Abnormality repairing method and electronic equipment

Abstract

The invention discloses an exception repairing method and electronic equipment. The method comprises the steps that in response to monitoring that a target abnormal event occurs in the operation process of the electronic equipment, target key data is acquired from the target abnormal event, and the target key data at least can represent an abnormal problem occurring in the electronic equipment; generating a target repair scheme based on a target retrieval result corresponding to the target key data, wherein the target repair scheme comprises at least one instruction set for repairing the abnormal problem; and based on the target repair scheme, executing a repair operation on a target component of the electronic equipment, wherein the target component is a component associated with the abnormal problem.

Description

Technical Field

[0001] This application relates to the field of problem repair technology, and in particular to an anomaly repair method and electronic device. Background Technology

[0002] Anomalies are common during computer system operation, especially when new software or games are run for the first time. These anomalies may involve various factors such as hardware incompatibility, driver conflicts, and system configuration errors, affecting system stability and user experience. To improve system availability and stability, these anomalies need to be identified and fixed promptly.

[0003] In related technologies, automated repair tools based on preset rule bases are commonly used to handle common anomalies. These tools achieve rapid responses to some anomalies by matching known failure modes and executing fixed repair strategies. However, when faced with new or complex anomalies, their success rate is low due to a lack of flexible data analysis capabilities and targeted repair mechanisms.

[0004] Furthermore, when automatic repair fails, users often rely on unstructured information sources such as online forums to obtain solutions. However, these information sources lack unified standards and accuracy, causing users to spend a lot of time identifying effective solutions and making it difficult to ensure that the selected solution is suitable for the current device environment, thus affecting repair efficiency. Summary of the Invention

[0005] In view of this, this application provides an anomaly repair method and an electronic device.

[0006] The technical solution of this application embodiment is implemented as follows:

[0007] This application provides an anomaly repair method, the method including:

[0008] In response to the detection of a target abnormal event during the operation of an electronic device, key target data is obtained from the target abnormal event. The key target data can at least characterize the abnormal problem that occurred in the electronic device.

[0009] A target repair plan is generated based on the target retrieval results corresponding to the target key data. The target repair plan includes at least one set of instructions for repairing abnormal problems.

[0010] Repair operations are performed on target components of electronic devices based on the target repair plan. The target component is the component associated with the abnormal problem.

[0011] In the above-mentioned anomaly repair method, in response to the detection of a target anomaly event during the operation of an electronic device, target key data is obtained from the target anomaly event, including at least one of the following: obtaining the event type of the target anomaly event, determining the intelligent agent matching the event type, and using the intelligent agent to extract the target key data from the monitoring data corresponding to the event type; in response to the detection of a target anomaly event during the operation of an electronic device, obtaining the target log data corresponding to the target anomaly event, and obtaining the target key data from the target log data; in response to the detection of a target anomaly warning prompt during the operation of an electronic device, obtaining the component operation data of the component associated with the target anomaly warning prompt, and obtaining the target key data from the component operation data.

[0012] In the above-mentioned anomaly repair method, in response to the detection of a target anomaly event during the operation of the electronic device, target key data is obtained from the target anomaly event, including at least one of the following: in response to the detection of a target anomaly interface during the operation of the electronic device, the image data corresponding to the target anomaly interface and the content of the first prompt word are input to the first intelligent agent, so that the first intelligent agent can extract key information from the target anomaly interface with reference to the content of the first prompt word to obtain target key data; in response to the detection of a target anomaly interface during the operation of the electronic device, the first intelligent agent is used to identify and analyze the target anomaly interface, and the obtained multi-dimensional information is processed into target key data according to the target field format; in response to the detection of a target anomaly interface during the operation of the electronic device, the first intelligent agent is activated to call at least one processing model to analyze the interface information in the target anomaly interface, and target key data is extracted from the interface information with reference to the content of the first prompt word.

[0013] In the above-mentioned anomaly repair method, a target repair plan is generated based on the target retrieval results corresponding to the target key data, including: constructing a target retrieval strategy based on the target key information, and processing the retrieval data obtained by retrieving from the target database through the target retrieval strategy into target retrieval results; and generating a target repair plan based on the target retrieval results and the content of the second prompt words.

[0014] In the above-mentioned anomaly repair method, constructing a target retrieval strategy based on target key information includes at least one of the following: constructing a target retrieval strategy based on target key information and first operation log data of electronic devices, wherein the first operation log data includes operation log data of target components corresponding to the target key information and operation log data of electronic devices associated with the target components; constructing a target retrieval strategy based on target key information and second operation log data of electronic devices, wherein the second operation log data includes log data corresponding to target anomalies; and constructing a target retrieval strategy based on target key information using a first intelligent agent or a second intelligent agent; wherein the target retrieval strategy includes search keywords, search formulas, search methods, search weights, and the most important search terms in the search database. One or more; and / or, processing the retrieval data obtained in the target database through the target retrieval strategy into target retrieval results, including at least one of the following: when the retrieval data generated by the first agent or the second agent meets the first condition, performing a retrieval operation in the target database using the target retrieval strategy, and merging the retrieved result data into target retrieval results; performing a retrieval in multiple databases using the target retrieval strategy, and merging the retrieved result data according to the corresponding configured weights to obtain target retrieval results; performing a retrieval in multiple databases using the target retrieval strategy, sorting the retrieved result data according to the score, and processing the result data located in the target sequence into target retrieval results.

[0015] In the above-mentioned anomaly repair method, a target repair scheme is generated based on the target retrieval results corresponding to the target key data, including: using a second intelligent agent to construct a target retrieval strategy based on the target key information and its corresponding log data, and processing the retrieval data obtained by retrieving in the target database through the target retrieval strategy into target retrieval results; and using a third intelligent agent or a second intelligent agent to generate a target repair scheme based on the target retrieval results and the content of the second prompt words.

[0016] In the above-mentioned anomaly repair method, generating a target repair scheme based on the target retrieval results and the content of the second prompt words includes at least one of the following: inputting the target retrieval results and the content of the second prompt words into a second intelligent agent or a third intelligent agent, so that the second intelligent agent or the third intelligent agent can generate and process the target retrieval results with reference to the content of the second prompt words to obtain a target repair scheme; determining repair parameters for the anomaly problem of the target component based on the target retrieval results and the content of the second prompt words, and generating a target repair scheme for repairing the anomaly problem of the target component based on the repair parameters; wherein, the target repair scheme includes at least an instruction set and / or automatic repair code for calling the target repair tool.

[0017] In the above-mentioned anomaly repair method, the repair operation on the target component of the electronic device is performed based on the target repair scheme, including at least the following: using a fourth intelligent agent to perform the repair operation on the target component of the electronic device based on the target repair scheme; determining the target interactive component in the electronic device for interaction based on the target repair scheme, controlling the target interactive component to perform the operation corresponding to the target repair scheme to repair the target component; calling the corresponding target repair tool to repair the target component based on the target repair scheme, or executing the automatic repair code in the target repair scheme to repair the target component.

[0018] The above-mentioned anomaly repair method further includes at least one of the following: in response to obtaining the result of failure to repair the target component, inputting the result and the target repair plan to the first intelligent agent, and re-executing the repair process of the target component; after monitoring the target anomaly event, outputting a first prompt message, and the first intelligent agent can respond to the first feedback data of the target user to perform the operation of obtaining the target key data; after obtaining the target repair plan, outputting a second prompt message, and the third or fourth intelligent agent can respond to the second feedback data of the target user to perform the operation of repairing the target component.

[0019] This application provides an electronic device, including at least one processor and at least one intelligent agent capable of running on the at least one processor. The at least one intelligent agent is capable of invoking at least one processing model deployed in the electronic device to perform the following operations:

[0020] In response to the detection of a target abnormal event during the operation of an electronic device, key target data is obtained from the target abnormal event. The key target data can at least characterize the abnormal problem that occurred in the electronic device.

[0021] A target repair plan is generated based on the target retrieval results corresponding to the target key data. The target repair plan includes at least one set of instructions for repairing abnormal problems.

[0022] Repair operations are performed on target components of electronic devices based on the target repair plan. The target component is the component associated with the abnormal problem.

[0023] In the aforementioned electronic device, at least one intelligent agent is also capable of invoking at least one processing model to perform the following operations: after monitoring a target anomaly event, locating the target code data where the anomaly occurred based on the knowledge graph data associated with the key data in the target anomaly event, and generating automatic repair code for fixing the anomaly based on the target code data; or,

[0024] When target input data for a target component of an electronic device is identified, a target code file matching the user intent represented by the target input data is generated using a target processing model. The target code file can be used to optimize the target code data of the target component to optimize the functional services that the target component can provide.

[0025] In the aforementioned electronic device, the process of generating a target code file that matches the user intent represented by the target input data using a target processing model includes: obtaining optimization suggestions generated by the target processing model for the target prompt word data; obtaining a code generation template generated by the target processing model with reference to the target knowledge graph data; processing the optimization suggestions into an initial code file using the code generation template; reviewing and / or testing the initial code file; and outputting the initial code file as a target code file if the initial code file passes the review and / or testing.

[0026] Alternatively, it may include: constructing a target knowledge graph based on the target optimization scheme associated with the target prompt word data, wherein the target knowledge graph includes at least the mapping relationship between user intent and corresponding knowledge; obtaining target knowledge data from the target knowledge graph based on the target code generation template; and generating a target code file based on the target knowledge data according to at least one of the method call relationship, data dependency relationship or event sequence relationship in the target code generation template.

[0027] This application provides a computer-readable storage medium that stores one or more computer programs, which can be executed by one or more processors to implement the above-described anomaly repair method.

[0028] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the technical solutions provided in the embodiments of this application. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of this application, 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 application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0030] Figure 1 A flowchart illustrating an anomaly repair method provided in an embodiment of this application;

[0031] Figure 2 This application provides an exemplary flowchart for obtaining target key data. Figure 1 ;

[0032] Figure 3 This application provides an exemplary flowchart for obtaining target key data. Figure 2 ;

[0033] Figure 4 A schematic flowchart illustrating the generation of a target repair solution provided in this application embodiment. Figure 1 ;

[0034] Figure 5 This application provides an exemplary flowchart for constructing a target retrieval strategy. Figure 1 ;

[0035] Figure 6 This application provides an exemplary flowchart for constructing a target retrieval strategy. Figure 2 ;

[0036] Figure 7 A schematic flowchart illustrating the generation of a target repair solution provided in this application embodiment. Figure 2 ;

[0037] Figure 8 A schematic flowchart illustrating the generation of a target repair solution provided in this application embodiment. Figure 3 ;

[0038] Figure 9 A flowchart illustrating an exemplary implementation of a target repair scheme is provided in this application embodiment. Figure 1 ;

[0039] Figure 10 A flowchart illustrating an exemplary implementation of a target repair scheme is provided in this application embodiment. Figure 2 ;

[0040] Figure 11 A schematic diagram of an exemplary user feedback page provided for embodiments of this application. Figure 1 ;

[0041] Figure 12 A schematic diagram of an exemplary user feedback page provided for embodiments of this application. Figure 2 ;

[0042] Figure 13 A flowchart illustrating an exemplary anomaly repair method provided in this application embodiment;

[0043] Figure 14 This is a schematic diagram illustrating an exemplary process for generating code files, provided as an embodiment of this application. Detailed Implementation

[0044] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It is understood that the specific embodiments described herein are merely for explaining the relevant application and not for limiting the application. Furthermore, it should be noted that, for ease of description, only the parts relevant to the application are shown in the accompanying drawings.

[0045] This application provides an anomaly repair method, implemented by an electronic device, such as... Figure 1 As shown, the process includes the following steps S101 to S103:

[0046] Step S101: In response to the detection of a target abnormal event during the operation of the electronic device, obtain target key data from the target abnormal event. The target key data can at least characterize the abnormal problem that occurred in the electronic device.

[0047] In the embodiments of this application, the target abnormal event can be an abnormal event determined by the electronic device through monitoring tools. For example, the target abnormal event can be: abnormal pop-up prompt, system error, blue screen, system high temperature alarm, abnormal sound acquisition or output, abnormal system or application update, abnormal network connection, abnormal hardware compatibility, abnormal camera, etc., or other events that prevent the electronic device from operating normally.

[0048] For example, if an abnormal pop-up is captured by a monitoring tool (e.g., Windows Event Viewer or Process Monitor), it indicates a target abnormal event during the operation of the electronic device: an abnormal pop-up notification; if the CPU or GPU temperature is monitored in real time by a monitoring tool (e.g., HW Monitor or a temperature sensor), and the temperature exceeds a preset threshold, it indicates a target abnormal event: a system high-temperature alarm; if an abnormal sound is detected by a monitoring tool (e.g., an AI (artificial intelligence) voiceprint recognition system) through spectrum analysis, it indicates a target abnormal event: abnormal sound output; if a network connection failure or TCP retransmission / DNS resolution failure is detected by a monitoring tool (e.g., Ping, Traceroute, Wireshark), it indicates a target abnormal event: abnormal network connection; if the hardware status is checked by a monitoring tool (e.g., Device Manager), or if the hardware specifications and compatibility list are analyzed using a monitoring tool (e.g., CPU-Z / HWinfo), and a hardware abnormality or incompatibility is detected, it indicates a target abnormal event: hardware incompatibility.

[0049] In the embodiments of this application, after monitoring an abnormal event that occurs during the operation of an electronic device, target key data can be obtained from the abnormal event. For example, the target key data may be error codes, error reason keywords, pop-up window content, abnormal logs, abnormal sensor data, abnormal operation data, etc.

[0050] For example, if the target abnormal event is an abnormal pop-up notification, you can view the "Windows Log" or "Application and Services Log" through Windows Event Viewer, and then select the event identifier, source, or detection filter to view detailed information, which may include: error description, error path, etc. Alternatively, you can also extract the content of the abnormal pop-up notification window to determine the target key data. An exemplary method is to capture an image of the abnormal pop-up notification window and identify the pop-up content from the image. For example, if the pop-up window displays "DirectX not initialized correctly," the electronic device will extract keywords such as "DirectX" and "initialization failure" as the target key data.

[0051] For example, if the target abnormal event is a system high temperature alarm, the electronic device can obtain the CPU or GPU temperature data monitored by the temperature sensor and its relationship with a preset threshold, or abnormal sensor data such as the fan speed monitored by the fan speed sensor, as key target data. This key target data indicates that the system temperature of the electronic device is too high, that is, it can pinpoint which component or function of the electronic device is malfunctioning.

[0052] For example, if the target exception event is a blue screen, the error code will be displayed directly on the screen. The error code can be obtained directly or from the event viewer. For instance, the obtained error code could be 0x0000007E (SYSTEM_THREAD_EXCEPTION_NOT_HANDLED) or 0x000000F4 (CRITICAL_OBJECT_TERMINATION), representing the specific function that is malfunctioning. If the obtained error code is 0x0000007E (SYSTEM_THREAD_EXCEPTION_NOT_HANDLED), it indicates that a driver or system service triggered an unhandled exception. If the obtained error code is 0x000000F4 (CRITICAL_OBJECT_TERMINATION), it indicates that a critical process terminated unexpectedly. In other words, the target critical data can pinpoint the specific exception problem currently occurring in the electronic device.

[0053] For example, if the target exception event is a game launch failure, the target key data obtained from the target exception event may include specific descriptions such as the inability to load D3DCompiler_47.dll or an outdated graphics card driver version.

[0054] For example, based on key target data, an electronic device can pinpoint what kind of abnormal problem has occurred, or at least pinpoint which component or function of the electronic device is malfunctioning, such as unavailable Wi-Fi, overheating system, inability to activate the camera, inability to load or lag of a certain application, etc.

[0055] In the embodiments of this application, the electronic device is a device with anomaly repair function, which may be a tablet computer, a laptop computer, a handheld computer, a personal digital assistant (PDA), a desktop computer, etc. No specific electronic device is limited here.

[0056] Step S102: Generate a target repair plan based on the target retrieval results corresponding to the target key data. The target repair plan includes at least one instruction set for repairing abnormal problems.

[0057] In the embodiments of this application, after the electronic device determines the target key data, it can use a retrieval agent to construct a retrieval strategy based on the target key data, perform a retrieval in the target database, and then organize the different retrieval results to obtain the target retrieval results.

[0058] For example, if the target key data is a system overheating alarm, then the target key data can be divided into structured data (such as error codes, device models) and unstructured data (such as log text or user feedback). The core data includes "high temperature alarm", "electronic devices", "temperature sensor data", and "heat dissipation system status". Of course, "overheating" can also be expanded to include "temperature threshold exceeded", "heat dissipation failure", etc. If the target key data includes "blue screen", it can be broken down into "driver conflict", "memory error", etc. Then, a search expression can be built based on the core data or the expanded data. The search expression can be a simple combination or a high-level search expression can be constructed using wildcards. Then, different search results can be obtained by searching the target database based on the constructed search expression.

[0059] For example, the target key data is: frequent high-temperature alarms from the server, and the cause needs to be identified. The search query is constructed as follows: Extract keywords: high-temperature alarm, server, temperature sensor, cooling system; Expand keywords: overheating, fan failure, CPU utilization, data center environment; Construct the search query: "high-temperature alarm" AND "server" AND ("temperature sensor" OR "cooling system").

[0060] In the embodiments of this application, when constructing a search query based on target key data, key event information recorded by the system is also combined, such as game installation events, the latest system update record, hardware configuration change records, etc., and the search is performed on platforms (target databases) such as professional forums, knowledge bases, and manufacturer websites via network connection. For example, if the target key data is DirectX initialization failure, the search keywords may include DirectX cannot be initialized, DirectX version incompatibility, graphics card driver conflict, etc.

[0061] In the embodiments of this application, the search results obtained by the electronic device based on the target key data may include technical documents, user experience sharing, official solutions, etc. The electronic device can organize the technical documents, user experience sharing, official solutions, etc. to obtain the target search results.

[0062] For example, after obtaining different search results, electronic devices can also filter and organize these results, such as by deduplication, sorting, setting weights, and performing data fusion processing, to obtain the target search result. For instance, based on the search results, results with high similarity to the search query can be identified as the target search result, or results with higher weights can be selected from the different search results based on their weights.

[0063] In the embodiments of this application, the electronic device generating a target repair solution based on the target retrieval results may include at least one instruction set for repairing abnormal problems. This instruction set may be a single operation instruction or a program instruction set consisting of multiple operation instructions. For example, it may be a single instruction or a set of multiple instructions for downloading a repair tool, or it may be repair software or a repair tool automatically generated based on an intelligent agent, which can be considered a program instruction set. It may also be instructions that communicate with the BIOS or corresponding hardware of the electronic device, or an instruction set that generates hardware or software configuration parameters. At least one instruction set exists in the form of a Windows batch script (repair bat script) for automating the execution of a series of commands.

[0064] For example, if the cause of the excessively high system temperature is abnormal fan speed, then the target repair solution includes at least one set of instructions that may include instructions for generating instructions containing fan speed configuration parameters and instructions for updating fan speed configuration parameters.

[0065] For example, if the blue screen on an electronic device is caused by a DirectX error, then the target repair solution may include at least one set of instructions such as downloading and installing a DirectX repair tool, selecting the "Detect and Repair" button, restarting the computer after the repair is completed, and verifying whether the problem is resolved.

[0066] Step S103: Perform repair operations on the target component of the electronic device based on the target repair plan. The target component is the component associated with the abnormal problem.

[0067] In the embodiments of this application, an intelligent agent interacts with the corresponding target component. If there is a hardware problem with the electronic device, the hardware variables can be updated by interacting with the BIOS. For example, if the cause of the system overheating is abnormal fan speed, then the target repair scheme includes at least one instruction set that can generate instructions containing fan speed configuration parameters and update fan speed configuration parameters. Then, the fan repair operation is to reconfigure the hardware (fan) configuration parameters or adjust the hardware configuration parameters to solve the problem of overheating.

[0068] For example, if the target abnormal event of an electronic device is a system startup problem, and the target key data extracted is: unable to boot, black screen, error, then the corresponding repair operations can be: adjusting the boot order, updating the BIOS, enabling hardware compatibility mode, etc.

[0069] In the embodiments of this application, if a software problem occurs, the problem can be solved by changing the application's configuration parameters or the operating parameters of the hardware on which the application depends through interaction between the intelligent agent and the corresponding application. This also includes calling or downloading relevant repair tools based on program instruction sets to repair the target component, such as using virus scanning tools or driver repair tools to scan for and remove viruses or repair drivers. For example, if at least one instruction set is to download and install a DirectX repair tool, select the "Detect and Repair" button, restart the computer after repair, and verify whether the problem is resolved, etc., the DirectX repair operation is performed, and the performed repair operation is a firmware update.

[0070] In embodiments of this application, the electronic device can also be an intelligent agent that directly calls the plugin in the file package to perform repair. For example, if the target abnormal event is that the electronic device cannot play audio and video files of a specific format (such as MKV, FLAC), and the determined target search result is that the native decoder does not support the specific format, then the VLC plugin can be installed. In this way, the media player on the electronic device can automatically recognize and play MKV files. Alternatively, if the target abnormal event is that the electronic device cannot recognize or write to an NTFS formatted USB flash drive, and the determined target search result is: the system only supports FAT32 by default, then the corresponding repair operation can be: install NTFS-3G through the plugin manager, and after restarting the USB flash drive, it can operate normally.

[0071] For example, repair operations may also include: hardware replacement, firmware re-flashing, firmware recovery, parameter tuning, etc.

[0072] In embodiments of this application, the target component is an internal component of the electronic device (e.g., the camera or microphone itself) or a software module (e.g., a camera firmware update or a camera application) directly related to the anomaly. For example, for a DirectX initialization failure, the target component might be a graphics card driver, the DirectX runtime library, or related system services. The corresponding repair operations typically involve modifying, replacing, or updating the target component to restore normal system operation.

[0073] In the embodiments of this application, the instruction set for repairing abnormal problems varies depending on the type of the target component. If it is an internal component of an electronic device, it may be necessary to replace the hardware. If the software program has a problem, it is necessary to rewrite the program, adjust the parameters, etc.

[0074] In the embodiments of this application, the target component can also be a related component that causes the abnormality. For example, the target file for high temperature alarm repair can be a heat dissipation component, a heat dissipation control application, or a camera firmware update or a camera application repair. Different target components correspond to different instruction sets for repairing abnormal problems.

[0075] Thus, in the event of an electronic device malfunction, the embodiments of this application achieve fully automated processing from abnormal event monitoring, key information extraction, repair plan generation to repair operation execution. The anomaly repair method provided by the embodiments of this application not only improves the intelligence level and processing efficiency of anomaly repair, but also reduces manual intervention, lowers the consumption of technical support resources, and enhances the user experience.

[0076] In some embodiments, when the electronic device performs step S101 above, it "in response to monitoring that a target abnormal event occurs during the operation of the electronic device, and obtains target key data from the target abnormal event," such as... Figure 2 As shown, the following step S201 can also be performed:

[0077] Step S201: Obtain the event type of the target abnormal event, determine the intelligent agent that matches the event type, and use the intelligent agent to extract the target key data from the monitoring data corresponding to the event type.

[0078] In the embodiments of this application, corresponding intelligent agents can be set for different event types. An intelligent agent refers to an artificial intelligence model or program module with specific functions. In this way, when a target abnormal event is determined, an intelligent agent matching the event type can be determined. For example, if the target abnormal event is an abnormal problem presented on the interface, such as a blue screen or an abnormal pop-up prompt, an intelligent agent with image recognition and input can be used for identification and analysis. For example, the way to extract the target key data from the monitoring data corresponding to the event type can be the image monitoring data obtained by taking a screenshot of the abnormal scene, including but not limited to screenshot images, interface layout information, and related text content.

[0079] In the embodiments of this application, if the anomaly is not visible, such as excessively high system temperature or hardware failure, an intelligent agent capable of directly reading and analyzing log data can be used. For example, the way to extract the target key data from the monitoring data corresponding to the event type can be by reading the monitoring log data from EC, BIOS or other memory through the corresponding interface.

[0080] In the embodiments of this application, the event type is the type of abnormal event, which can identify which type of abnormality. For example, interface pop-up type abnormality, system error type abnormality, hardware operation abnormality, etc.

[0081] In the embodiments of this application, the intelligent agent can also be a multimodal data input agent, that is, the same intelligent agent can correspond to various event types. That is, the intelligent agent is compatible with multimodal input intelligent agents such as image recognition, text analysis or log parsing, so as to adapt to different input data and processing requirements.

[0082] In the embodiments of this application, the event type can also be a system-level anomaly, a hardware-level anomaly, a software-level anomaly, a network-level anomaly, or a storage-level anomaly. In this way, a corresponding intelligent agent is set up for different event types, and the corresponding monitoring data is obtained. For example, for system-level anomalies, the monitoring data that can be obtained can be based on system logs, and for hardware-level anomalies, the corresponding monitoring data is CPU / GPU temperature, fan speed, voltage data, memory data, etc.

[0083] In other embodiments, the electronic device, in performing step S101 above, "in response to monitoring that a target abnormal event occurs during the operation of the electronic device, obtains target key data from the target abnormal event," such as... Figure 2 As shown, the following step S202 can also be performed:

[0084] Step S202: In response to the detection of a target abnormal event during the operation of the electronic device, obtain the target log data corresponding to the target abnormal event, and extract the target key data from the target log data.

[0085] In the embodiments of this application, target log data refers to records generated by the system or application within the time range of the target abnormal event, used to describe the behavioral state of the electronic device before and after the abnormality occurred. This type of data typically includes timestamps, error codes, stack traces, system call sequences, memory usage, etc., and has high information density and structured characteristics. Regardless of the event type, the system will directly read and analyze the target log data that matches the time range of the target abnormal event, and extract the target key data from the target log data.

[0086] For example, if the target abnormal event is a system failure, the electronic device can obtain the target log data corresponding to the system, open the Windows Event Viewer (eventvwr.msc), filter error level events, locate key logs (such as the BugCheck event corresponding to a blue screen, and the Application Error event corresponding to a software crash), and extract the event ID, source, and description information (such as the Windows Error Report corresponding to event ID 1001).

[0087] For example, if the target abnormal event is a software malfunction, the electronic device can use ProcessMonitor to monitor process behavior in real time, capture abnormal file access and registry operations, run 360 System Rescue or Huorong Security, scan for malware and system file errors, check application logs (such as Apache's error.log and MySQL's mysqld.log), and then extract the target key data from the application logs.

[0088] In some other embodiments, the electronic device, in performing step S101 above, "in response to monitoring that a target abnormal event occurs during the operation of the electronic device, obtains target key data from the target abnormal event," such as... Figure 2 As shown, the following step S203 can also be performed:

[0089] Step S203: In response to the detection of a target anomaly warning during the operation of the electronic device, obtain the component operation data of the component associated with the target anomaly warning, and obtain the target key data from the component operation data.

[0090] In the embodiments of this application, if an electronic device issues a target abnormality warning during operation, for example, if the target abnormality warning indicates that the current camera cannot be raised, or that it is not connected to the network, the component operation data of the component associated with the target abnormality warning can be directly obtained, such as the operation data related to the camera and the network connectivity data.

[0091] In the embodiments of this application, by monitoring the operation of electronic devices in real time and providing early warnings of faults or abnormal operations based on the operating status of each operating component, the relevant operating data of the component can be located based on the warning, such as operating logs and user operation data or interaction data of the component. Then, the target key data can be obtained by analyzing these data.

[0092] For example, if the target anomaly warning indicates that the current camera cannot be raised, the camera's operating data can be located based on this warning, and then the operating data can be analyzed to obtain key target data (camera malfunction, camera pop-up component failure). If the target anomaly warning indicates that the microphone cannot be recognized, the microphone-related operating data or user operation data on the microphone can be directly obtained, and then the key target data (microphone disconnected from electronic device) can be extracted from it.

[0093] For example, the component operation data associated with the target anomaly warning can be the operation data of a software component or the operation data of a hardware component. For example, if the target anomaly warning is that the game is running abnormally, the running status, running logs, and player operation data of the game application can be obtained as component operation data; if the target anomaly warning is that the camera cannot take pictures, the working data of the camera can be obtained as component operation data.

[0094] For example, if the warning message indicates that the GPU temperature is too high, the associated components may include the graphics card driver, cooling fan, power management module, etc. Each component may generate corresponding operating data, and this component operating data can be obtained.

[0095] In the embodiments of this application, after acquiring component operation data, the electronic device can analyze and extract target key data from it. For example, if the target abnormality warning indicates a game operation abnormality, the acquired component operation data is the operation log. The type of abnormal log is identified from the operation log, including but not limited to warning logs, error logs, and security logs. Then, the corresponding target key data is analyzed for different types of logs. For example, if the operation log contains "Connection Timeout", then the network terminal or server response timeout is taken as the target key data. If the operation log contains "Cheat Engine Detected", then the cheat program is taken as the target key data. If the operation log contains "NullReferenceException" (Unity engine), then the object is called without initialization as the target key data.

[0096] The entity that performs the acquisition of target abnormal data and target key data in steps S201, S202 and S203 above can be the first intelligent agent in the electronic device or a multimodal intelligent agent in the electronic device.

[0097] In this way, through multi-level and multi-dimensional data collection and analysis, the system can efficiently identify and promptly handle abnormal events of electronic devices, thereby improving system stability and user experience.

[0098] In some embodiments, when the electronic device performs step S101 above, it "in response to monitoring that a target abnormal event occurs during the operation of the electronic device, and obtains target key data from the target abnormal event," such as... Figure 3 As shown, the following steps S301 can also be performed:

[0099] Step S301: In response to the detection of a target abnormal interface appearing during the operation of the electronic device, the image data corresponding to the target abnormal interface and the content of the first prompt word are input to the first intelligent agent, so that the first intelligent agent can extract key information from the target abnormal interface with reference to the content of the first prompt word and obtain target key data.

[0100] In the embodiments of this application, the target abnormal interface refers to a pop-up interface caused by abnormal operation or abnormal problems during the operation of an electronic device due to software conflicts, system errors, hardware compatibility issues, etc. For example, error windows popped up by the system, blue screen interfaces, application crash prompt interfaces, driver loading failure interfaces, etc. are all target abnormal interfaces.

[0101] In the embodiments of this application, the first prompt word content can be a preset fixed prompt word content, or it can be prompt word data generated according to the type of abnormal interface; for example, if the abnormal interface is a prompt in the form of a pop-up window, then the first prompt word content (preset prompt) can be: Please understand the interface in this scenario that is most like an abnormal pop-up window, and output the key information of the pop-up window.

[0102] For example, if the type of the abnormal interface is a blue screen, the first prompt word (preset prompt) could be: Please understand the blue screen interface in this scenario and output the key information of the blue screen.

[0103] In the embodiments of this application, if a target abnormal event is detected during the operation of an electronic device, and if the target abnormal event is displayed on the interface, a screenshot of the current display interface can be taken (e.g., using Windows' Snipping Tool or a third-party tool Snagit) to obtain image data. Alternatively, image data of the current display interface can be obtained through an image acquisition device such as a camera or a webcam. The obtained image data and the content of the first prompt word are then input into the first intelligent agent. The first intelligent agent can extract key information from the target abnormal interface by referring to the content of the first prompt word to obtain key target data.

[0104] In the embodiments of this application, the target abnormal interface may be a pop-up abnormal window or a system error prompt window (such as a pop-up window or the entire interface for system high temperature alarm, abnormal sound acquisition or output, abnormal system or application update, abnormal network connection, abnormal hardware compatibility, abnormal camera, etc.); a blue screen or system freeze, application freeze interface, or system / application error interface.

[0105] In the embodiments of this application, since the malfunction of the electronic device is manifested on the visual interface, the corresponding first intelligent agent is either a problem understanding intelligent agent or an image recognition intelligent agent. The image recognition intelligent agent is a dedicated intelligent agent set based on the event type, i.e., the type of the visual interface, and is capable of extracting key problem descriptions from complex interface information, such as error codes and error reason keywords.

[0106] Of course, the first intelligent agent can also be a problem-understanding intelligent agent, capable of outputting corresponding key information for different inputs. The problem-understanding intelligent agent can input text information, error codes, or other multi-dimensional data.

[0107] In this way, by inputting the image data and the first prompt word corresponding to the target abnormal interface into the first intelligent agent, the first intelligent agent can extract key information from the target abnormal interface, thereby accurately identifying the core problem in the abnormal interface and improving the efficiency of problem localization.

[0108] In other embodiments, the electronic device, in performing step S101 above, "in response to monitoring that a target abnormal event occurs during the operation of the electronic device, obtains target key data from the target abnormal event," such as... Figure 3 As shown, the following steps S302 can also be performed:

[0109] Step S302: In response to the detection of a target abnormal interface during the operation of the electronic device, the first intelligent agent is used to identify and analyze the target abnormal interface, and the obtained multi-dimensional information is processed into target key data according to the target field format.

[0110] In the embodiments of this application, if an abnormal pop-up or system error message appears during the operation of an electronic device, the first intelligent agent can obtain key information about the target from the target abnormal interface in the form of image data and the content of the first prompt word as in step S301. Of course, in the case of a target abnormal interface, the first intelligent agent can also directly identify and analyze the abnormal pop-up or system error message to obtain multi-dimensional information. For example, if the abnormal pop-up is a pop-up indicating an abnormal operation of a game application, the first intelligent agent can use the interface or SDK provided by the game application to obtain the pop-up content by calling relevant functions. For example, some applications provide API interfaces that allow the acquisition of the text information of the pop-up. For example, if the target abnormal pop-up is a web page pop-up, the first intelligent agent can call an automated testing tool such as Selenium to simulate browser operation and capture the pop-up content.

[0111] In the embodiments of this application, the target abnormal interface typically includes key information such as error codes, warning texts, and icon prompts. This key information reflects the specific fault state of the current system or application. The first intelligent agent can identify and analyze the abnormal information in the Windows GUI interface to obtain multi-dimensional information, including the text content of pop-ups, interface layout, and icons. This multi-dimensional information is then processed into target key data according to the target field format. The multi-dimensional information includes, but is not limited to, visual elements such as text content, icon styles, layout structure, and color changes in the target abnormal interface, as well as dynamic information such as the time of appearance, duration, and whether other abnormal behaviors accompany the target abnormal interface. Together, these multi-dimensional information constitute a comprehensive description of the target abnormal interface.

[0112] In the embodiments of this application, the first intelligent agent can monitor the display interface in real time. After the target abnormal interface appears, the first intelligent agent calls the window handle to obtain a screenshot of the target abnormal interface, and then identifies multi-dimensional information such as the text content, interface layout, and icons of the pop-up window.

[0113] For example, the first intelligent agent can perform data preprocessing and cleaning (duplicate removal or anomaly detection) on multi-dimensional information, then perform feature extraction and selection, and finally arrange and combine the data according to the target field format to obtain the target key data. Here, the target field format refers to the standard format used for uniformly organizing and storing the target key data, and the target field format can be set according to the actual application scenario.

[0114] For example, the target key data may include multiple fields such as error type, occurrence time, related process name, error code, and scope of impact. By converting multi-dimensional information into the target field format, subsequent modules can quickly read and process the target key data, thereby improving the overall processing efficiency.

[0115] In this way, the first intelligent agent achieves standardized output of abnormal information through the above operations, improving the readability and usability of the data. The standardized output can better support the subsequent data processing and repair strategy generation.

[0116] In some other embodiments, the electronic device, in performing step S101 above, "in response to monitoring that a target abnormal event occurs during the operation of the electronic device, obtains target key data from the target abnormal event," such as... Figure 3 As shown, the following steps S303 can also be performed:

[0117] Step S303: In response to the detection that an abnormal target interface appears during the operation of the electronic device, the first intelligent agent is activated to call at least one processing model to analyze the interface information in the abnormal target interface, and extract the target key data from the interface information with reference to the content of the first prompt word.

[0118] In the embodiments of this application, if an abnormal interface appears during the operation of an electronic device, a first intelligent agent is activated to call at least one processing model to analyze the interface information in the target abnormal interface. That is, the abnormal information in the Windows GUI interface is analyzed, including the text content of pop-ups, interface layout, icons, and other multi-dimensional information, and the target key information is extracted based on the content of the first prompt word. The processing model can be a large language model or a multimodal model.

[0119] In the embodiments of this application, the first prompt word content is used not only to guide the first intelligent agent to understand the target abnormal interface content, but also to guide the analysis direction of the processing model. For example, the first prompt word content can instruct the processing model to prioritize identifying error codes or focusing on specific types of interface elements, thereby ensuring that the extracted information meets actual needs.

[0120] For example, the first prompt can also be a prompt generated based on key content information from big data. This first prompt can extract target key data from the interface information. Through a preset prompt such as "Please understand the interface information in the pop-up scenario and output the key information of the pop-up," it extracts key problem descriptions from complex interface information, such as error codes and error reason keywords, and passes the extracted target key information to the key information retrieval agent. The key information of the pop-up may include information such as the application that caused the exception, the exception in the file path, the exception path, etc.

[0121] In the embodiments of this application, after the interface information and the content of the first prompt word in the analyzed target abnormal interface are input into the first intelligent agent, the first intelligent agent can output the corresponding target key data.

[0122] In the embodiments of this application, since the malfunction of the electronic device can be reflected in the interface information, the corresponding first intelligent agent is either a problem understanding intelligent agent or an interface information processing intelligent agent (image recognition intelligent agent). The information processing intelligent agent is a dedicated intelligent agent set based on event type, i.e., text type, capable of extracting key problem descriptions from complex interface information, such as error codes and error reason keywords. The problem understanding intelligent agent is one that can output corresponding key information for different inputs. The problem understanding intelligent agent can input text information, error codes, or other multi-dimensional data.

[0123] Thus, by introducing an intelligent processing mechanism that combines interface information with the content of the first prompt word, efficient identification of the target abnormal interface and extraction of key information from the target abnormal interface are achieved.

[0124] In some embodiments, when the electronic device performs the step S102 described above, "generating a target repair plan based on the target retrieval results corresponding to the target key data," as follows: Figure 4 As shown, the following steps S401 and S402 can be performed:

[0125] Step S401: Construct a target retrieval strategy based on the target key data, and process the retrieval data obtained from the target database through the target retrieval strategy into target retrieval results.

[0126] In the embodiments of this application, the electronic device constructs a target retrieval strategy by generating keywords or query conditions for database retrieval based on key information of the user's current problem (such as error codes, error reason keywords, etc.) and key event information recorded by the system (such as game installation time, system update records, hardware configuration, etc.).

[0127] For example, when a user's computer displays an error message indicating that DirectX was not initialized correctly while running a game, the first agent extracts the key information "DirectX initialization failed" from the error message and combines it with information provided by the user, such as the graphics card model, operating system version, and the last system update time, to construct a more precise combination of search keywords, such as "DirectX initialization failed + NVIDIA GTX 1650 + Windows 10 22H2". This method of constructing a targeted search strategy (consistent with the search expression mentioned above) can improve the relevance and accuracy of search results, avoiding the return of a large number of irrelevant forum discussions or outdated target search results.

[0128] In the embodiments of this application, the target retrieval strategy may include semantic retrieval or image retrieval in addition to the keyword retrieval described above. For example, if the obtained target key data exists in the form of an image, retrieval can be performed based on the image, or semantic recognition can be performed on the image, and a retrieval expression can be constructed based on the recognized semantic data.

[0129] For example, for text-based questions, natural language processing models can also be used for semantic retrieval to identify words such as crash, black screen, and app crash in the user's description, and map these words to standard terms to improve retrieval results.

[0130] In the embodiments of this application, the configuration weights of each search keyword in the search query can also be set. For example, in the above example, if the target key data is a system overheating alarm, then the target key data can be divided into structured data (such as error codes, device models) and unstructured data (such as log text or user feedback). Then, the core data includes "high temperature alarm", "electronic equipment", "temperature sensor data", and "heat dissipation system status". Of course, "overheating" can also be expanded to include "temperature threshold exceeded", "heat dissipation failure", etc. Then, the configuration search weights for the core data can be set higher, and the configuration search weights for the expanded keywords can be appropriately lower.

[0131] For example, in the above example, if the target key data includes "blue screen", which can be broken down into "driver conflict", "memory error", etc., then the configuration weight of "blue screen" can be set higher, and the configuration weight of "driver conflict", "memory error", etc. can be lower than the configuration weight of "blue screen".

[0132] In the embodiments of this application, after obtaining the target retrieval strategy, the retrieved data can be retrieved in the target database, which may be a diagnostic database, log database, diagnostic records, technical forum database, etc.

[0133] For example, the target database can also be a database corresponding to the target key data. For instance, when the key information of the problem is a game crash caused by driver incompatibility, the target database may be the official target search results or community forums provided by the manufacturer; while when the key information of the problem is an error pop-up caused by system setting errors, relevant records may be extracted from the local log database.

[0134] In the embodiments of this application, the electronic device processes the obtained search data into target search results. The process of processing search data and generating target search results includes operations such as cleaning invalid data, deduplication, classification, and sorting. For example, from search data retrieved from multiple sources, duplicate content is removed, non-Chinese forum posts are filtered, and target search results are sorted by publication time to prioritize the display of the latest search data. The target search results can be in various forms such as images, text, links, and code snippets.

[0135] In the embodiments of this application, the electronic device can sort multiple retrieved data based on information such as the relevance of the retrieved data to the target retrieval strategy and the update time, and then determine the target retrieval result from them based on the reordering strategy.

[0136] Step S402: Generate a target repair plan based on the target retrieval results and the content of the second prompt words.

[0137] In the embodiments of this application, the target retrieval results obtained by the electronic device are 5 retrieval data. Natural language processing can be used to process and summarize the 5 retrieval data to obtain the repair process. Combined with script generation technology and the limitation Prompt "to solve this problem, a corresponding batch script needs to be generated for me", a corresponding repair batch script is generated for the specific problem, and then the repair batch script is executed to repair the target abnormal event.

[0138] In the embodiments of this application, the target repair solution, in addition to the batch script itself, may also include execution order instructions, dependency check logic, rollback mechanisms, and other content. For example, before executing the repair script, the script automatically detects whether the user's operating system version supports the relevant instructions and provides alternative repair paths. Furthermore, the repair script records the status of each step during execution for subsequent analysis or backtracking of the problem's cause.

[0139] In this way, a closed-loop process is achieved, from identifying key target data of the problem to automatic repair, improving the efficiency and accuracy of automated operation and maintenance.

[0140] In some embodiments, when the electronic device performs the "constructing a target retrieval strategy based on target key information" step S401 above, such as Figure 5As shown, the following steps S501 can also be performed:

[0141] Step S501: Construct a target retrieval strategy based on the target key information and the first operation log data of the electronic device. The first operation log data includes the operation log data of the target component corresponding to the target key information and the operation log data of the electronic device associated with the target component.

[0142] In the embodiments of this application, the electronic device constructs a target retrieval strategy based on key target information and the first operating log data of the electronic device. The first log data may include, but is not limited to, key event information recorded by the system, such as game installation time, system update records, hardware configuration, etc.

[0143] In the embodiments of this application, the operation log data of the target component (e.g., graphics card, sound card, hard disk) corresponding to the target key information includes the operation log data of the component that has abnormal problems, such as the running status, user operation data, system resources called, communication data between other related components, etc.

[0144] In the embodiments of this application, the operation log data of the electronic device associated with the target component may be the device operation log since the component was installed in the electronic device, or it may be system updates, firmware updates, software updates or related component configuration data, etc.

[0145] For example, if the target key information is that the GPU temperature is too high, then in addition to obtaining the GPU's operation log data, the operation log data of components associated with the GPU, such as graphics card drivers, cooling fans, power management modules, etc., can also be obtained. Then, a target retrieval strategy can be constructed based on the obtained operation log data and the target key data.

[0146] In the embodiments of this application, after obtaining the first operation log data, the electronic device can also process the first operation log data to obtain keywords, and then combine them with target key information to construct a target retrieval strategy.

[0147] For example, if the target key information is a game startup error, then in addition to obtaining the game application's runtime log data (e.g., game installation time, software update time, configuration information), it is also possible to obtain the runtime log data of components associated with the game application, such as system configuration, speakers, network connections, etc., and then construct a target retrieval strategy based on the obtained runtime log data and target key data.

[0148] In other embodiments, when the electronic device performs the above-described step S401, "constructing a target retrieval strategy based on target key information," as follows: Figure 5As shown, the following step S502 can also be performed:

[0149] Step S502: Construct a target retrieval strategy based on the target's key information and the second operation log data of the electronic device. The second operation log data includes log data corresponding to the target's abnormal events.

[0150] In the embodiments of this application, the second operation log data focuses on the operation records of the electronic device within a specific time period related to the target abnormal event. The second operation log data typically selects information within a certain time range before and after the abnormality occurs to avoid excessive irrelevant data interfering with the search results. The second operation log data includes, but is not limited to, error codes, abnormality trigger time points, and related process call chains, which helps to quickly locate the cause of the abnormality and improve search efficiency.

[0151] For example, if an application crashes, the electronic device extracts and analyzes log data from a preset time period before and after the crash to quickly identify possible triggering factors. The preset time period can be 3 minutes, 5 minutes, or 1 minute, or it can be set based on the historical time of the program crash. For example, the preset time period can be set based on actual needs and application scenarios, and this application does not limit it in this regard.

[0152] In the embodiments of this application, after the electronic device obtains the second operation log data, it can also perform data processing on the second operation log data to extract key information, and then construct a target retrieval strategy based on the extracted key information and target key information.

[0153] For example, if the key target information is camera call anomaly, then in addition to obtaining the camera's operation log data (e.g., camera firmware update time, program flashing time, configuration information), it is also possible to obtain operation data related to the electronic device, such as other applications calling the camera within a preset time period of camera call anomaly, or the electronic device's system update, and then construct a target retrieval strategy based on the electronic device's system update and camera call anomaly.

[0154] In some other embodiments, when the electronic device performs the above-described step S401, "constructing a target retrieval strategy based on target key information," as follows: Figure 5 As shown, the following steps S503 can also be performed:

[0155] Step S503: Construct a target retrieval strategy based on key target information using a first or second intelligent agent.

[0156] In the embodiments of this application, a first intelligent agent or a second intelligent agent: the second intelligent agent may be independent of the first intelligent agent or may be part of the first intelligent agent. Exemplarily, the relationship between the first intelligent agent and the second intelligent agent can be set based on actual needs and application scenarios, and this application does not limit this.

[0157] In the embodiments of this application, a target retrieval strategy can be constructed by inputting target key information into a first intelligent agent or a second intelligent agent. For example, the first or second intelligent agent can invoke a large language model (artificial intelligence processing model) to directly construct a target retrieval strategy based on the target key information. The large language model is a pre-trained model capable of constructing a target retrieval strategy based on the target key information.

[0158] In the embodiments of this application, steps S501, S502, and S503 include at least one of the following: search keywords, search formula, search method, search weight, and search database. The search method may include keyword search, semantic search, image search, etc.

[0159] For example, different configuration weights can be set for the first running log data and the target key data in step S501. For instance, the configuration weight set for the target key data can be higher than the configuration weight set for the first running log data; conversely, the configuration weight set for the target key data can be lower than the configuration weight set for the first running log data. The configuration weights can be set based on actual needs. Alternatively, different configuration weights can be set for the search terms corresponding to the target key data, or different configuration weights can be set for the search terms corresponding to the first running log data.

[0160] For example, different configuration weights can be set for the second operation log data and the target key data in step S502. For instance, the configuration weight set for the target key data can be higher than the configuration weight set for the second operation log data; conversely, the configuration weight set for the target key data can be lower than the configuration weight set for the second operation log data. The configuration weights can be set based on actual needs. Alternatively, different configuration weights can be set for the search terms corresponding to the target key data, or for the search terms corresponding to the second operation log data.

[0161] In some embodiments, when the electronic device performs the above step S401, "processing the search data retrieved in the target database through the target search strategy into target search results", such as Figure 6 As shown, the following steps S601 can also be performed:

[0162] Step S601: If the retrieval data generated by the first agent or the second agent meets the first condition, perform a retrieval operation in the target database using the target retrieval strategy, and fuse the retrieval results into the target retrieval result.

[0163] In the embodiments of this application, the first condition refers to the situation where the database carried by the first or second intelligent agent is unable to identify the abnormal problem or provide an effective repair solution based on the retrieved data, that is, a new abnormal problem has occurred.

[0164] In the embodiments of this application, the intelligent agent carries a preset database, which stores repair solutions corresponding to different search data. If the search data generated by the first intelligent agent or the second intelligent agent can find the corresponding repair solution in the preset database, then the repair operation is performed directly based on the repair solution. If the intelligent agent cannot identify the abnormal problem or provide a repair solution based on the search data, then the first condition is considered to be met. At this time, it is necessary to use an external network database or expert database for retrieval, and then merge and process the retrieved data into the target retrieval result.

[0165] In the embodiments of this application, the target database can be an external network database, such as a forum or vendor case library; it can also be an expert database. The target database typically contains a large number of solutions and experience sharing provided by real users or experts, which can provide rich reference for problem repair. The first or second intelligent agent will select the most suitable data source according to the problem type to improve the authority and practicality of the search results.

[0166] In the embodiments of this application, the result data obtained by performing a search operation in the target database based on the target retrieval strategy can be deduplicated and fused into the target retrieval result. For example, if the similarity between two result data is greater than 90%, one of the result data can be deleted. Alternatively, the two result data can be fused to obtain a single result data; thus, after fusing multiple result data, the data becomes more concise.

[0167] In some embodiments, when the electronic device performs the above step S401, "processing the search data retrieved in the target database through the target search strategy into target search results", such as Figure 6 As shown, the following step S602 can also be performed:

[0168] Step S602: Use the target retrieval strategy to search multiple databases, and then merge the retrieved results according to the corresponding configured weights to obtain the target retrieval results.

[0169] In the embodiments of this application, the target database includes a forum, a vendor case study database, and an expert database. For example, the vendor case study database is assigned the highest weight, followed by the expert database, and the forum has the lowest weight. The different weights of the multiple databases can be configured based on big data scoring, historical access frequency, credibility, and other users' evaluations.

[0170] In the embodiments of this application, after obtaining the target retrieval strategy, a search is first performed in the vendor case database to obtain the first result data. Then, a search is performed in the expert database to obtain the second result data. After that, a search is performed in the forum to obtain the third result data. Of course, the order of retrieval in each database is not so strict. It can be done simultaneously, sequentially, or in a mixed manner.

[0171] In the embodiments of this application, after obtaining the first result data, the second result data, and the third result data, the obtained data are fused based on the configuration weight of the corresponding retrieval database. For example, the first three result data can be obtained from the first result data, the first two result data can be obtained from the second result data, and the first result data can be obtained from the third result data. Then, the six result data are fused to obtain the target retrieval data.

[0172] In some embodiments, when the electronic device performs the above step S401, "processing the search data retrieved in the target database through the target search strategy into target search results", such as Figure 6 As shown, the following step S603 can also be performed:

[0173] Step S603: Use the target retrieval strategy to search multiple databases, sort the retrieved results data according to the score, and process the results data located in the target sequence into target retrieval results.

[0174] In the embodiments of this application, the electronic device uses a target retrieval strategy to search multiple databases and obtain multiple result data. The obtained multiple result data are sorted according to a rating. The rating can be a user rating, a weighted rating of result data obtained from different databases, or a rating based on information such as historical visits and user feedback.

[0175] In embodiments of this application, after the electronic device sorts the results according to the scores, the result data located in the target sequence is processed into the target retrieval results. For example, the top 5 results are determined as the result data of the target sequence, and then fused together to determine the target retrieval results; or the top 3 results are determined as the result data of the target sequence, and then fused together to determine the target retrieval results. Of course, other numbers of result data from target sequences can also be used. For example, the result data of the target sequence can be set based on actual needs and application scenarios, and this application does not limit this.

[0176] Thus, by combining key target information with multi-dimensional operational log data and introducing the auxiliary analysis of a first or second intelligent agent, a more accurate and personalized target retrieval strategy can be constructed. Furthermore, by setting weights and scores for the database, the highest quality result data is selected from multiple search result data and processed into target retrieval results, which can significantly improve the targeting and effectiveness of the retrieval process. This allows for more efficient acquisition of solutions highly relevant to user problems, reducing invalid searches and the need for manual intervention.

[0177] In some embodiments, when the electronic device performs the above step S103, such as Figure 7 As shown, the following steps S701 and S702 can also be performed:

[0178] Step S701: The second intelligent agent constructs a target retrieval strategy based on the target key information and its corresponding log data, and processes the retrieval data obtained from the target database through the target retrieval strategy into target retrieval results.

[0179] In the embodiments of this application, the relationship between the second intelligent agent and the first intelligent agent has been discussed above and will not be repeated here.

[0180] In the embodiments of this application, if the target key information is a game application startup error, the corresponding log data can be data such as game installation time, system update records, game configuration information, etc. A target retrieval strategy is constructed based on the target key information and its corresponding log data. The constructed target retrieval strategy can be "startup error" AND "game installation time" AND "system update records" AND "game configuration information". Based on this target retrieval strategy, search results are obtained in external databases, expert databases, forums or other databases. Then, the first 3 search results are taken and processed by data fusion to obtain the target retrieval result.

[0181] For example, when a user encounters a problem where the game fails to launch, the second AI will comprehensively analyze the latest update record, the installed game version, and key information from the error message to construct search terms such as Steam game + unable to launch + error code 0x80070005, and select which professional forums or knowledge bases to search first.

[0182] For example, the target database can be a collection of external data sources storing a large amount of information about PC problem-solving experiences, user feedback, and vendor technical documents, including but not limited to technical forums, official support websites, and third-party communities. After preprocessing, the target database is used by the second intelligent agent to quickly match solution paths for relevant problems. For example, common target databases may include Steam community forums, Reddit PC technology sections, Zhihu technology channels, and Microsoft technical support centers.

[0183] For example, after retrieving multiple sets of search data, the data can be structured, deduplicated, scored, and sorted to obtain the target search results.

[0184] For example, the target search results can be diagnostic plans or diagnostic cases formed according to a specific format, or they can include the target component where an abnormal problem has occurred + the abnormal problem + the corresponding solution. For example, a game application + startup failure + restart.

[0185] Step S702: Generate a target repair plan based on the target retrieval results and the content of the second prompt words using a third or second intelligent agent.

[0186] In the embodiments of this application, the electronic device may be equipped with a third intelligent agent or the second intelligent agent may be used to perform the repair solution generation step. Of course, both the second and third intelligent agents are sub-intelligent agents of the first intelligent agent, and both are part of the first intelligent agent. Alternatively, the second and third intelligent agents may be independent of the first intelligent agent; the second intelligent agent may also operate independently of the third intelligent agent, and the third intelligent agent may reuse the functions of the second intelligent agent. That is, in some cases, if the second intelligent agent already possesses sufficiently strong generation capabilities, there is no need to introduce a separate third intelligent agent; instead, the second intelligent agent can directly complete the two steps of retrieval and solution generation. By allowing the second intelligent agent to directly execute the retrieval and solution generation tasks, system complexity can be reduced and response speed improved.

[0187] In the embodiments of this application, the third and second intelligent agents can be intelligent modules responsible for transforming the target retrieval results into specific executable repair solutions, typically possessing strong text understanding and script generation capabilities. The third or second intelligent agent can understand solutions from different sources based on enhanced natural language models (such as GPT, BERT, etc.) and abstract the solutions into a unified instruction set.

[0188] In the embodiments of this application, natural language processing and script generation technologies are used, combined with the limited prompt "to solve this problem, a corresponding batch script needs to be generated for me", to generate a corresponding repair batch script for the specific problem.

[0189] For example, if the retrieved solution suggests updating the graphics card driver, the third-party AI will translate the solution into a specific command-line operation or batch script, or call an automation tool to update the graphics card driver.

[0190] In the embodiments of this application, the second prompt content refers to the guiding statements or templates used in the process of generating the target repair solution, which are used to control the format, style or functional scope of the generated content.

[0191] For example, based on the above search results, please generate a batch script suitable for Windows 10 system to solve the game startup failure problem. This type of second prompt content can be provided by the system to help the agent clarify the task goal, and the system can further ensure that inapplicable or invalid content is avoided.

[0192] In the embodiments of this application, the third intelligent agent can use different second prompt words to guide the generation of a target repair solution based on different problem types. For example, for network connectivity problems, the second prompt word might be to generate a script that detects local network settings and automatically repairs DNS configuration; while for system permission problems, the second prompt word might be to generate a command that modifies the access permissions of a specific folder.

[0193] In this way, intelligent, automated, and personalized solutions can be generated in complex PC problem-solving scenarios. Compared with traditional methods that rely on manual judgment or simple rule bases, the embodiments of this application can more efficiently locate the root cause of the problem and provide a highly adaptable and successful target repair solution, thereby significantly improving user experience and system maintenance efficiency.

[0194] In some embodiments, when the electronic device performs the above step S702 of "generating a target repair plan based on the target retrieval results and the content of the second prompt word", such as Figure 8 As shown, the following steps S801 can also be performed:

[0195] Step S801: Input the target search results and the content of the second prompt word into the second agent or the third agent, so that the second agent or the third agent can generate and process the target search results with reference to the content of the second prompt word to obtain the target repair solution.

[0196] In the embodiments of this application, the relationship between the second intelligent agent and the third intelligent agent has been discussed above and will not be repeated here.

[0197] In the embodiments of this application, the second and third intelligent agents can handle different problem types. For example, the second intelligent agent can focus on generating repair scripts suitable for Windows systems, while the third intelligent agent can generate solutions for specific configuration problems in the game's operating environment. By inputting the target retrieval results and the content of the second prompt words, the second or third intelligent agent will combine a preset Prompt semantic model to perform in-depth analysis of the relevant retrieval information and generate an adapted target repair solution based on the specific situation of the user's computer.

[0198] In the embodiments of this application, the second prompt typically includes limiting conditions. For example, it may request the generation of a batch script to resolve a specific anomaly corresponding to the target search result, or the generation of a target repair solution suitable for the current hardware configuration. The second prompt guides the second or third agent to generate a target repair output that better meets the user's needs. By using the second prompt as guidance, the second or third agent can dynamically adapt to different user devices and problem types, improving the relevance and applicability of the target repair solution.

[0199] In embodiments of this application, the target repair scheme includes at least an instruction set and / or automatic repair code for invoking the target repair tool.

[0200] For example, if the target search result is a solution to the problem of game startup failure, the second suggestion could be: Please generate a batch script suitable for Windows 10 or Windows 11 based on the above search results.

[0201] For example, if the target search result is a blue screen troubleshooting process, the second prompt could be something like: "Based on the above search results, please generate a command set to call the repair tool, then download and install the DirectX repair tool based on the command set, select the 'Detect and Repair' button, restart the computer after the repair is complete, and verify whether the problem is resolved."

[0202] In some embodiments, when the electronic device performs the above step S702 of "generating a target repair plan based on the target retrieval results and the content of the second prompt word", such as Figure 8 As shown, the following steps S802 can also be performed:

[0203] Step S802: Based on the target retrieval results and the content of the second prompt words, determine the repair parameters for the abnormal problems of the target component, and generate a target repair scheme for repairing the abnormal problems of the target component based on the repair parameters.

[0204] In the embodiments of this application, the repair parameters refer to the set of key parameters used to describe how to specifically perform the repair operation when generating the target repair scheme.

[0205] For example, if the problem is caused by a corrupted driver file, the repair parameters might include the driver file's path, version number, update source, and other information. The repair parameters, along with the second prompt ("Please generate automatic repair code to resolve the corrupted driver file"), are extracted and integrated from the target search results. This yields a target repair solution for the abnormal problem of the target component, which is then applied to the corresponding target component using the repair parameters specified in the target repair solution.

[0206] In the embodiments of this application, the instruction set of the target repair tool may be a set of operation instructions for calling system built-in or third-party repair tools, for example, calling the Windows Update API to perform system updates, using batch scripts to perform registry modifications, etc.

[0207] In the embodiments of this application, the automatic repair code is a custom script or program code directly embedded in the target repair solution. For example, it can be used to rewrite configuration files, restore damaged system settings, etc.

[0208] In the embodiments of this application, complex repair processes can also be completed without manual user intervention by combining the instruction set of the target repair tool with automatic repair code.

[0209] In the embodiments of this application, the instruction set and automatic repair code of the target repair tool are suitable for different scenarios. The instruction set of the target repair tool is suitable for calling existing system tools or services, and has high stability and compatibility; the automatic repair code is suitable for handling specific problems, providing greater flexibility and customization capabilities. By combining the instruction set of the target repair tool with the automatic repair code, this approach can cover more types of system problems and improve overall repair efficiency.

[0210] For example, all repairs can be performed using the instruction set of a target repair tool or through automated repair code. For instance, there are instruction sets for target repair tools used to update system configuration variables, repair target files, update target drivers, and repair network connectivity. The instruction set of a target repair tool typically consists of a series of commands that can invoke low-level system interfaces or external tools to quickly locate and resolve problems.

[0211] Thus, by inputting the target search results and second prompt words into a second or third intelligent agent, or by generating repair parameters based on the target search results and second prompt words, a target repair solution can be further generated. This fully utilizes the generation capabilities of the second or third intelligent agent and the accuracy of the repair parameters, thereby generating a target repair solution that is more tailored to the user's device environment and actual problems, significantly improving the success rate and automation of PC problem repair.

[0212] In some embodiments, when the electronic device performs the above step S103, such as Figure 9 As shown, it may also include the following step S901:

[0213] Step S901: The fourth intelligent agent performs a repair operation on the target component of the electronic device based on the target repair scheme.

[0214] In the embodiments of this application, the fourth agent may be a Model Context Protocol (MCP) executing agent, which may be part of the first agent, or part of the second or third agent. The relationship between the first agent, the second agent, the third agent and the fourth agent may be set based on actual needs and application scenarios, and this application does not limit this.

[0215] For example, if the target repair solution is a batch script for a game that fails to launch on a Windows 10 system, the fourth agent can directly run the batch script to repair the game application on the electronic device.

[0216] For example, if the target repair solution is the instruction set that resolves the blue screen, then based on the instruction set, the following instructions are used to repair the blue screen: download and install the DirectX repair tool, select the "Detect and Repair" button, restart the computer after the repair is completed, and verify whether the problem is resolved.

[0217] For example, after obtaining the target repair solution, the fourth intelligent agent can simulate the repair operation for the target component, obtain the corresponding risk assessment, and then provide the risk assessment to the target object operating the electronic device to confirm whether to perform the repair operation. If it is authorized to perform, then the repair operation is performed.

[0218] For example, after generating the target repair plan, if the repair plan requires repairing the network connection, it is also necessary to check whether the user's network configuration allows the operation to repair the network connection, so as to avoid the operation to repair the network connection from failing due to insufficient permissions or configuration conflicts.

[0219] In some embodiments, when the electronic device performs the above step S103, such as Figure 9 As shown, it may also include the following step S902:

[0220] Step S902: Based on the target repair plan, determine the target interactive component used for interaction in the electronic device, and control the target interactive component to perform the operation corresponding to the target repair plan in order to repair the target component.

[0221] In the embodiments of this application, the electronic device can interact with the underlying layer of the user's computer system and automatically perform corresponding repair actions.

[0222] In the embodiments of this application, the target interaction component refers to a hardware or software component in an electronic device used in association with the repair operation. The target interaction component may be the target component of the electronic device itself, or it may be other components associated with the target component of the electronic device, such as a BIOS, an operating system kernel module, or a configuration interface for a specific application. When the fourth intelligent agent performs the repair operation, it first identifies and locates the target interaction component, and then sends control commands to the target interaction component to execute the repair operation.

[0223] For example, if the problem is caused by driver incompatibility, the target interactive component may be the graphics card driver installation interface; if the error occurs during game runtime, it may involve the game client's configuration file.

[0224] In the embodiments of this application, if the target repair solution is to reconfigure the hardware parameters of a certain hardware, then the electronic device can determine which hardware it is based on the target repair solution, and then update the hardware variables by interacting with the BIOS.

[0225] For example, if a fan driver malfunctions, the fan configuration parameters can be updated by interacting with the BIOS. Similarly, if the CPU temperature is too high and the determined solution is to modify the fan speed, the fan speed parameters can also be updated by interacting with the BIOS. This allows for adjustments to be made to the problematic component itself, or to the parameters of components associated with the problematic component.

[0226] In the embodiments of this application, if a software problem occurs, the problem can be solved by changing the application's configuration parameters or the operating parameters of the hardware on which the application depends through the interaction between the intelligent agent and the corresponding application.

[0227] For example, if a game fails to launch due to version incompatibility, and its runtime parameters need to be modified to resolve the issue, a fourth intelligent agent can be used to modify the game application's runtime parameters.

[0228] In some embodiments, when the electronic device performs the above step S103, such as Figure 9 As shown, it may also include the following step S903:

[0229] Step S903: Based on the target repair plan, execute the corresponding target repair tool to repair the target component, or execute the automatic repair code in the target repair plan to repair the target component.

[0230] In the embodiments of this application, the target repair tool refers to a third-party or built-in software tool specifically designed to solve specific types of problems, such as virus scanning tools, driver update tools, system restore tools, etc. When the target repair solution suggests using an external tool for repair, the fourth intelligent agent will automatically call or download the corresponding tool and perform the repair operation according to the instructions of the target repair solution.

[0231] For example, if the target repair solution requires virus scanning and removal to resolve the current abnormal event, and the electronic device does not have a virus scanning tool installed, then you can download and install it first, and then call it to scan and remove the virus; if the target repair solution requires driver repair tool to repair the driver, and the electronic device has a driver repair tool installed, you can directly call it to repair the driver.

[0232] In embodiments of this application, if the target repair solution itself contains executable automatic repair code (such as a batch script, PowerShell script, plugin, etc.), the fourth agent can directly execute the automatic repair code to complete the repair task. For example, a batch script can be executed to reconfigure network settings or reinstall a service.

[0233] In the embodiments of this application, the automatic repair code may be a fourth intelligent agent directly calling the plugin in the file package to perform the repair; or the target repair scheme itself may be an executable automatic repair code that can be used to repair the target components of the electronic device that has a problem.

[0234] For example, firmware reflashing and file recovery can both be accomplished using automated repair code. The application of automated repair code makes the repair process more efficient and automated, thereby reducing reliance on additional external tools.

[0235] Thus, by introducing a fourth intelligent agent to perform repair operations, controlling the target interactive components to perform interactive repairs, and combining the target repair tools or automatic repair code to complete multiple repair paths, efficient, safe, and intelligent PC problem repair can be achieved. This can effectively reduce the time users rely on manual support, thereby improving the overall user experience and system operation and maintenance efficiency.

[0236] In some embodiments, such as Figure 10 As shown, the electronic device can also perform the following step S1001:

[0237] Step S1001: In response to the result of failure to repair the target component, input the result and the target repair plan into the first intelligent agent and re-execute the repair process for the target component.

[0238] In the embodiments of this application, if the repair of the target component fails based on the target repair plan, the failure result and the repair plan will be input into the first intelligent agent, and the repair process of the target component will be re-executed.

[0239] For example, if the application is malfunctioning, the target repair solution is to update the application. However, if the application is still malfunctioning after the update, the electronic device can input the cause of the malfunction and the repair solution, such as updating the application, into the first intelligent agent to re-execute the repair process of the target program.

[0240] For example, if the network function is abnormal, the target repair solution is to replace the router if the router is faulty. However, if the network function is still abnormal after replacing the router, the electronic device can re-enter the cause of the abnormality and the repair solution into the first intelligent agent to carry out the repair process.

[0241] Step S1002: After monitoring the abnormal event of the target, the first prompt message is output, and the first intelligent agent can respond to the first feedback data of the target user to perform the operation of obtaining the target's key data.

[0242] In the embodiments of this application, if an abnormal event is detected, a first prompt message can be output. The first prompt message can be a pop-up notification, or it can be an audio notification, a light notification, or a vibration notification, or a combination of the above-mentioned prompts.

[0243] For example, when the network latency is high, the system can vibrate and output a first prompt message: "The current network latency is high," and provide an interface for selecting whether to perform a repair process. The selection interface can be a clickable button, or clicking and touching any location, etc. In this way, the system can know the first feedback data of the target user after the current target abnormal event. If the first feedback data indicates that the target user agrees to perform the repair operation himself, then the first intelligent agent can respond to the target user's first feedback data and perform the operation of obtaining the target's key data.

[0244] For example, such as Figure 11 As shown, a crash of target application 1 was detected, and the reason obtained was 111, which means that the OLL file may not be registered. Then there is an area 121 that the target user can select, which can be clicked "Close" or "Ignore". If the target user clicks the corresponding error bar 122, it means that he agrees to the self-repair operation.

[0245] Step S1003: After obtaining the target repair solution, output the second prompt information, and the third or fourth intelligent agent can respond to the second feedback data of the target user to perform the repair operation on the target component.

[0246] In embodiments of this application, after obtaining the target repair solution, a second prompt message can be output, for example, such as... Figure 12 As shown, after target application 1 crashes, the identified causes of the anomaly are: GPU memory overload, missing essential game runtime libraries, and outdated game patches. Therefore, the corresponding target repair solutions are: system resource cleanup, component repair / runtime library updates, and game patch updates. For each repair instruction in each target repair solution, a corresponding feedback entry point can be set, such as... Figure 12 As shown in 121, after the user clicks, the second feedback data indicates that the target user agrees to perform the corresponding repair operation on the target component.

[0247] In the embodiments of this application, after obtaining the target repair solution, in addition to outputting the corresponding repair solution, the advantages and disadvantages of the repair solution can also be output. This helps the target user to more comprehensively understand the impact of the repair operation, thereby increasing user engagement.

[0248] An exemplary implementation of anomaly repair is as follows: When an anomaly is detected in a game application on an electronic device, Process Monitor is used to capture the anomaly pop-up and obtain the target key data from it: the game is experiencing significant latency due to an outdated game patch. Then, a target retrieval strategy is constructed by combining key system events, such as the game application's installation / download time, update time, and game version: "game patch + significant latency + game version + update time + installation / download event". This strategy is then used to search the target database, obtaining multiple search results. These multiple search results are then reordered based on weights, ratings, etc., and three search results are selected. These three search results are then merged to obtain the target search result: the game patch needs to be updated. Based on the need to update the game patch and the corresponding prompt, an instruction value for updating the game patch is generated, resulting in a target repair plan. The game application is then repaired based on this target repair plan.

[0249] Another exemplary implementation of anomaly repair is as follows: When an electronic device experiences a blue screen error, acquire key system events within 5 minutes before and after the blue screen, as well as the error codes and content displayed on the blue screen. Based on these key system events, error codes, and error content, construct a target retrieval strategy. Then, based on this strategy, search target databases such as forums, vendor case libraries, and expert databases to obtain multiple search results. Sort these results based on historical access counts, page views, and ratings. Merge the top three results from the sorted database to obtain the target search result: repair bad sectors. Combine the target search result with a command to repair bad sectors to obtain the target repair solution: use the chkdsk / r command to scan and repair bad sectors. Finally, perform a repair operation on the electronic device based on the target repair solution.

[0250] In an embodiment of this application, an exemplary implementation of anomaly repair is as follows: In response to the detection of a target anomaly event, a first intelligent agent of the electronic device obtains key target data from the target anomaly event and provides it to a second intelligent agent of the electronic device; the second intelligent agent provides the target retrieval results obtained based on the key target information to a third intelligent agent of the electronic device; the third intelligent agent generates a target repair plan based on the target retrieval results and provides it to a fourth intelligent agent of the electronic device; the fourth intelligent agent performs a repair operation on the target component of the electronic device based on the target repair plan. Thus, the repair operation for the target anomaly event of the electronic device can be achieved by using multiple intelligent agents. Of course, the first, second, third, and fourth intelligent agents can belong to the same intelligent agent, or the second, third, and fourth intelligent agents can be extensions of the first intelligent agent, or the first, second, third, and fourth intelligent agents can be independent of each other. The relationship between the first, second, third, and fourth intelligent agents can be set based on actual needs and application scenarios, and this application does not limit this.

[0251] For example, in response to an electronic device detecting a functional deficiency in an IoT device (e.g., a camera), the electronic device's first intelligent agent retrieves key target data from the corresponding abnormal data of the IoT device: If the latest camera parameters are missing, this data is given to the second intelligent agent. The second intelligent agent then retrieves the target search results based on the key target information. If a plugin supporting new features is missing, this data is given to the third intelligent agent. The third intelligent agent generates a target repair plan based on the target search results. If a cross-compilation plugin (such as PlatformIO) is called to generate custom firmware and flash it, this plan is given to the fourth intelligent agent. The fourth intelligent agent then performs a repair operation on the target component of the electronic device based on the target repair plan: Device parameters are configured in PlatformIO, firmware is compiled and generated, and then flashed to the camera via USB.

[0252] like Figure 13 As shown, an exemplary implementation of anomaly repair is provided, which is implemented through four agents, including a problem understanding agent 131 (which can correspond to the first agent discussed above), a key information retrieval agent 132 (which can correspond to the second agent discussed above), an enhancement generation agent 133 (which can correspond to the third agent discussed above), and an MCP execution agent 134 (which can correspond to the fourth agent discussed above), including:

[0253] Problem-understanding agent 131 executes step S1301:

[0254] Step S1301, Problem Understanding.

[0255] Here, the GUI Recognize model 1311 identifies and processes pop-ups appearing on the interface. If an anomaly occurs, the problem understanding agent 131 is activated, using a multimodal model to analyze the anomaly information in the Windows GUI interface, including the text content of the pop-up, the interface layout, icons, and other multi-dimensional information. This multi-dimensional information, along with the preset prompt "Please understand the interface most like an anomaly pop-up in this scenario and output the key information of the pop-up," is input into the Large Language Model (LLM) 1312. This LLM extracts key problem descriptions from the complex interface information, such as error codes and error reason keywords, and passes the extracted key information (Key Information Extract) 1313 to the key information retrieval agent 132. Alternatively, the LLM 1312 can also obtain key data from the multi-dimensional information in the CleanData Bug Comments database 1314. This includes GUI Screen Shots Annotation 1315 for training the GUI model 1311 to obtain a trained GUI model 1311; and Entity Extraction 1316 for training the large language model 1312 to obtain a trained large language model 1312.

[0256] Key information retrieval agent 132 executes step S1302:

[0257] Step S1302: Information retrieval.

[0258] Here, after receiving the key information of the problem, the key information retrieval agent 132 combines it with key event information recorded by the system, such as game installation time, system update records, and hardware configuration, to construct search keywords. Through a connection with external networks, it searches on professional technical forums, knowledge bases, and manufacturer official websites (Online Retrieval) 1321 to obtain information related to the problem, such as solutions, user experience sharing, and technical documents. The retrieved content is then reorganized using a reranking strategy 1322 before being transmitted to the enhancement generation agent 133. This content reorganization can be based on the weights of different databases (including databases such as Microsoft 1323, Reddit 1324, and Computer Hope 1325) and the importance of the content.

[0259] Enhanced agent generation 133 execution step S1303:

[0260] Step S1303: Scheme generation.

[0261] Here, the enhanced generative agent 133, based on the received retrieval information (OnlineUpdate Knowledge Base 1331 and a preset knowledge base 1332), utilizes Natural Language Processing and Script Generation (LLM) technology 1333, combined with the limited prompt "to solve this problem, a corresponding batch script needs to be generated for me," to generate a corresponding repair batch script 1334 for the specific problem. The script content covers operation instructions such as system setting adjustments, file repair, and driver updates, ensuring that the script can effectively solve the problem 1335, and then passes the generated script, and / or the generated tool call instructions, to the MCP execution agent 134.

[0262] Step S1304: Execution of the plan.

[0263] Here, after receiving the repair script 1334, the MCP execution agent 134 uses the Large Language Model Virtual Run (LLMVirtual Run) 1341 to interact with the user's computer system underlying layer according to the instructions in the script, automatically executing the corresponding repair actions, and can call the Large Language Model 1345. After execution, the Clarify Result 1346 is fed back to the target user 130. If the target user confirms the execution, then the repair script 1334 is executed. During the execution process, information such as data 1342 (e.g., configuration parameters), tools 1343 (e.g., automatically generated tools or existing tools on electronic devices), and devices 1344 (hardware components on electronic devices) may be called to perform repair operations. During the execution process, the execution status and system feedback are monitored in real time. If the execution is successful, the user is informed that the problem has been resolved; if new problems occur or the execution fails, the new exception information is transmitted to the problem understanding agent again, and the problem repair process is restarted until the problem is resolved.

[0264] This application focuses on the specific scenario of device operation, addressing underlying PC system issues and constructing a complete closed-loop process from problem understanding, information retrieval, repair solution generation to automated execution, demonstrating strong professionalism and scenario adaptability. In terms of technical implementation details, conventional solutions rarely involve multimodal model analysis of the GUI interface or the generation of underlying system operation commands. In this application's embodiments, the problem understanding agent uses a multimodal model to analyze Windows GUI error information, accurately locating problems by integrating multi-dimensional information such as text and images; the enhanced generation agent generates batch scripts for underlying system repair. Furthermore, in terms of information processing, this solution's key information retrieval agent combines key system event information for retrieval, resulting in search results more closely aligned with the user's actual computer situation.

[0265] This application provides an anomaly repair method, comprising: responding to the detection of a target anomaly event during the operation of an electronic device; obtaining target key data from the target anomaly event, wherein the target key data at least characterizes the anomaly problem occurring in the electronic device; generating a target repair scheme based on the target retrieval results corresponding to the target key data, wherein the target repair scheme includes at least one instruction set for repairing the anomaly problem; and performing a repair operation on a target component of the electronic device based on the target repair scheme, wherein the target component is a component associated with the anomaly problem. The anomaly repair method provided by this application not only improves the intelligence level and processing efficiency of anomaly repair but also reduces manual intervention, lowers the consumption of technical support resources, and enhances the user experience.

[0266] This application provides an electronic device, including at least one processor and at least one intelligent agent capable of running on the at least one processor. The at least one intelligent agent can invoke at least one processing model deployed in the electronic device to perform the following operations: in response to monitoring a target abnormal event occurring during the operation of the electronic device, obtaining target key data from the target abnormal event, the target key data being able to at least characterize the abnormal problem occurring in the electronic device; generating a target repair scheme based on the target retrieval results corresponding to the target key data, the target repair scheme including at least one instruction set for repairing the abnormal problem; and performing repair operations on a target component of the electronic device based on the target repair scheme, the target component being a component associated with the abnormal problem.

[0267] In one embodiment of this application, at least one intelligent agent can also invoke at least one processing model to perform the following operations: after monitoring a target abnormal event, locate the target code data of the abnormal problem based on the knowledge graph data associated with the key data in the target abnormal event, and generate automatic repair code for repairing the abnormal problem based on the target code data.

[0268] In the embodiments of this application, knowledge graph data is a semantic network that stores and expresses entities and their relationships in a structured form, which can transform complex data relationships into easily understandable graphical structures.

[0269] In the embodiments of this application, the knowledge graph data contains the relationships between functions, components, and code. By constructing and maintaining the knowledge graph data, the functions that affect the current abnormal event can be located to the components and the corresponding problematic code, and then automatic repair code to fix the abnormal problem can be directly generated based on the problematic code.

[0270] In the embodiments of this application, the use of knowledge graph data enables the search not only to rely on keyword matching for fuzzy search when faced with new problems, but also to find the most relevant problem-solving code through semantic reasoning.

[0271] For example, when a user's computer displays a DirectX initialization failure error pop-up while running a game, the knowledge graph data can be used to find common features (e.g., the corresponding application) between the DirectX initialization failure error and similar previous problems. For example, whether it involves incompatible graphics card driver versions or is related to operating system updates, the specific code location can then be pinpointed based on these common features.

[0272] In practice, there is a close relationship between knowledge graph data and target code data. Knowledge graph data provides contextual information and historical experience, helping the system identify code regions that may be involved in the current anomaly; target code data further refines the specific location of the problem, providing a basis for generating targeted automatic repair code. Combining knowledge graph data with target code data effectively improves the system's diagnostic accuracy and repair performance.

[0273] Thus, by introducing knowledge graph data and combining it with target code data to generate automatic repair code, the accuracy of identifying abnormal issues and the specificity of repair can be improved, and the time spent on invalid attempts and manual troubleshooting can be reduced. This approach enables a more efficient and intelligent PC problem repair process.

[0274] In one embodiment of this application, at least one intelligent agent can also invoke at least one processing model to perform the following operations: upon identifying target input data for a target component of an electronic device, using a target processing model to generate a target code file that matches the user intent represented by the target input data, wherein the target code file can be used to optimize the target code data of the target component to optimize the functional services that the target component can provide.

[0275] In embodiments of this application, if it is detected that a target user inputs data to a target component of an electronic device, a target processing model can be used to generate a target code file that matches the user intent represented by the target input data. The target component may include software components and hardware components. Software components may be applications, drivers, or operating systems; hardware components may be cameras, microphones, displays, heat dissipation modules, communication modules, keyboards and mice, processors, etc. Exemplarily, the target input data may also include user-submitted input regarding software application anomalies, UI adjustments, function optimizations, and personalized adaptations, as well as user-submitted input regarding hardware component anomalies and function optimizations.

[0276] For example, if a target user experiences lag while using game software, and the electronic device may not have detected the abnormal event, the target user can input their needs for the game software, such as adjusting network connection or improving game speed. The electronic device can then use a target processing model to generate a target code file that matches the user's intent represented by the target input data.

[0277] For example, the target processing model can be an AI model deployed on the ontology or a cloud model. The target processing model can be an AI agent model used for diagnosis and repair, a large model used for diagnosis, repair, or code generation, or, of course, the aforementioned intelligent agent. If it is the aforementioned intelligent agent, code generation and repair plan execution can be performed based on the steps described above; the only difference is that one is an actively initiated repair, and the other is a passively initiated repair.

[0278] For example, the firmware of hardware components can also be updated based on the target input data input by the user. If the target user thinks that the image captured by the current camera is blurry, but the camera is a newly replaced camera, the target user can input the requirements for the camera and adjust the imaging parameters of the camera to make the captured image clearer. After receiving the target input data, the electronic device can use the AIAgent model to generate imaging parameter update code based on the requirements. This code can be used to optimize the target imaging data of the camera to optimize the functional services that the camera can provide.

[0279] For example, updating the firmware of hardware components can also involve updating or repairing the firmware code of the t-con or scalar in the display.

[0280] In the embodiments of this application, in the process of generating the target processing model, in addition to considering code templates and modification suggestions, the configuration information of electronic devices and usage environment data can also be considered to optimize code generation. For example, the application code used may be different for different operating systems and environment data, or the configuration information may also be different.

[0281] For example, the threshold for determining overheating of the central processing unit may be different under different temperature environments, and the network speed requirements of game applications may be different for electronic devices with different performance. These can all be used as reference information for the target code.

[0282] In the embodiments of this application, the target code can be an initially generated code file or a code file after code review and unit testing. If it is an initially generated code file, then subsequent processes such as testing can be performed.

[0283] In the embodiments of this application, the target code file can be used to fix abnormal problems or optimization problems existing in the target component, such as the target application. The target code file can be directly used to replace the source code in the application or hardware firmware. For example, if the clarity of the camera is to be adjusted, only the clarity-related parameters in the imaging algorithm, such as contrast and exposure, can be adjusted. Of course, a complete imaging algorithm can also be directly generated and flashed into the camera firmware to update the imaging algorithm.

[0284] In the embodiments of this application, by changing or replacing the code segment of the target application (such as the code in the camera application that optimizes the camera's imaging quality, or the code in the network connection application that switches the signal strength threshold or changes the triggering conditions for switching network connections) or the firmware code of the hardware (including changing the hardware's configuration variables to improve the hardware's functional parameters, such as the CPU power range or the maximum power supply of the battery, etc.), it is possible to solve the abnormal problems or user needs that the user perceives.

[0285] This not only enables the repair of electronic devices upon monitoring for abnormal events, but also allows for optimization of electronic devices based on user needs, thus improving the flexibility of the repair process.

[0286] In one embodiment of this application, generating a target code file that matches the user intent represented by the target input data using a target processing model can involve the following steps: obtaining optimization suggestions generated by the target processing model for the target prompt word data; obtaining a code generation template generated by the target processing model with reference to the target knowledge graph data; processing the optimization suggestions into an initial code file using the code generation template; reviewing and / or testing the initial code file; and outputting the initial code file as the target code file if the initial code file passes the review and / or test.

[0287] like Figure 14 As shown in the embodiments of this application, the AI ​​Coding Assistant 140 inputs the target prompt 141 into the target processing model (LLM) 142. The target processing model 142 analyzes the target prompt 141 data and generates parsed requirements 143 related to the user's intent. These parsed requirements may include optimizations to algorithms, data structures, code, and logic. For example, if the user's intent is to fix an anomaly, the target model can obtain fix suggestions 1417 based on the target prompt, such as modifying the code structure of the code corresponding to the anomaly or changing the algorithm.

[0288] For example, the target prompt can be the target input data entered by the user, or it can be key data extracted based on the target input data entered by the user.

[0289] In embodiments of this application, the target processing model 142 can generate code templates 145 by referencing the target knowledge graph data (MultiKnowledge Base) 144. For example, after receiving a target prompt word processed based on user intent, the target processing model can query relevant entities and relationships in the target knowledge graph based on the target prompt word, and use the queried content to perform reasoning to generate a code template 145 matching the user intent. For example, the code template may include a project structure framework, class / function framework, database operation framework, business logic code snippets, configuration code, etc., and the code template can provide a framework and guidance for subsequent code generation or code optimization.

[0290] In embodiments of this application, the target knowledge graph data may include, for example, syntax rules, best practices, common patterns, etc., and the target code generation template can provide a framework and guidance for subsequent code generation.

[0291] In embodiments of this application, the target processing model 142 can also utilize the code generation template 145 to concretize the optimization suggestions into code form, forming an initial code file (Code) 146, and then optimize it. For example, by analyzing the optimization suggestions, extracting key information, and converting the key information into parameter values ​​in code form, the determined parameter values ​​(the results after analysis) are passed to the code template. Based on the code framework and placeholders in the code generation template, the parameter values ​​are filled into the corresponding positions in the code generation template, generating the initial code file 146.

[0292] In the embodiments of this application, the initial code file (Code) 146 can undergo code review 147. Code review (CR) refers to the process of evaluating code changes, which typically occurs before the code is merged into the main branch. It involves reviewing the code to improve its quality. Through code review, potential problems in the code can be identified and corrected, thereby improving code quality. Code review can be conducted manually and / or using automated tools. For example, team members can discuss and review the initial code file, or automated tools such as static code analysis tools and security scans can be used to review the code.

[0293] For example, the review process may include: submitting code changes to the system after the initial code file 146 is generated; performing automated pre-checks on the code using automated tools; obtaining the check results; assigning the pre-check results to reviewers; having reviewers further check the code; and obtaining code review results. The code review results may be presented in the form of a report, which may include, for example, code style issues, potential errors, security vulnerabilities, etc.

[0294] In the embodiments of this application, unit testing 148 can be performed on the initial code file. Unit testing refers to verification testing of the smallest testable unit (such as a function, method, or class) in the software to ensure that each unit can correctly perform its expected function independently of other units. Unit testing of the initial code file can be performed by writing test cases. For example, consistent test cases are written for each unit, covering normal conditions, boundary conditions, and abnormal conditions, and the test cases are run using a test framework to obtain the test results for that unit.

[0295] In embodiments of this application, the initial code file is further processed based on the results of review and / or testing: if the initial code file passes the review and / or testing, it is output as the target code file. If the initial code file fails the review and / or testing, the code file is regenerated. The code file is regenerated with reference to suggestions in the review and / or testing results, and the newly generated code file is reviewed and / or tested. If the review and / or testing pass, the newly generated code file is output as the target code file. Along with the target code output, a report 149, an optimized code scheme 1410, and suggestions 1411 are also provided.

[0296] For example, such as Figure 14 As shown, the target user 1416 can input requirements 1412, code optimization 1413, problem analysis 1414, etc. into the intent understanding model 1415.

[0297] In this way, the goal-processing model is used to automate code generation, reducing coding time. Knowledge graph data and prompt word data guide the goal-processing model in code generation, effectively improving the flexibility and accuracy of automated code generation. Furthermore, review and evaluation processes are introduced to promptly identify and correct errors in the code, improving its accuracy.

[0298] In one embodiment of this application, generating a target code file that matches the user intent represented by the target input data using a target processing model may further include the following steps: constructing a target knowledge graph based on the target optimization scheme associated with the target prompt word data, wherein the target knowledge graph includes at least the mapping relationship between the user intent and the corresponding knowledge; obtaining target knowledge data from the target knowledge graph based on a target code generation template; and generating the target code file based on the target knowledge data according to at least one of the method call relationship, data dependency relationship, or event sequence relationship in the target code generation template.

[0299] In the embodiments of this application, the electronic device can construct a target knowledge graph based on the target optimization scheme associated with the target prompt word data. Exemplarily, the implementation of constructing the target knowledge graph involves a knowledge graph construction module, a requirement parsing and knowledge mapping module, a code template matching and customization module, and a code verification and optimization module. Specifically, the knowledge graph construction module is responsible for executing the knowledge graph construction steps, including data collection, entity and relation extraction, and knowledge graph storage and management; the requirement parsing and knowledge mapping module implements the requirement parsing and knowledge mapping steps, including a natural language processing engine, requirement parsing tools, and knowledge mapping algorithms; the code template matching and customization module manages the code template library, performs template matching and customization, and interacts with the knowledge graph to obtain relevant knowledge; the code generation and reasoning module executes the code generation and reasoning steps, and includes a built-in code filling algorithm, knowledge reasoning engine, and code integration tools; and the code verification and optimization module verifies and optimizes the generated code, including static code analysis tools, performance optimization algorithms, and user feedback processing mechanisms.

[0300] In the embodiments of this application, after constructing the target knowledge graph, target knowledge data is obtained from the target knowledge graph based on the target code generation template. For example, the target code generation template includes information such as variables of the target optimization scheme related to multi-target prompt word data, or configuration parameters, etc. Target knowledge data is obtained from the target knowledge graph based on the required information in the target code generation template, and then the obtained target knowledge data is used to generate target code files according to the method call relationships, data dependency relationships, or event sequence relationships in the target code generation template.

[0301] For example, if the imaging algorithm needs to be updated, the corresponding adjustment information parameters, such as contrast adjustment parameters, are obtained from the target format map based on the user's intent. Then, the corresponding contrast parameters are replaced with the corresponding parameters in the target code generation module to obtain the target code file. Alternatively, the repair solution can be to first adjust the contrast and then adjust the pixels based on the contrast. In this case of dependency, the target knowledge data will be embedded into the target code generation template according to the dependency in the target code generation template to obtain the target code file.

[0302] In this way, based on the relationships between the target knowledge graph and the relationships in the target code generation template, the corresponding knowledge data is combined with the target code generation template to generate the target code file, and then the repair work is carried out based on the target code file, which improves the efficiency of the repair work.

[0303] This application provides an electronic device that, in response to monitoring a target anomaly event during operation, obtains target key data from the target anomaly event. The target key data at least characterizes the anomaly problem occurring in the electronic device. Based on the target retrieval results corresponding to the target key data, a target repair plan is generated. The target repair plan includes at least one instruction set for repairing the anomaly problem. Based on the target repair plan, a repair operation is performed on a target component of the electronic device, where the target component is associated with the anomaly problem. The electronic device provided by this application not only improves the intelligence and processing efficiency of anomaly repair but also reduces manual intervention, lowers the consumption of technical support resources, and enhances the user experience.

[0304] This application provides a computer-readable storage medium storing one or more computer programs, which can be executed by one or more processors to implement the above-described anomaly repair method. The computer-readable storage medium can be volatile memory, such as random-access memory (RAM); or non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD); or it can be a device including one or any combination of the above-described memories, such as a mobile phone, computer, tablet device, personal digital assistant, etc.

[0305] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0306] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0307] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0308] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0309] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An anomaly repair method, comprising: In response to the detection of a target abnormal event during the operation of an electronic device, target key data is obtained from the target abnormal event, and the target key data can at least characterize the abnormal problem of the electronic device; A target repair plan is generated based on the target retrieval results corresponding to the target key data, and the target repair plan includes at least one instruction set for repairing the abnormal problem; Based on the target repair scheme, a repair operation is performed on a target component of the electronic device, the target component being the component associated with the anomaly.

2. The method according to claim 1, wherein the step of obtaining target key data from the target abnormal event in response to monitoring an abnormal event occurring during the operation of the electronic device includes at least one of the following: Obtain the event type of the target abnormal event, determine the intelligent agent that matches the event type, and use the intelligent agent to extract the target key data from the monitoring data corresponding to the event type; In response to the detection of a target abnormal event during the operation of an electronic device, target log data corresponding to the target abnormal event is obtained, and target key data is obtained from the target log data; In response to a target anomaly warning alert detected during the operation of an electronic device, component operation data of the component associated with the target anomaly warning alert is obtained, and the target key data is obtained from the component operation data.

3. The method according to claim 1, wherein the step of obtaining target key data from the target abnormal event in response to monitoring an abnormal event occurring during the operation of the electronic device includes at least one of the following: In response to the detection of a target abnormal interface appearing during the operation of an electronic device, the image data corresponding to the target abnormal interface and the content of the first prompt word are input to the first intelligent agent, so that the first intelligent agent can extract key information from the target abnormal interface with reference to the content of the first prompt word to obtain the target key data; In response to the detection of a target abnormal interface during the operation of an electronic device, a first intelligent agent is used to identify and analyze the target abnormal interface, and the obtained multi-dimensional information is processed into the target key data according to the target field format; In response to the detection of a target abnormal interface appearing during the operation of an electronic device, a first intelligent agent is activated to call at least one processing model to analyze the interface information in the target abnormal interface, and extract the target key data from the interface information with reference to the content of the first prompt word.

4. The method according to claim 1 or 3, wherein generating the target repair scheme based on the target retrieval results corresponding to the target key data includes: A target retrieval strategy is constructed based on the target key data, and the retrieval data obtained from the target database through the target retrieval strategy is processed into the target retrieval results. The target repair solution is generated based on the target search results and the content of the second prompt words.

5. The method according to claim 4, wherein, A target retrieval strategy is constructed based on the aforementioned key target information, including at least one of the following: The target retrieval strategy is constructed based on the target key information and the first operation log data of the electronic device. The first operation log data includes the operation log data of the target component corresponding to the target key information and the operation log data of the electronic device associated with the target component. The target retrieval strategy is constructed based on the target key information and the second operation log data of the electronic device, wherein the second operation log data includes log data corresponding to the target abnormal events; A target retrieval strategy is constructed based on the target key information using a first or second intelligent agent; The target retrieval strategy includes at least one of the following: retrieval keywords, retrieval formula, retrieval method, retrieval weight, and retrieval database. And / or, The search data obtained from the target database through the target retrieval strategy is processed into the target retrieval results, including at least one of the following: If the retrieval data generated by the first or second intelligent agent meets the first condition, the target retrieval strategy is used to perform a retrieval operation in the target database, and the retrieval result data is fused and processed into the target retrieval result. The target retrieval strategy is used to search multiple databases, and the retrieved results are fused according to the corresponding configured weights to obtain the target retrieval results; The target retrieval strategy is used to search multiple databases, and the retrieved results are sorted according to their scores. The results located in the target sequence are then processed into the target retrieval results.

6. The method according to claim 4, wherein generating a target repair scheme based on the target retrieval results corresponding to the target key data includes: A second intelligent agent is used to construct a target retrieval strategy based on the target key information and its corresponding log data, and the retrieval data obtained from the target database through the target retrieval strategy is processed into the target retrieval result. The target repair scheme is generated using a third agent or the second agent based on the target retrieval results and the content of the second prompt words.

7. The method according to claim 6, wherein, The target repair solution is generated based on the target retrieval results and the content of the second prompt words, including at least one of the following: The target retrieval result and the second prompt word content are input into the second agent or the third agent, so that the second agent or the third agent can generate the target retrieval result by referring to the second prompt word content, and obtain the target repair solution. Based on the target retrieval results and the content of the second prompt word, repair parameters for the abnormality of the target component are determined, and a target repair scheme for repairing the abnormality of the target component is generated based on the repair parameters. The target repair scheme includes at least an instruction set and / or automatic repair code for invoking the target repair tool.

8. The method according to any one of claims 1 to 3 or 5 to 7, wherein performing the repair operation on the target component of the electronic device based on the target repair scheme comprises at least one of the following: A fourth intelligent agent is used to perform repair operations on the target components of the electronic device based on the target repair scheme; Based on the target repair scheme, a target interactive component in the electronic device is identified for interaction, and the target interactive component is controlled to perform an operation corresponding to the target repair scheme in order to repair the target component; Based on the target repair scheme, the corresponding target repair tool is invoked to repair the target component, or the automatic repair code in the target repair scheme is executed to repair the target component.

9. The method of claim 1, further comprising at least one of the following: In response to the result that the repair of the target component has failed, the result and the target repair plan are input to the first intelligent agent, and the repair process for the target component is re-executed; After monitoring the abnormal event of the target, the first prompt message is output, and the first intelligent agent can respond to the first feedback data of the target user to perform the operation of obtaining the key data of the target; After obtaining the target repair solution, a second prompt message is output, and a third or fourth intelligent agent can respond to the second feedback data from the target user to perform the repair operation on the target component.

10. An electronic device, comprising at least one processor and at least one intelligent agent capable of running on said at least one processor, said at least one intelligent agent being able to invoke at least one processing model deployed in the electronic device to perform the following operations: In response to the detection of a target abnormal event during the operation of an electronic device, target key data is obtained from the target abnormal event, and the target key data can at least characterize the abnormal problem of the electronic device; A target repair plan is generated based on the target retrieval results corresponding to the target key data, and the target repair plan includes at least one instruction set for repairing the abnormal problem; Based on the target repair scheme, a repair operation is performed on a target component of the electronic device, the target component being the component associated with the anomaly.

11. The electronic device of claim 10, wherein the at least one intelligent agent is further capable of invoking at least one processing model to perform at least one of the following operations: After detecting a target anomaly, the target code data causing the anomaly is located based on the knowledge graph data associated with the key data in the target anomaly event, and automatic repair code is generated based on the target code data to fix the anomaly; or, When target input data for a target component of an electronic device is identified, a target processing model is used to generate a target code file that matches the user intent represented by the target input data. The target code file can be used to optimize the target code data of the target component to optimize the functional services that the target component can provide.

12. The electronic device according to claim 10, wherein, Generating a target code file that matches the user intent represented by the target input data using a target processing model includes: Obtain optimization suggestions generated by the target processing model based on the target prompt word data; Obtain the code generation template generated by the target processing model based on the target knowledge graph data; The optimization suggestions are processed into an initial code file using the code generation template. The initial code file is reviewed and / or tested, and if the initial code file passes the review and / or test, it is output as the target code file. Or, including: A target knowledge graph is constructed based on the target optimization scheme associated with the target prompt word data. The target knowledge graph includes at least the mapping relationship between the user intent and the corresponding knowledge. Target knowledge data is obtained from the target knowledge graph based on the target code generation template; Based on the target knowledge data, the target code file is generated according to at least one of the method call relationship, data dependency relationship, or event sequence relationship in the target code generation template.

Images