storage location corresponding to the target address based on the final number of concurrent uploads and the upload priority according to the upload connection.

[0011] Preferably, an intelligent analysis system based on Android system upgrade anomalies includes an analysis module comprising: Directed graph building unit, used for: An information processing engine is built based on a predefined grammar rule library, and the extracted key information is input into the information processing engine to map unstructured log text sequences into structured atomic event sequences. Based on the temporal relationship of atomic event sequences, the correlation between atomic events and the temporal dependency and causal probability between atomic events are determined, and the correlation is used as nodes and the temporal dependency and causal probability are used as edges to construct a directed graph. Analysis unit, used for: The directed graph is input into the first-level network of the preset AI analysis model, and deep feature representations of nodes and edges in the directed graph are extracted based on the first-level network. The state feature vector in the Android system upgrade process is determined based on the deep feature representations. The first-level network is a graph neural network. The state feature vector is compared with the baseline vector of the pre-established normal upgrade process, and the deviation clusters in the feature space are located based on the second-level network of the preset AI analysis model. Based on the location results, the original directed graph is traced to determine the corresponding key nodes and edges. Based on the determination results, the initial abnormal events and subsequent propagation paths corresponding to the state deviation are identified. The second-level network is an anomaly detection and attribution network. The third-level network based on the AI ​​analysis model calls the internally stored fault mode knowledge graph to perform matching analysis on the initial abnormal event and subsequent propagation path, and obtains the analysis results containing the fault link based on the matching analysis results. The fault mode knowledge graph associates abnormal events, system component states and potential root causes in the form of logical rules, and the third-level network is a root cause reasoning network. The suggestion generation unit is used to generate a library of handling solutions associated with each abnormal node in the fault link based on the dynamic index of the fault link, and to sort and logically connect the handling solutions of each node in the fault link to obtain a sequence of handling suggestions.

[0012] Preferably, an intelligent analysis system based on Android system upgrade anomalies includes an analysis unit comprising: Initial feature initialization subunit, used for: Based on the nodes and edges of the directed graph, the initial feature vector of each node is extracted and initialized. The initial feature vector of each node is formed by concatenating the embedding vectors of the predefined action subject, operation object and status code in the corresponding atomic event. The initial feature vector of the edge is represented by the causal probability value. An adjacency matrix is ​​constructed based on the connection relationships of the directed graph, whereby the adjacency matrix is ​​used to define the message propagation path between nodes; Feature-determining subunits are used for: Based on a graph neural network, message passing between nodes is performed according to the adjacency matrix, where each node gathers the feature information of all its neighboring nodes after weighting by edge weights; Based on the graph neural network, the feature information of the neighboring nodes after each node is aggregated and the feature of the current node itself are nonlinearly transformed. Based on the nonlinear transformation result, the feature information of the neighboring nodes and the feature of the current node itself are fused and updated to obtain the deep feature representation of the current node. Perform global pooling on the deep feature representations of all nodes in the directed graph to obtain the state feature vector during the Android system upgrade process.

[0013] This invention provides an intelligent analysis method based on Android system upgrade anomalies, including: When an Android system fails during an OTA upgrade, collect the corresponding system logs and program execution logs. Package the collected system logs and program execution logs into a compressed file; Request and obtain the target address and authentication information for uploading the compressed file from the server, and upload the compressed file to the server according to the target address and authentication information; The server decompresses the compressed file and extracts key information from the decompression logs based on predefined keywords. Key information is input into a preset AI analysis model for analysis, resulting in analysis results and handling suggestions for upgrade failure events; The analysis results, processing suggestions, and storage information of the compressed file are associated and saved, and queries and log downloads are performed based on the device identifier.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This solution effectively addresses the problems of poor timeliness in log collection, large log volume and difficulty in filtering, and inconvenient operation during traditional OTA upgrades. By embedding a trigger mechanism within the upgrade program, it automatically collects system logs and program execution logs when an upgrade failure is detected. These logs are then packaged and sent to the cloud server via fragmented upload. The server decompresses the log packages, extracts key information based on predefined keywords, and imports them into an AI model for analysis, generating failure causes and solution suggestions. Finally, the analysis results and suggestions are stored in association with device identifiers, allowing users to query details or download the original logs by device number. This effectively automates and intelligently integrates log collection, uploading, analysis, and querying, improving the efficiency and accuracy of upgrade anomaly location and providing data support for subsequent system optimization.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of an intelligent analysis system based on Android system upgrade anomalies in an embodiment of the present invention; Figure 2 This is an OTA failure log analysis diagram of an intelligent analysis system based on Android system upgrade anomalies in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the entire process of an intelligent analysis system based on Android system upgrade anomalies in an embodiment of the present invention. Figure 4 This is a flowchart of an intelligent analysis method based on Android system upgrade anomalies in an embodiment of the present invention. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0019] Example 1: This example provides an intelligent analysis system based on Android system upgrade anomalies, such as... Figure 1 As shown, it includes: The log collection module is used to collect system logs and program execution logs corresponding to the upgrade process when a failure event occurs during the OTA upgrade of the Android system. The packaging module is used to package the collected system logs and program runtime logs into a compressed file; The upload module is used to request and obtain the target address and authentication information for uploading the compressed file from the server, and upload the compressed file to the server according to the target address and authentication information; The decompression module is used to decompress compressed files based on the server and extract key information from the decompression logs based on predefined keywords; The analysis module is used to input key information into a preset AI analysis model for analysis, and obtain analysis results and handling suggestions for upgrade failure events; The associated storage module is used to associate and save the analysis results, processing suggestions, and storage information of the compressed file, and to query and download logs based on the device identifier.

[0020] In this embodiment, upon detecting an OTA upgrade failure, log collection, packaging, and uploading to the cloud are automatically triggered. The server decompresses the logs, extracts key information using keywords, and then performs intelligent analysis using an AI model to generate analysis reports and suggestions. The results are then stored in association with the log files, facilitating subsequent querying or downloading of the original logs for secondary analysis by device identifier. This achieves automated and intelligent log collection, significantly improving the efficiency and accuracy of upgrade anomaly localization.

[0021] In this embodiment, OTA stands for Over-the-Air, meaning "over-the-air download technology." It is a technology that uses mobile communication networks (such as 2G, 3G, 4G, 5G) or Wi-Fi networks to remotely push software updates and system upgrades to mobile devices (such as Android phones and tablets) without requiring users to manually connect to external devices such as computers.

[0022] In this embodiment, the overall process from detecting an upgrade failure to triggering log collection, packaging, uploading to the server, and then the server analyzing, recording, and providing query and download services is illustrated in the flowchart below. Figure 2 As shown.

[0023] In this embodiment, the complete operation flowchart is as follows: Figure 3 As shown, the specific process includes: collecting logs after an upgrade failure, packaging the logs, requesting information to be uploaded to the cloud server, performing chunked uploads, the server parsing the logs and extracting key information, AI analysis to generate reports and suggestions, recording the information to the database, and supporting operators to query failure details or download log files by device identifier.

[0024] In this embodiment, the system log refers to a log file that records the operating system's running status, events, and errors. For example, the "Kernel panic" message output by the Android kernel or system service crash records.

[0025] In this embodiment, the program runtime log refers to the record of debugging, error, or status information generated during the execution of a specific application or process. For example, the "Installation aborted" error details output by the OTA upgrade program during the installation phase.

[0026] In this embodiment, the target address refers to the specific network location identifier on the server used to receive the compressed file. For example, an HTTP URL.

[0027] In this embodiment, key information refers to the core content directly related to the upgrade failure extracted from the logs. For example, error code "0x80070002", exception stack trace, or failure timestamp "2023-10-01 12:00:00".

[0028] In this embodiment, the processing suggestion refers to a specific solution or operation step provided based on the analysis results. For example, "re-download the upgrade package and verify its integrity" or "enter recovery mode to perform a cache clearing operation".

[0029] In this embodiment, the device identifier refers to a string that uniquely identifies an Android device.

[0030] The working principle and beneficial effects of the above technical solution are as follows: It effectively solves the problems of poor timeliness of log capture, large volume of logs and difficulty in filtering, and inconvenient operation in the traditional OTA upgrade process; This technical solution automatically collects system logs and program running logs when an upgrade failure is detected by building a trigger mechanism in the upgrade program, packages them, and sends them to the cloud server through a segmented upload method; The server decompresses the log package, extracts key information based on predefined keywords, and imports it into an AI model for analysis to generate failure reasons and solution suggestions; Finally, the analysis results and suggestions are associated with the device identifier and stored, supporting querying details or downloading the original logs by device number; It effectively realizes the full automation and intelligence of log collection, uploading, analysis, and querying, improves the efficiency and accuracy of upgrade anomaly location, and provides data support for subsequent system optimization.

[0031] Example 2: Based on Example 1, this example provides an intelligent analysis system for Android system upgrade anomalies, including a log collection module: The log collection unit is used to collect system logs and program execution logs corresponding to the upgrade process when a failure event is detected. The collected system logs include: The first system log is obtained from the system runtime cache by simulating log capture commands; When the OTA upgrade process is in recovery upgrade mode, the system log also includes a second system log obtained from the recovery partition that records details of the system installation process; The program execution log is obtained from a specified log file written during the runtime of the upgrade program.

[0032] In this embodiment, two types of logs are automatically collected when an upgrade fails: first, system logs obtained from the system cache by simulating the logcat command; and second, detailed installation process logs extracted from the / cache / recovery / last_log file when upgrading in recovery mode. The program runtime log is written to a specified file in real time by the upgrade program itself during runtime. This multi-source log collection mechanism ensures the comprehensiveness and completeness of exception information.

[0033] In this embodiment, the first system log refers to the regular logs that are captured in real time from the Android system runtime cache by simulating system commands.

[0034] In this embodiment, the second system log refers to the log obtained from the device's recovery system partition in recovery upgrade mode, which specifically records details of the installation operation process. In this embodiment, the specified log file refers to the log file whose path is pre-defined and written by the specific application responsible for performing the upgrade task during its operation. The beneficial effects of the above technical solution are: by distinguishing the log sources during system operation and special upgrade modes, and clarifying the acquisition path of program logs, more comprehensive and accurate log collection is achieved, ensuring that no key failure information is missed, and laying a more reliable data foundation for subsequent intelligent analysis.

[0035] Example 3: Based on Example 1, this example provides an intelligent analysis system for Android system upgrade anomalies, including an upload module comprising: The authentication information acquisition unit is used to acquire the authentication information from the server, including: cloud service provider identifier, storage node information, and temporary key; The upload connection building unit is used to build an upload connection with the target cloud storage service based on the cloud service provider identifier, storage node information, and temporary key. An upload unit is used to upload the compressed file to the server based on the target address and the upload connection.

[0036] In this embodiment, a chunked upload method is used when uploading compressed files. First, the upload link and relevant authentication information (such as cloud service provider, region, node, bucket, token, temporary key, etc.) are requested from the server. Then, a secure connection with the cloud storage service is established based on this information, enabling efficient and reliable uploading of large files. This method improves the stability and efficiency of the upload process and is suitable for log file transmission in mobile network environments.

[0037] In this embodiment, the cloud service provider identifier refers to a unique code used to distinguish different cloud computing vendors.

[0038] In this embodiment, storage node information refers to the specific location identifier used to store data in the cloud storage service.

[0039] The beneficial effects of the above technical solution are as follows: For example, by dynamically obtaining authentication information including cloud service providers and temporary keys, it is possible to securely and efficiently establish a connection with a specified cloud storage service, ensure the reliability of log file uploads, and improve compatibility and access security for different cloud service platforms.

[0040] Example 4: Based on Example 1, this example provides an intelligent analysis system for Android system upgrade anomalies. The decompression module is used for: The decompression unit is used to decompress the received compressed files on the server side; The retrieval unit is used to retrieve the decompressed log content based on a predefined set of keywords; Key information determination unit, used for: Extract the log line containing any keyword from the keyword set and the context information corresponding to the log line from the log content; Log lines and their corresponding context information are considered key information.

[0041] In this embodiment, after decompressing the log file, the server retrieves the log content using preset keywords (such as file anomaly, permission anomaly, signature anomaly, installation anomaly, battery storage anomaly, etc.), extracting log lines containing the keywords and their surrounding context information as input for subsequent AI analysis. This keyword-based log extraction method effectively filters irrelevant information, focusing on anomaly-related logs and improving analysis efficiency and accuracy.

[0042] In this embodiment, the keyword set refers to a pre-configured combination of specific words or phrases used to locate potential problems in logs. For example, a list containing words such as "error," "failed," "exception," and "abort" is used to quickly filter log entries that indicate anomalies or failures.

[0043] In this embodiment, context information refers to other log content in the log file that is adjacent to the line containing the retrieved keyword and has a certain correlation. For example, when a line containing the error "Installation failed" is retrieved, its context information includes the last few operation steps performed by the system before the error occurred (the first few lines) and the system status output that followed (the last few lines).

[0044] The beneficial effects of the above technical solution are: by decompressing the logs on the server side, searching based on a preset set of keywords, and extracting the matching log lines and their context as key information, the solution effectively focuses on log fragments related to the problem, reduces redundant data processing, and provides more accurate and complete input data for subsequent analysis.

[0045] Example 5: Based on Example 1, this example provides an intelligent analysis system for Android system upgrade anomalies, including an associated storage module, comprising: The association unit is used to associate the analysis results and processing suggestions with the device identifier where the upgrade failure event occurred, and at the same time record the storage path of the compressed file on the server. A storage unit is used to store the associated result data and the storage path in the database; The download access unit is used to receive query requests based on the device identifier stored, retrieve and return the corresponding analysis results and processing suggestions from the database according to the device identifier, and provide download access to compressed files according to the associated storage path.

[0046] In this embodiment, the analysis reports and solution suggestions generated by AI analysis are associated with device identifiers (such as device IDs), and the storage path and download link of the original log files on the cloud server are recorded. All associated information is stored in a database, allowing developers to query failure details by device ID, view analysis reports, and download the original log files for further analysis. This mechanism facilitates problem tracing and subsequent upgrade strategy optimization.

[0047] In this embodiment, the associated result data refers to a structured data record composed of analysis results, processing suggestions, device identifiers, and storage paths.

[0048] In this embodiment, the storage path refers to the specific location identifier of the compressed file in the server's storage system.

[0049] In this embodiment, the database refers to a software system used for structured storage, management, and querying of related result data.

[0050] In this embodiment, download access refers to a functional interface that provides compressed file reading and transfer services to authorized requesters based on the storage path. For example, an endpoint that can be invoked via a device identifier returns a download link for the corresponding log file.

[0051] The beneficial effects of the above technical solution are: by associating and storing the analysis results and processing suggestions with the device and log files, and providing query and download services based on the device identifier, centralized management and efficient tracing of historical upgrade issues are achieved, which facilitates quick problem location and acquisition of raw data.

[0052] Example 6: Based on Example 3, this example provides an intelligent analysis system for Android system upgrade anomalies, including an upload unit comprising: The communication channel construction subunit is used to construct a communication channel with the target cloud storage service based on the target address and upload connection. The piecewise parameter constraints determine the sub-units, used for: Information such as the size of compressed files, the network type of the device that experienced the upgrade failure, and real-time bandwidth information are obtained based on the communication channel. Based on the storage node information corresponding to the target address and the cloud service provider identifier, determine the sharding parameter constraints supported by the target cloud storage service. The matching subunit is used to match the network type with the preset policy library to determine the baseline fragment size and baseline concurrency corresponding to the network type; The target fragment size determination subunit is used for: The upload duration of a single fragment is determined based on real-time bandwidth information, and the minimum number of fragments required to meet the target total upload duration is determined based on the size information of the compressed file and the upload duration of a single fragment. The minimum number of fragments is compared with the baseline concurrency, and the maximum result is taken as the final number of concurrent fragment uploads. The target fragment size is determined based on the fragmentation parameter constraints and the final number of concurrent fragment uploads.

[0053] In this embodiment, the communication channel refers to the network path established between the target cloud storage service and the established upload connection for data transmission.

[0054] In this embodiment, the fragmentation parameter constraints refer to the technical parameter restrictions stipulated by the target cloud storage service for fragmented upload operations. For example, each fragment must be between 100KB and 5GB in size, and all fragments except the last one must be the same size.

[0055] In this embodiment, network type refers to the type of communication technology used by the device to access the Internet. For example, "Wi-Fi", "4G cellular network" or "5G cellular network".

[0056] In this embodiment, real-time bandwidth information refers to the actual data transmission rate available for the device's current network connection at the planned upload time.

[0057] In this embodiment, the preset policy library refers to a pre-configured set of rules that maps different network types to recommended upload parameters. For example, the library defines that when the network type is "Wi-Fi", the baseline fragment size is 2MB and the baseline concurrency is 4; when it is "4G", the baseline fragment size is 512KB and the baseline concurrency is 2.

[0058] In this embodiment, the baseline fragment size refers to the recommended initial fragment data size obtained from a preset policy library based on the network type.

[0059] In this embodiment, the baseline concurrency refers to the number of shard tasks recommended to be uploaded simultaneously, obtained from a preset policy library based on the network type.

[0060] In this embodiment, the single-segment upload time refers to the estimated time required to upload a single segment based on the real-time uplink bandwidth and the proposed segment size.

[0061] In this embodiment, the number of concurrent uploads per shard refers to the final total number of shard tasks that can simultaneously send upload requests to the cloud storage service.

[0062] The beneficial effects of the above technical solution are as follows: During the upload process, by dynamically analyzing the size of the compressed file, the current network type of the device, and the real-time bandwidth status, and in combination with the fragmentation limitations of the cloud storage service itself, the optimal fragment size and concurrent upload quantity are intelligently matched and calculated; the adaptive fragmentation strategy based on the real-time network environment and cloud service capabilities can effectively overcome the transmission bottleneck under weak or unstable network conditions, and maximize the use of available network bandwidth while ensuring compliance with cloud protocol constraints, thereby improving the success rate, transmission efficiency, and overall stability of large log file uploads.

[0063] Example 7: Based on Example 6, this example provides an intelligent analysis system for Android system upgrade anomalies. The upload unit further includes: The upload priority determination subunit is used to split the compressed file according to the target fragment size to obtain multiple data fragments. At the same time, it assigns upload priority to multiple data fragments according to the position of the data fragments in the compressed file. The data fragment upload subunit is used to upload multiple data fragments to the server storage location corresponding to the target address based on the final number of concurrent uploads and the upload priority according to the upload connection.

[0064] In this embodiment, the upload priority can be, for example, based on the data fragment located at the beginning of the file that contains the initial state log of the upgrade process, which is assigned a higher upload priority. For example, the system assigns a "high" priority to the first data fragment (the beginning of the file), a "medium" priority to the middle fragments, and a "low" priority to the last fragment, and will send the higher priority fragments first during upload.

[0065] In this embodiment, data sharding refers to dividing a complete compressed log file into several independent data blocks according to a determined target shard size.

[0066] The beneficial effects of the above technical solution are as follows: by assigning different upload priorities to data fragments according to their position in the file, priority transmission of critical data is achieved; the server can obtain and start analyzing the header log information most likely to reveal the cause of failure before the complete file transmission is completed, thereby shortening the startup delay of problem analysis; at the same time, it can ensure that the most critical data fragments are successfully delivered, improve the reliability of core diagnostic information upload, and buy valuable time for rapid response and fault diagnosis.

[0067] Example 8: Based on Example 1, this example provides an intelligent analysis system for Android system upgrade anomalies. The analysis module includes: Directed graph building unit, used for: An information processing engine is built based on a predefined grammar rule library, and the extracted key information is input into the information processing engine to map unstructured log text sequences into structured atomic event sequences. Based on the temporal relationship of atomic event sequences, the correlation between atomic events and the temporal dependency and causal probability between atomic events are determined, and the correlation is used as nodes and the temporal dependency and causal probability are used as edges to construct a directed graph. Analysis unit, used for: The directed graph is input into the first-level network of the preset AI analysis model, and deep feature representations of nodes and edges in the directed graph are extracted based on the first-level network. The state feature vector in the Android system upgrade process is determined based on the deep feature representations. The first-level network is a graph neural network. The state feature vector is compared with the baseline vector of the pre-established normal upgrade process, and the deviation clusters in the feature space are located based on the second-level network of the preset AI analysis model. Based on the location results, the original directed graph is traced to determine the corresponding key nodes and edges. Based on the determination results, the initial abnormal events and subsequent propagation paths corresponding to the state deviation are identified. The second-level network is an anomaly detection and attribution network. The third-level network based on the AI ​​analysis model calls the internally stored fault mode knowledge graph to perform matching analysis on the initial abnormal event and subsequent propagation path, and obtains the analysis results containing the fault link based on the matching analysis results. The fault mode knowledge graph associates abnormal events, system component states and potential root causes in the form of logical rules, and the third-level network is a root cause reasoning network. The suggestion generation unit is used to generate a library of handling solutions associated with each abnormal node in the fault link based on the dynamic index of the fault link, and to sort and logically connect the handling solutions of each node in the fault link to obtain a sequence of handling suggestions.

[0068] In this embodiment, the information processing engine refers to a software module built according to a predefined grammar rule base, used to map unstructured log text sequences into structured atomic event sequences.

[0069] In this embodiment, the atomic event sequence refers to a sequence of structured events, derived from log information, that includes elements such as the action subject, the operation object, and the status code, arranged in chronological order.

[0070] In this embodiment, a directed graph refers to a graph constructed with atomic events as nodes and temporal dependencies and causal probabilities between events as edges, used to represent the relationships between events during the upgrade process.

[0071] In this embodiment, a node refers to an element in a directed graph that represents an atomic event.

[0072] In this embodiment, an edge refers to a node in a directed graph that represents the temporal dependency and causal probability relationship between events.

[0073] In this embodiment, temporal dependence and causal probability refer to the temporal sequence of atomic events and the measure of the probability of causing other events to occur.

[0074] In this embodiment, the first-level network refers to the graph neural network in the preset AI analysis model, which is used to learn deep feature representations of nodes and edges from a directed graph.

[0075] In this embodiment, deep feature representation refers to the data vector extracted through a graph neural network that represents the high-level abstract features of nodes and edges.

[0076] In this embodiment, the state feature vector refers to a fixed-dimensional feature vector that represents the state of the entire upgrade process, generated through a global pooling operation.

[0077] In this embodiment, the baseline vector refers to the feature vector representing a state without abnormalities, which is learned in advance from the normal upgrade process.

[0078] In this embodiment, the second-level network refers to the anomaly detection and attribution network in the preset AI analysis model, which is used to locate off-center clusters in the feature space.

[0079] In this embodiment, the off-cluster refers to the set of state feature vectors that differ significantly from the reference vector in the feature space.

[0080] In this embodiment, key nodes and edges refer to atomic events and their relationships that play a decisive role in state deviation in a directed graph.

[0081] In this embodiment, the initial abnormal event and subsequent propagation path refer to the first abnormal event that occurs and the sequence of subsequent abnormal events triggered by that event, which is traced back from the key node.

[0082] In this embodiment, the third-level network refers to the root cause reasoning network in the preset AI analysis model, which is used to call the fault mode knowledge graph for matching analysis.

[0083] In this embodiment, the fault mode knowledge graph refers to a knowledge base that associates abnormal events, system component states, and potential root causes in the form of logical rules.

[0084] In this embodiment, the fault chain refers to a sequence of critical event nodes listed in causal and chronological order, starting from the root cause event, until the upgrade fails.

[0085] In this embodiment, the disposal plan library refers to a data set that stores disposal measures associated with various abnormal nodes.

[0086] In this embodiment, the processing suggestion sequence refers to the sequence of processing steps generated by sorting and logically concatenating the handling plans for each node in a faulty link.

[0087] The beneficial effects of the above technical solution are as follows: by transforming unstructured log information into a structured sequence of atomic events and constructing a directed graph model, deep features can be extracted using graph neural networks, which can comprehensively and accurately characterize the state of the system upgrade process. By comparing with normal benchmarks, abnormal deviation clusters can be quickly located and traced back to key nodes and edges, thereby accurately identifying the initial abnormal events and propagation paths. Combined with the fault mode knowledge graph, root cause reasoning is performed to generate detailed fault chains, and then a dynamic indexing solution library is generated to form a targeted sequence of processing suggestions. This significantly improves the analysis efficiency and diagnostic accuracy of Android system upgrade failure problems, realizes automated and intelligent fault location and handling, reduces manual intervention, and speeds up problem resolution.

[0088] Example 9: Based on Example 8, this example provides an intelligent analysis system for Android system upgrade anomalies. The analysis unit includes: Initial feature initialization subunit, used for: Based on the nodes and edges of the directed graph, the initial feature vector of each node is extracted and initialized. The initial feature vector of each node is formed by concatenating the embedding vectors of the predefined action subject, operation object and status code in the corresponding atomic event. The initial feature vector of the edge is represented by the causal probability value. An adjacency matrix is ​​constructed based on the connection relationships of the directed graph, whereby the adjacency matrix is ​​used to define the message propagation path between nodes; Feature-determining subunits are used for: Based on a graph neural network, message passing between nodes is performed according to the adjacency matrix, where each node gathers the feature information of all its neighboring nodes after weighting by edge weights; Based on the graph neural network, the feature information of the neighboring nodes after each node is aggregated and the feature of the current node itself are nonlinearly transformed. Based on the nonlinear transformation result, the feature information of the neighboring nodes and the feature of the current node itself are fused and updated to obtain the deep feature representation of the current node. Perform global pooling on the deep feature representations of all nodes in the directed graph to obtain the state feature vector during the Android system upgrade process.

[0089] In this embodiment, the initial feature vector refers to a vector composed of the embedded vectors of the predefined action subject, operation object and status code in the atomic event, used to represent the initial state of the node.

[0090] In this embodiment, the embedded vector refers to the continuous numerical vector representation that maps discrete symbols such as the action subject, the operation object, and the status code to the embedded vector through the embedding technology.

[0091] In this embodiment, the causal probability value refers to the probability measure of the causal relationship between atomic events represented by the edge, and is used to initialize the characteristics of the edge.

[0092] In this embodiment, the adjacency matrix refers to a matrix constructed based on the connection relationship of a directed graph. Its elements indicate whether there are edges between nodes and the direction of the edges, which are used to define the message propagation path between nodes.

[0093] In this embodiment, the message propagation path refers to the route for transmitting feature information between nodes in a graph neural network, determined by the adjacency matrix.

[0094] In this embodiment, edge weight refers to the causal probability value on the edge, which is used to weight the feature information of neighboring nodes during message transmission.

[0095] In this embodiment, a neighboring node refers to another node in the directed graph that is directly connected to the current node through an edge.

[0096] In this embodiment, nonlinear transformation refers to the nonlinear activation function used in graph neural networks to transform the converged features and their own features.

[0097] In this embodiment, fusion and updating refers to the process of combining the neighbor features that have undergone nonlinear transformation with the node's own features to generate a new feature representation of the node.

[0098] In this embodiment, deep feature representation refers to the high-level abstract features of nodes obtained through message passing and updating of graph neural networks.

[0099] In this embodiment, global pooling refers to the operation of aggregating (such as summing or averaging) the deep feature representations of all nodes to generate a global vector of fixed dimensions.

[0100] The beneficial effects of the above technical solution are as follows: by initializing the features of nodes and edges, constructing an adjacency matrix to define message propagation paths, using graph neural networks for message passing and feature fusion, extracting deep feature representations, and generating state feature vectors through global pooling, it can accurately capture the semantics and correlations of events during the upgrade process, providing high-quality feature representations for subsequent anomaly detection and root cause analysis, and improving analysis accuracy and efficiency.

[0101] Example 10: This example provides an intelligent analysis method based on Android system upgrade anomalies, such as... Figure 4 As shown, it includes: Step 1: When an Android system fails during an OTA upgrade, collect the corresponding system logs and program execution logs. Step 2: Package the collected system logs and program execution logs into a compressed file; Step 3: Request and obtain the target address and authentication information for uploading the compressed file from the server, and upload the compressed file to the server according to the target address and authentication information; Step 4: Decompress the compressed file on the server and extract key information from the decompression logs based on predefined keywords; Step 5: Input the key information into the preset AI analysis model for analysis to obtain the analysis results and handling suggestions for the upgrade failure event; Step 6: Save the analysis results, processing suggestions, and storage information of the compressed file together, and perform queries and log downloads based on the device identifier.

[0102] The working principle and beneficial effects of the above technical solution are as follows: It effectively solves the problems of poor timeliness of log capture, large volume of logs and difficulty in filtering, and inconvenient operation in the traditional OTA upgrade process; This technical solution automatically collects system logs and program running logs when an upgrade failure is detected by building a trigger mechanism in the upgrade program, packages them, and sends them to the cloud server through a segmented upload method; The server decompresses the log package, extracts key information based on predefined keywords, and imports it into an AI model for analysis to generate failure reasons and solution suggestions; Finally, the analysis results and suggestions are associated with the device identifier and stored, supporting querying details or downloading the original logs by device number; It effectively realizes the full automation and intelligence of log collection, uploading, analysis, and querying, improves the efficiency and accuracy of upgrade anomaly location, and provides data support for subsequent system optimization.

[0103] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An intelligent analysis system based on Android system upgrade anomalies, characterized in that, include: The log collection module is used to collect system logs and program execution logs corresponding to the upgrade process when a failure event occurs during the OTA upgrade of the Android system. The packaging module is used to package the collected system logs and program runtime logs into a compressed file; The upload module is used to request and obtain the target address and authentication information for uploading the compressed file from the server, and upload the compressed file to the server according to the target address and authentication information; The decompression module is used to decompress compressed files based on the server and extract key information from the decompression logs based on predefined keywords; The analysis module is used to input key information into a preset AI analysis model for analysis, and obtain analysis results and handling suggestions for upgrade failure events; The associated storage module is used to associate and save the analysis results, processing suggestions, and storage information of the compressed file, and to query and download logs based on the device identifier.

2. The intelligent analysis system based on Android system upgrade anomalies according to claim 1, characterized in that, The log collection module includes: The log collection unit is used to collect system logs and program execution logs corresponding to the upgrade process when a failure event is detected. The collected system logs include: The first system log is obtained from the system runtime cache by simulating log capture commands; When the OTA upgrade process is in recovery upgrade mode, the system log also includes a second system log obtained from the recovery partition that records details of the system installation process; The program execution log is obtained from a specified log file written during the runtime of the upgrade program.

3. The intelligent analysis system based on Android system upgrade anomalies according to claim 1, characterized in that, The upload module includes: The authentication information acquisition unit is used to acquire the authentication information from the server, including: cloud service provider identifier, storage node information, and temporary key; The upload connection building unit is used to build an upload connection with the target cloud storage service based on the cloud service provider identifier, storage node information, and temporary key. An upload unit is used to upload the compressed file to the server based on the target address and the upload connection.

4. The intelligent analysis system based on Android system upgrade anomalies according to claim 1, characterized in that, The decompression module is used for: The decompression unit is used to decompress the received compressed files on the server side; The retrieval unit is used to retrieve the decompressed log content based on a predefined set of keywords; Key information determination unit, used for: Extract the log line containing any keyword from the keyword set and the context information corresponding to the log line from the log content; Log lines and their corresponding context information are considered key information.

5. The intelligent analysis system based on Android system upgrade anomalies according to claim 1, characterized in that, The associated save module includes: The association unit is used to associate the analysis results and processing suggestions with the device identifier where the upgrade failure event occurred, and at the same time record the storage path of the compressed file on the server. A storage unit is used to store the associated result data and the storage path in the database; The download access unit is used to receive query requests based on the device identifier stored, retrieve and return the corresponding analysis results and processing suggestions from the database according to the device identifier, and provide download access to compressed files according to the associated storage path.

6. The intelligent analysis system based on Android system upgrade anomalies according to claim 3, characterized in that, Upload unit, including: The communication channel construction subunit is used to construct a communication channel with the target cloud storage service based on the target address and upload connection. The piecewise parameter constraints determine the sub-units, used for: Information such as the size of compressed files, the network type of the device that experienced the upgrade failure, and real-time bandwidth information are obtained based on the communication channel. Based on the storage node information corresponding to the target address and the cloud service provider identifier, determine the sharding parameter constraints supported by the target cloud storage service. The matching subunit is used to match the network type with the preset policy library to determine the baseline fragment size and baseline concurrency corresponding to the network type; The target fragment size determination subunit is used for: The upload duration of a single fragment is determined based on real-time bandwidth information, and the minimum number of fragments required to meet the target total upload duration is determined based on the size information of the compressed file and the upload duration of a single fragment. The minimum number of fragments is compared with the baseline concurrency, and the maximum result is taken as the final number of concurrent fragment uploads. The target fragment size is determined based on the fragmentation parameter constraints and the final number of concurrent fragment uploads.

7. The intelligent analysis system based on Android system upgrade anomalies according to claim 6, characterized in that, The upload unit further includes: The upload priority determination subunit is used to split the compressed file according to the target fragment size to obtain multiple data fragments. At the same time, it assigns upload priority to multiple data fragments according to the position of the data fragments in the compressed file. The data fragment upload subunit is used to upload multiple data fragments to the server storage location corresponding to the target address based on the final number of concurrent uploads and the upload priority according to the upload connection.

8. The intelligent analysis system based on Android system upgrade anomalies according to claim 1, characterized in that, The analysis module includes: Directed graph building unit, used for: An information processing engine is built based on a predefined grammar rule library, and the extracted key information is input into the information processing engine to map unstructured log text sequences into structured atomic event sequences. Based on the temporal relationship of atomic event sequences, the correlation between atomic events and the temporal dependency and causal probability between atomic events are determined, and the correlation is used as nodes and the temporal dependency and causal probability are used as edges to construct a directed graph. Analysis unit, used for: The directed graph is input into the first-level network of the preset AI analysis model, and deep feature representations of nodes and edges in the directed graph are extracted based on the first-level network. The state feature vector in the Android system upgrade process is determined based on the deep feature representations. The first-level network is a graph neural network. The state feature vector is compared with the baseline vector of the pre-established normal upgrade process, and the deviation clusters in the feature space are located based on the second-level network of the preset AI analysis model. Based on the location results, the original directed graph is traced to determine the corresponding key nodes and edges. Based on the determination results, the initial abnormal events and subsequent propagation paths corresponding to the state deviation are identified. The second-level network is an anomaly detection and attribution network. The third-level network based on the AI ​​analysis model calls the internally stored fault mode knowledge graph to perform matching analysis on the initial abnormal event and subsequent propagation path, and obtains the analysis results containing the fault link based on the matching analysis results. The fault mode knowledge graph associates abnormal events, system component states and potential root causes in the form of logical rules, and the third-level network is a root cause reasoning network. The suggestion generation unit is used to generate a library of handling solutions associated with each abnormal node in the fault link based on the dynamic index of the fault link, and to sort and logically connect the handling solutions of each node in the fault link to obtain a sequence of handling suggestions.

9. The intelligent analysis system based on Android system upgrade anomalies according to claim 8, characterized in that, Analysis unit, including: Initial feature initialization subunit, used for: Based on the nodes and edges of the directed graph, the initial feature vector of each node is extracted and initialized. The initial feature vector of each node is formed by concatenating the embedding vectors of the predefined action subject, operation object and status code in the corresponding atomic event. The initial feature vector of the edge is represented by the causal probability value. An adjacency matrix is ​​constructed based on the connection relationships of the directed graph, whereby the adjacency matrix is ​​used to define the message propagation path between nodes; Feature-determining subunits are used for: Based on a graph neural network, message passing between nodes is performed according to the adjacency matrix, where each node gathers the feature information of all its neighboring nodes after weighting by edge weights; Based on the graph neural network, the feature information of the neighboring nodes after each node is aggregated and the feature of the current node itself are nonlinearly transformed. Based on the nonlinear transformation result, the feature information of the neighboring nodes and the feature of the current node itself are fused and updated to obtain the deep feature representation of the current node. Perform global pooling on the deep feature representations of all nodes in the directed graph to obtain the state feature vector during the Android system upgrade process.

10. An intelligent analysis method based on Android system upgrade anomalies, characterized in that, include: When an Android system fails during an OTA upgrade, collect the corresponding system logs and program execution logs. Package the collected system logs and program execution logs into a compressed file; Request and obtain the target address and authentication information for uploading the compressed file from the server, and upload the compressed file to the server according to the target address and authentication information; The server decompresses the compressed file and extracts key information from the decompression log based on predefined keywords. Key information is input into a preset AI analysis model for analysis, resulting in analysis results and handling suggestions for upgrade failure events; The analysis results, processing suggestions, and storage information of the compressed file are associated and saved, and queries and log downloads are performed based on the device identifier.

Images