A Computer Parameter Self-Calibration Method and System Based on Real-Time Indicator Feedback

Abstract

This invention relates to the field of artificial intelligence technology, and discloses a method and system for computer parameter self-calibration based on real-time indicator feedback. The method includes: storing performance indicator data in a time-series format to obtain real-time indicator change curves; extracting dynamic features of the real-time indicator change curves within a preset time window, and comparing the similarity of the dynamic features with normal operation features to obtain abnormal fluctuation indicators and abnormal features; traversing the currently effective set of configuration parameters of the computer to obtain a set of associated configuration parameters; generating specific adjustment instructions for the associated configuration parameter set based on the correspondence between the abnormal features and the parameter settings in the associated configuration parameter set; incrementally adjusting the parameter settings based on the specific adjustment instructions to obtain the adjusted settings of the associated configuration parameter set; and recording the abnormal features, parameter settings, and adjusted settings as a successful calibration case. This invention can improve the efficiency of computer parameter self-calibration.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a computer parameter self-calibration method and system based on real-time indicator feedback. Background Technology

[0002] In the existing technology system for computer parameter calibration, the data collection dimensions for computer performance indicators are relatively singular, and non-serialized storage methods are mostly used. This makes it impossible to form continuous real-time indicator change curves, and it is difficult to accurately capture the dynamic change patterns of multi-dimensional events such as CPU scheduling and memory access. As a result, the perception of the computer's operating status is lagging and one-sided, and it is impossible to identify abnormal fluctuations in the operation process in a timely manner. At the same time, the existing technology lacks quantitative analysis methods for the dynamic characteristics of performance indicators. Judging abnormal operation based solely on simple threshold comparisons is difficult to distinguish between normal fluctuations and abnormal fluctuations, which can easily lead to misjudgments or omissions, and bring incorrect guidance to subsequent parameter calibration.

[0003] Current computer parameter calibration methods mostly involve manual adjustment or static adjustment based on fixed rules. After identifying operational anomalies, they cannot quickly and accurately match configuration parameters related to abnormal fluctuations, often resulting in excessively large parameter adjustment ranges or insufficient correlation. Furthermore, the adjustment process lacks real-time feedback monitoring of abnormal indicators, and is mostly a one-time batch adjustment, unable to perform incremental or iterative parameter optimization based on indicator changes, resulting in low calibration efficiency and accuracy. In addition, the relevant data after calibration is not systematically linked, recorded, and indexed for storage, making it impossible to form reusable calibration cases. Each time a similar anomaly is encountered, parameter analysis and adjustment must be repeated, causing a large waste of resources and preventing the effective accumulation and reuse of parameter calibration experience. Therefore, how to improve the efficiency of computer parameter calibration has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a computer parameter self-calibration method and system based on real-time index feedback to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a computer parameter self-calibration method based on real-time index feedback, comprising:

[0006] S1. The timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events are collected into the computer's performance index data, and the performance index data is stored in a time-series format to obtain the real-time index change curve of the computer.

[0007] S2. Within a preset time window, extract the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve to obtain the dynamic characteristics of the real-time indicator change curve, and compare the similarity of the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer.

[0008] S3. Using the abnormal fluctuation index as an index, traverse the currently effective set of configuration parameters of the computer to obtain the associated set of configuration parameters of the computer;

[0009] S4. Based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, determine the parameter adjustment direction and adjustment range of the abnormal feature, and generate a specific adjustment instruction for the associated configuration parameter set;

[0010] S5. Based on the specific adjustment instruction, the parameter setting value is incrementally adjusted, and the feedback change of the abnormal fluctuation index is monitored in real time to obtain the adjusted setting value of the associated configuration parameter set.

[0011] S6. The abnormal feature, the parameter setting value, and the adjusted setting value are associated and recorded as a successful calibration case of the computer.

[0012] In a preferred embodiment, the step of aggregating the timestamps and attributes of the computer's CPU scheduling, memory access, disk I / O, and network data transmission and reception events into the computer's performance index data, and storing the performance index data in a time-series format to obtain the computer's real-time index change curve, includes:

[0013] Raw event records are captured from the central processing unit scheduler, memory management unit, block device layer and network protocol stack in the computer. The raw event records include a high-precision timestamp of the time when the event occurred and an event type identifier.

[0014] Based on the event type identifier, extract the context switching count of the CPU scheduling event, the page fault type of the memory access event, the request size and priority of the disk input / output event, and the message length and transmission protocol type of the network data transmission and reception event to obtain the key attributes of the computer.

[0015] According to the high-precision timestamp, the key attributes are filled into the computer's circular buffer, and an interrupt is triggered when the circular buffer overflows to notify the computer's user-mode program to read the data.

[0016] Remove incomplete records from the circular buffer and correct any misalignments in the records within the circular buffer to obtain the computer's performance metrics data.

[0017] Based on the event type identifier, the performance index data is mapped to a two-dimensional coordinate system with time as the horizontal axis and index value as the vertical axis, and adjacent index data points are connected by straight line segments to obtain the real-time index change curve of the computer.

[0018] In a preferred embodiment, the step of extracting the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve within a preset time window to obtain the dynamic characteristics of the real-time indicator change curve, and comparing the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer, including:

[0019] Using a preset time window as a sliding window, the real-time indicator change curve is divided into curve segments, and the curve segments are discretized and sampled to obtain the first sequence of the computer.

[0020] The first sequence is analyzed for trend to obtain the trend direction identifier of the curve segment;

[0021] Historical indicator segments corresponding to the preset time window are extracted from the computer's operation log, and the historical indicator segments are discretized and sampled to obtain the computer's second sequence;

[0022] Calculate the dynamic characteristic deviation between the first sequence and the second sequence to obtain the abnormal fluctuation index of the computer;

[0023] The abnormal fluctuation index is compared and analyzed with the fluctuation tolerance threshold of the computer to obtain the abnormal segments of the computer, and the abnormal segments and the trend direction identifiers corresponding to the abnormal segments are recorded as the abnormal features of the computer.

[0024] In a preferred embodiment, the formula for calculating the abnormal fluctuation index is:

[0025] ;

[0026] in, This indicates the abnormal fluctuation index. This represents the total number of discrete sampling points within the preset time window. Indicates the first sequence of the first sequence The time series value of each sampling point Indicates the second sequence of the first... The time series value of each sampling point Indicates the second sequence of the first... Historical standard deviation of the sampling points This represents a pre-defined non-zero minimum constant. Indicates the first sequence of the first sequence The instantaneous rate of change of each sampling point This represents the historical average rate of change of the second sequence.

[0027] In a preferred embodiment, the step of using the abnormal fluctuation index as an index to traverse the currently effective set of configuration parameters of the computer to obtain the associated configuration parameter set of the computer includes:

[0028] The abnormal fluctuation index is decomposed to obtain the abnormal feature components of the computer and the abnormal intensity corresponding to the abnormal feature components.

[0029] Obtain the currently effective set of configuration parameters for the computer, wherein the set of configuration parameters is pre-configured with a set of parameter feature tags;

[0030] For the configuration parameter set, the parameter feature label set is matched with the abnormal feature components, and the abnormal intensity is weighted and fused to obtain the initial relevance of the configuration parameter set;

[0031] The initial relevance is passed to the upper-level and lower-level configuration parameters of the configuration parameter set to obtain the extended relevance of the configuration parameter set;

[0032] Based on the extended relevance, the configuration parameter set is sorted and filtered to obtain the initial associated configuration parameter set of the computer;

[0033] Redundancy is removed from the initial set of associated configuration parameters to obtain the set of associated configuration parameters for the computer.

[0034] In a preferred embodiment, the step of removing redundancy from the initial associated configuration parameter set to obtain the associated configuration parameter set of the computer includes:

[0035] Obtain the parameter type identifier and parameter scope of the configuration parameters in the initial associated configuration parameter set;

[0036] Based on the parameter type identifier, cluster analysis is performed on the associated configuration parameter set to obtain the parameter group of the computer;

[0037] Configuration parameters with overlapping scopes in the parameter group are marked as redundant parameter pairs for the computer;

[0038] The computer's associated configuration parameter set is obtained by removing the configuration parameters with low contribution from the redundant parameter pairs in the initial associated configuration parameter set.

[0039] In a preferred embodiment, determining the parameter adjustment direction and adjustment range of the abnormal feature based on the correspondence between the abnormal feature and the parameter setting values ​​in the associated configuration parameter set, and generating a specific adjustment instruction for the associated configuration parameter set, includes:

[0040] Extract the anomaly type identifier and anomaly degree of the aforementioned anomaly features;

[0041] The associated configuration parameter set is structured and parsed to obtain the parameter type identifier, current setting value, adjustable range, and adjustment step size of the associated configuration parameter set;

[0042] A correlation analysis is performed on the anomaly type identifier and the parameter type identifier to obtain the matching relationship between the anomaly type identifier and the parameter type identifier, and the adjustment direction of the associated configuration parameter set is determined based on the matching relationship;

[0043] The degree of abnormality is mapped to a level to obtain the adjustment range level of the associated configuration parameter set;

[0044] The adjustment range of the associated configuration parameter set is determined based on the adjustment range level and the adjustment step size;

[0045] The adjustable range and the adjustment magnitude are combined and encoded to obtain the specific adjustment instructions for the associated configuration parameter set.

[0046] In a preferred embodiment, the step of incrementally adjusting the parameter setting value based on the specific adjustment instruction and monitoring the feedback changes of the abnormal fluctuation index in real time to obtain the adjusted setting value of the associated configuration parameter set includes:

[0047] Based on the specific adjustment instruction, obtain the parameter identifier, adjustment direction identifier, and adjustment range identifier of the associated configuration parameter set;

[0048] Based on the dependencies between configuration parameters in the associated configuration parameter set, determine the sequential adjustment order of the associated configuration parameter set;

[0049] The current parameter to be adjusted in the associated configuration parameter set is selected according to the serial adjustment order, and the basic adjustment step size of the computer is determined according to the degree of deviation between the current abnormal characteristic value of the computer and the normal operation characteristic of the computer.

[0050] An incremental adjustment instruction containing the parameter identifier and the basic step size is sent to the computer to perform a single modification and record the intermediate setting value after modification.

[0051] The average rate of change of the abnormal fluctuation index is continuously collected, and the average rate of change is compared with the trend of change of the abnormal features to obtain the abnormal change comparison result of the computer.

[0052] Based on the comparison results of the abnormal changes, the intermediate setting value is iteratively optimized to obtain the adjusted setting value of the associated configuration parameter set.

[0053] In a preferred embodiment, the step of associating and recording the abnormal feature, the parameter setting value, and the adjusted setting value as a successful calibration case for the computer includes:

[0054] The abnormal features, the parameter settings, and the adjusted settings are integrated into the computer's original case data.

[0055] The original case data is structured and encoded to obtain the standard case record format of the computer;

[0056] The standard case record format is compared item by item with the computer's historical successful cases to obtain similar cases of the computer. The confidence of the similar cases is increased and the timestamps of the similar cases are updated to obtain the case record of the computer.

[0057] Based on the abnormal indicator types and adjusted parameter types in the case records, multi-level index entries are constructed, and the storage location pointers of the case records are respectively attached to the corresponding nodes of the abnormal indicator type index table and the parameter type index table to obtain the successful calibration cases of the computer.

[0058] To address the above problems, the present invention also provides a computer parameter self-calibration system based on real-time index feedback, the system comprising:

[0059] The data aggregation module is used to aggregate the timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events into the computer's performance index data, and to store the performance index data in a time-series format to obtain the computer's real-time index change curve.

[0060] The anomaly comparison module is used to extract the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve within a preset time window, obtain the dynamic characteristics of the real-time indicator change curve, and compare the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer.

[0061] The configuration parameter association module is used to traverse the currently effective configuration parameter set of the computer using the abnormal fluctuation index as an index, and obtain the associated configuration parameter set of the computer.

[0062] The adjustment instruction generation module is used to determine the parameter adjustment direction and adjustment range of the abnormal feature based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, and generate a specific adjustment instruction for the associated configuration parameter set.

[0063] The parameter adjustment module is used to incrementally adjust the parameter setting value based on the specific adjustment instruction, and monitor the feedback changes of the abnormal fluctuation index in real time to obtain the adjusted setting value of the associated configuration parameter set.

[0064] The case storage module is used to associate and record the abnormal features, the parameter settings, and the adjusted settings as successful calibration cases of the computer.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] 1. This invention achieves real-time, multi-dimensional acquisition and serialized analysis of computer performance indicators, accurately extracts dynamic features of indicator curves, and locates abnormal fluctuation indicators and features. Then, it quickly matches and associates configuration parameter sets through anomaly indicator indexes, and makes parameter adjustments more targeted after redundancy removal. At the same time, it clarifies the adjustment direction and magnitude based on the correspondence between abnormal features and parameter settings, and completes parameter optimization by combining incremental adjustment and real-time feedback monitoring. This significantly improves the accuracy and efficiency of computer parameter self-calibration, making parameter adjustments fit the abnormal problems of actual computer operation, and realizing the scientific and intelligent self-calibration process.

[0067] 2. This invention records successful calibration cases by associating the abnormal features of each calibration with the original parameter settings and the adjusted settings. It also forms a reusable calibration case library through structured coding, confidence updates of similar cases, and multi-level indexing. This provides a direct reference for subsequent parameter self-calibration of similar anomalies, reduces repetitive feature analysis and parameter matching steps, and continuously improves the response speed and processing efficiency of subsequent computer self-calibration. At the same time, it enables the self-calibration system to have self-learning and self-optimization capabilities, continuously adapting to changes in computer operating status and ensuring continuous improvement in self-calibration effectiveness. Attached Figure Description

[0068] Figure 1 This is a flowchart illustrating a computer parameter self-calibration method based on real-time index feedback provided in an embodiment of the present invention.

[0069] Figure 2 A functional block diagram of a computer parameter self-calibration system based on real-time index feedback provided in an embodiment of the present invention;

[0070] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0071] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0072] This application provides a computer parameter self-calibration method based on real-time indicator feedback. The execution subject of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the computer parameter self-calibration method based on real-time indicator feedback can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0073] Reference Figure 1 The diagram shown is a flowchart illustrating a computer parameter self-calibration method based on real-time indicator feedback according to an embodiment of the present invention. In this embodiment, the computer parameter self-calibration method based on real-time indicator feedback includes:

[0074] S1. The timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events are collected into the computer's performance index data, and the performance index data is stored in a time-series format to obtain the real-time index change curve of the computer.

[0075] In this embodiment of the invention, the step of aggregating the timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events into the computer's performance index data, and storing the performance index data in a time-series format to obtain the computer's real-time index change curve, includes:

[0076] Raw event records are captured from the central processing unit scheduler, memory management unit, block device layer and network protocol stack in the computer. The raw event records include a high-precision timestamp of the time when the event occurred and an event type identifier.

[0077] Based on the event type identifier, extract the context switching count of the CPU scheduling event, the page fault type of the memory access event, the request size and priority of the disk input / output event, and the message length and transmission protocol type of the network data transmission and reception event to obtain the key attributes of the computer.

[0078] According to the high-precision timestamp, the key attributes are filled into the computer's circular buffer, and an interrupt is triggered when the circular buffer overflows to notify the computer's user-mode program to read the data.

[0079] Remove incomplete records from the circular buffer and correct any misalignments in the records within the circular buffer to obtain the computer's performance metrics data.

[0080] Based on the event type identifier, the performance index data is mapped to a two-dimensional coordinate system with time as the horizontal axis and index value as the vertical axis, and adjacent index data points are connected by straight line segments to obtain the real-time index change curve of the computer.

[0081] The system captures various runtime events in real time from the computer's central processing unit scheduler, memory management unit, block device layer, and network protocol stack to form raw event records. While capturing each event, it simultaneously records a high-precision timestamp of the event occurrence and marks each event with a unique event type identifier, ensuring that each raw event record contains both a high-precision timestamp and an event type identifier.

[0082] Based on the event type identifiers carried in the original event logs, different types of computer operation events are accurately distinguished. For CPU scheduling events, the number of context switches is directly extracted; for memory access events, the page fault type is directly extracted; for disk I / O events, the request size and priority are directly extracted; and for network data transmission and reception events, the message length and transmission protocol type are directly extracted. The information extracted from these different types of events is integrated and summarized to form the key attributes of the computer.

[0083] Based on the time sequence of the high-precision timestamps, the extracted key computer attributes are sequentially written into the computer's circular buffer to complete the filling operation. The circular buffer receives key attribute data according to the first-in-first-out storage rule. When the storage capacity of the circular buffer reaches the preset upper limit and overflows, a hardware interrupt signal is immediately triggered on the computer. This interrupt signal is directly transmitted to the computer's user-mode program, thereby informing the user-mode program to perform a read operation on the key attribute data in the circular buffer.

[0084] The integrity of all key attribute records stored in the circular buffer is checked line by line. Incomplete records that do not fully contain the corresponding event attribute information are identified and deleted directly from the circular buffer. At the same time, the record order misalignment caused by data writing and reading operations in the circular buffer is checked. The misaligned records are reordered and reorganized according to the original time order of the high-precision timestamp. After the incomplete record removal and record misalignment correction are completed, the remaining complete and ordered key attribute records in the circular buffer are integrated to obtain the computer's performance index data.

[0085] The computer's performance metrics data are classified according to event type identifiers. The classified performance metrics data are then mapped one by one to a two-dimensional coordinate system with time as the horizontal axis and the metric value as the vertical axis. The high-precision timestamp in the performance metrics data determines its corresponding position on the horizontal axis of the two-dimensional coordinate system, and the specific metric value in the performance metrics data determines its corresponding position on the vertical axis of the two-dimensional coordinate system. After finding the coordinate point corresponding to each performance metrics data in the two-dimensional coordinate system, adjacent coordinate points are connected sequentially using straight line segments to finally form the real-time metric change curve of the computer.

[0086] The beneficial effects include directly capturing raw event records from the core operating modules of the computer, ensuring the originality and accuracy of performance indicator data; accurately extracting key attributes through event type identifiers and achieving ordered storage and timely interruptive data reading based on a circular buffer, effectively ensuring the efficiency and continuity of data storage; removing incomplete data and correcting misalignments in the circular buffer, further improving the integrity and orderliness of performance indicator data; and finally, completing data mapping based on event type identifiers and connecting adjacent data points to form real-time indicator change curves, which can intuitively and accurately present the real-time operating status of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception, providing accurate, continuous, and complete real-time indicator data support for subsequent computer parameter self-calibration.

[0087] S2. Within a preset time window, extract the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve to obtain the dynamic characteristics of the real-time indicator change curve, and compare the similarity of the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer.

[0088] In this embodiment of the invention, the step of extracting the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve within a preset time window to obtain the dynamic characteristics of the real-time indicator change curve, and comparing the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer, includes:

[0089] Using a preset time window as a sliding window, the real-time indicator change curve is divided into curve segments, and the curve segments are discretized and sampled to obtain the first sequence of the computer.

[0090] The first sequence is analyzed for trend to obtain the trend direction identifier of the curve segment;

[0091] Historical indicator segments corresponding to the preset time window are extracted from the computer's operation log, and the historical indicator segments are discretized and sampled to obtain the computer's second sequence;

[0092] Calculate the dynamic characteristic deviation between the first sequence and the second sequence to obtain the abnormal fluctuation index of the computer;

[0093] The abnormal fluctuation index is compared and analyzed with the fluctuation tolerance threshold of the computer to obtain the abnormal segments of the computer, and the abnormal segments and the trend direction identifiers corresponding to the abnormal segments are recorded as the abnormal features of the computer.

[0094] The formula for calculating the abnormal fluctuation index is as follows:

[0095] ;

[0096] in, This indicates the abnormal fluctuation index. This represents the total number of discrete sampling points within the preset time window. Indicates the first sequence of the first sequence The time series value of each sampling point Indicates the second sequence of the first... The time series value of each sampling point Indicates the second sequence of the first... Historical standard deviation of the sampling points This represents a pre-defined non-zero minimum constant. Indicates the first sequence of the first sequence The instantaneous rate of change of each sampling point This represents the historical average rate of change of the second sequence.

[0097] Using a preset time window of fixed duration as a sliding window, the sliding window is slid sequentially along the horizontal axis of the real-time indicator change curve to cut the entire real-time indicator change curve into multiple curve segments of equal duration. For each curve segment, the corresponding indicator value is selected at a fixed time interval, and the selected indicator values ​​are arranged in chronological order to obtain the computer's first sequence.

[0098] A continuous numerical change analysis is performed on the indicator values ​​arranged chronologically in the first sequence. Based on the numerical change states of the indicator values, such as continuous rise, continuous fall, or stable change, a corresponding trend direction indicator is matched for each first sequence. This trend direction indicator directly corresponds to the overall numerical change trend of the curve segment, thus completing the trend judgment of the first sequence and obtaining the trend direction indicator of the curve segment.

[0099] Historical performance index data with the same duration and time dimension as the preset time window are retrieved from the computer's operation log to form historical index segments. According to the same time interval for discretizing the curve segments, the corresponding historical index values ​​are selected for the historical index segments. The selected historical index values ​​are arranged in chronological order to obtain the computer's second sequence.

[0100] The index values ​​corresponding to the same time points in the first and second sequences are compared one by one. The numerical deviation of each time point is calculated and the deviation of all time points is integrated. The dynamic characteristic deviation quantity that reflects the overall difference between the two sequences is obtained by combining the differences in the rate of change and fluctuation amplitude of the first and second sequences. This dynamic characteristic deviation quantity is the abnormal fluctuation index of the computer.

[0101] The total number of discrete sampling points within the preset time window comes from the discretization sampling operation performed after dividing the real-time index change curve into curve segments using the preset time window as a sliding window. The first sequence contains the... The time series value of the sampling point comes from the discretized sampling result of the curve segment after dividing the real-time index change curve. In the second sequence, the first... The time series value of the sampling point comes from the discretized sampling results of historical indicator segments corresponding to a preset time window extracted from the computer operation log. The second sequence... The historical fluctuation standard deviation of the first sampling point is a value obtained by calculating the fluctuation degree of the historical index data of the corresponding sampling point in the second sequence. The preset non-zero minimum constant is a fixed value set in advance to avoid the case of zero denominator during the calculation process. The instantaneous rate of change of each sampling point is a value obtained by calculating the rate of change of the time series value of the corresponding sampling point in the first sequence, and the historical average rate of change of the second sequence is a value obtained by calculating the arithmetic mean of the historical rates of change of all sampling points in the second sequence.

[0102] By performing corresponding calculations on each sampling point within a preset time window, summing the results, and then taking the arithmetic mean, the deviation between the real-time collected indicator sequence and the historical normal indicator sequence can be quantitatively represented. At the same time, by combining the deviation between the instantaneous rate of change of the real-time indicator sequence and the average rate of change of the historical indicator sequence, the basic deviation result is corrected. This ensures that the final quantitative result can reflect both the difference in numerical value and the difference in rate of change between the real-time and historical indicators, thereby accurately obtaining a quantitative result that reflects the degree to which the computer's operating state deviates from the normal operating state.

[0103] The greater the difference between the time series value of the real-time indicator sequence and the time series value of the historical indicator sequence, the larger the calculated value will be, and the greater the final quantification result will be. The smaller the historical fluctuation standard deviation of the corresponding sampling point in the second sequence, the larger the calculated value will be when the numerical difference is fixed, and the greater the final quantification result will be. The greater the difference between the instantaneous rate of change of the sampling point in the first sequence and the historical average rate of change of the second sequence, the greater the correction of the basic deviation result will be, and the greater the final quantification result will be. When the calculation results of all sampling points within the preset time window show an overall increasing trend, the final quantification result will continue to rise. When the calculation results of all sampling points within the preset time window show an overall decreasing trend, the final quantification result will continue to fall. When the real-time indicator sequence and the historical indicator sequence are completely consistent in value and rate of change, the final quantification result is zero.

[0104] The abnormal fluctuation index of the computer is directly compared with the pre-set computer fluctuation tolerance threshold. If the abnormal fluctuation index exceeds the fluctuation tolerance threshold, the curve segment corresponding to the abnormal fluctuation index is marked as an abnormal segment of the computer. The marked abnormal segment is associated with the trend direction identifier corresponding to the abnormal segment and the two are recorded as a whole to obtain the abnormal characteristics of the computer.

[0105] The beneficial effects are as follows: by using a preset time window for sliding segmentation and equally spaced discretized sampling, the standardization of real-time indicator change curves is achieved, ensuring that the comparison basis of the first and second sequences is consistent. The trend direction indicators obtained through trend analysis can accurately reflect the change direction of curve segments. Combined with the second sequence extracted from the operation log, it provides an effective reference for normal operation characteristics for anomaly judgment. The abnormal fluctuation indicators obtained by calculating the dynamic feature deviation can quantitatively reflect the difference between real-time indicators and historical normal indicators. Then, by comparing with the fluctuation tolerance threshold, abnormal segments are identified and associated with trend direction indicators to form abnormal features. This can accurately locate the abnormal position and abnormal change trend in computer operation, providing a clear abnormal orientation basis for subsequent parameter calibration.

[0106] S3. Using the abnormal fluctuation index as an index, traverse the currently effective set of configuration parameters of the computer to obtain the associated set of configuration parameters of the computer;

[0107] In this embodiment of the invention, the step of using the abnormal fluctuation index as an index to traverse the currently effective set of configuration parameters of the computer to obtain the associated configuration parameter set of the computer includes:

[0108] The abnormal fluctuation index is decomposed to obtain the abnormal feature components of the computer and the abnormal intensity corresponding to the abnormal feature components.

[0109] Obtain the currently effective set of configuration parameters for the computer, wherein the set of configuration parameters is pre-configured with a set of parameter feature tags;

[0110] For the configuration parameter set, the parameter feature label set is matched with the abnormal feature components, and the abnormal intensity is weighted and fused to obtain the initial relevance of the configuration parameter set;

[0111] The initial relevance is passed to the upper-level and lower-level configuration parameters of the configuration parameter set to obtain the extended relevance of the configuration parameter set;

[0112] Based on the extended relevance, the configuration parameter set is sorted and filtered to obtain the initial associated configuration parameter set of the computer;

[0113] Redundancy is removed from the initial set of associated configuration parameters to obtain the set of associated configuration parameters for the computer.

[0114] The step of removing redundancy from the initial associated configuration parameter set to obtain the associated configuration parameter set of the computer includes:

[0115] Obtain the parameter type identifier and parameter scope of the configuration parameters in the initial associated configuration parameter set;

[0116] Based on the parameter type identifier, cluster analysis is performed on the associated configuration parameter set to obtain the parameter group of the computer;

[0117] Configuration parameters with overlapping scopes in the parameter group are marked as redundant parameter pairs for the computer;

[0118] The computer's associated configuration parameter set is obtained by removing the configuration parameters with low contribution from the redundant parameter pairs in the initial associated configuration parameter set.

[0119] The abnormal fluctuation index is decomposed into abnormal feature components corresponding to different operating dimensions of the computer. At the same time, the abnormal intensity of each abnormal feature component is quantified. The magnitude of the abnormal intensity directly reflects the degree of influence of the abnormal feature component on the overall abnormal fluctuation index.

[0120] All currently active configuration parameters are retrieved from the computer's system configuration file and runtime parameter storage area and integrated to form the computer's currently active configuration parameter set. Each configuration parameter in this set is pre-bound with a parameter feature tag set that can characterize its functional attributes and scope of action. The parameter feature tag set corresponds one-to-one with the configuration parameter and there are no omissions.

[0121] Iterate through each configuration parameter in the configuration parameter set, match its bound parameter feature tag set with the abnormal feature components one by one, filter out the configuration parameters that match the abnormal feature components, and for the successfully matched configuration parameters, calculate the corresponding abnormal intensity according to the preset fusion rules. The calculation result is the initial relevance of the configuration parameter in the configuration parameter set.

[0122] Based on the initial relevance, trace upwards to the parent configuration parameter to which the configuration parameter belongs and pass the initial relevance proportionally to the parent configuration parameter. Then associate downwards with the lower configuration parameters controlled by the configuration parameter and pass the initial relevance proportionally to the lower configuration parameters. Integrate the initial relevance of the parameter itself, the relevance passed to the parent configuration parameter, and the relevance passed to the lower configuration parameter to obtain the extended relevance of the configuration parameter set.

[0123] Sort all configuration parameters in the configuration parameter set according to the numerical order of extended relevance, set a relevance filtering threshold, remove configuration parameters with extended relevance below the threshold, retain configuration parameters with extended relevance higher than or equal to the threshold, and integrate the retained configuration parameters to form the initial associated configuration parameter set of the computer.

[0124] Extract the parameter type identifier and parameter scope of each configuration parameter from the initial set of associated configuration parameters. The parameter type identifier is used to distinguish the function type of the configuration parameter, and the parameter scope clearly indicates the computer operating module and data range to which the configuration parameter applies.

[0125] The configuration parameters in the initial associated configuration parameter set are classified and aggregated according to the parameter type identifier. Configuration parameters with the same type identifier are grouped together to form multiple parameter groups based on parameter type. The configuration parameters in each parameter group have the same parameter type identifier.

[0126] Check the scope of each configuration parameter in each parameter group one by one, compare the scope range of different configuration parameters, and mark two or more configuration parameters with overlapping scopes as redundant parameter pairs of the computer. During the marking process, record the scope overlap range of the redundant parameter pairs completely.

[0127] For each pair of redundant parameters, the contribution is evaluated. The contribution evaluation is determined based on the weight and priority of the configuration parameters on the abnormal fluctuation index. The configuration parameters with smaller contribution values ​​are removed from the redundant parameter pairs, and the configuration parameters with larger contribution values ​​are retained. After removing the configuration parameters with smaller contribution values ​​from all redundant parameter pairs, the remaining configuration parameters are integrated to form the computer's associated configuration parameter set.

[0128] The beneficial effects are as follows: by accurately locating the abnormal dimension through feature decomposition of the abnormal fluctuation index, and by combining the pre-configured parameter feature label set to achieve accurate matching between configuration parameters and abnormal features, the correlation propagation extends and covers the upper and lower level configuration parameters to ensure the comprehensiveness of the screening of associated parameters. Then, by clustering and redundancy removal based on parameter type and scope, duplicate and low-contribution configuration parameters are removed. The final set of associated configuration parameters can accurately match the abnormal dimension corresponding to the abnormal fluctuation index and has the characteristics of no redundancy and high correlation. This provides a precise and concise parameter adjustment object for subsequent computer parameter self-calibration, improving the efficiency and accuracy of parameter calibration.

[0129] S4. Based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, determine the parameter adjustment direction and adjustment range of the abnormal feature, and generate a specific adjustment instruction for the associated configuration parameter set;

[0130] In this embodiment of the invention, determining the parameter adjustment direction and adjustment range of the abnormal feature based on the correspondence between the abnormal feature and the parameter setting values ​​in the associated configuration parameter set, and generating a specific adjustment instruction for the associated configuration parameter set, includes:

[0131] Extract the anomaly type identifier and anomaly degree of the aforementioned anomaly features;

[0132] The associated configuration parameter set is structured and parsed to obtain the parameter type identifier, current setting value, adjustable range, and adjustment step size of the associated configuration parameter set;

[0133] A correlation analysis is performed on the anomaly type identifier and the parameter type identifier to obtain the matching relationship between the anomaly type identifier and the parameter type identifier, and the adjustment direction of the associated configuration parameter set is determined based on the matching relationship;

[0134] The degree of abnormality is mapped to a level to obtain the adjustment range level of the associated configuration parameter set;

[0135] The adjustment range of the associated configuration parameter set is determined based on the adjustment range level and the adjustment step size;

[0136] The adjustable range and the adjustment magnitude are combined and encoded to obtain the specific adjustment instructions for the associated configuration parameter set.

[0137] Anomaly type identifiers are extracted from the abnormal features of the computer to characterize the operational dimension category to which the anomaly belongs. At the same time, anomaly degree is extracted to reflect the actual impact of the anomaly. The anomaly type identifier and the anomaly degree together constitute the core identification information of the anomaly features.

[0138] For each configuration parameter in the associated configuration parameter set, a structured parsing operation is performed. The parameter type identifier, the current effective setting value, the upper and lower limits of the adjustable value, and the minimum numerical unit of each adjustment are extracted from the parameter configuration information to ensure that the four parsing information of each configuration parameter are completely extracted and correspond one by one.

[0139] The extracted anomaly type identifiers are correlated one by one with the parameter type identifiers of each configuration parameter in the associated configuration parameter set. The corresponding matching relationship between the anomaly type identifier and the parameter type identifier is determined by feature matching. For the configuration parameter that is successfully matched, the adjustment direction corresponding to the configuration parameter is directly determined based on the matching relationship and the underlying logic of computer operation.

[0140] The extracted anomaly levels are classified and mapped. Based on the actual numerical value of the anomaly level, different level intervals are divided. The anomaly level in the corresponding level interval is directly mapped to a fixed adjustment range level, ensuring that each anomaly level can be matched with a unique corresponding adjustment range level.

[0141] The adjustment level obtained from the mapping is combined with the adjustment step size of the corresponding configuration parameter in the associated configuration parameter set for calculation. The adjustment step size is used as the basic unit, and the number of basic units to be adjusted is determined according to the adjustment level. The combined calculation result is the adjustment range of the corresponding configuration parameter in the associated configuration parameter set.

[0142] The adjustable range of each configuration parameter in the associated configuration parameter set is linked and integrated with the determined adjustment range. The two pieces of information are combined and encoded according to the preset encoding rules. The encoded information includes the numerical range limit of parameter adjustment and the specific adjustment value. The encoding result is the specific adjustment instruction for the associated configuration parameter set.

[0143] The beneficial effects are as follows: by extracting anomaly type identifiers and anomaly severity, accurate quantitative identification of anomaly features is achieved; the structured parsing of associated configuration parameter sets can fully obtain all the basic information required for parameter adjustment; determining the adjustment direction based on identifier matching ensures the pertinence and rationality of parameter adjustment; mapping the anomaly severity to the adjustment magnitude level and combining it with the adjustment step size to determine the adjustment magnitude achieves standardized quantification of parameter adjustment magnitude; finally, specific adjustment instructions are generated through the combination encoding of adjustable range and adjustment magnitude, giving the adjustment instructions clear execution standards and range constraints, providing a clear, standardized and directly executable operational basis for subsequent accurate incremental adjustment of parameters, and effectively improving the accuracy and executability of parameter adjustment instruction generation.

[0144] S5. Based on the specific adjustment instruction, the parameter setting value is incrementally adjusted, and the feedback change of the abnormal fluctuation index is monitored in real time to obtain the adjusted setting value of the associated configuration parameter set.

[0145] In this embodiment of the invention, the step of incrementally adjusting the parameter setting value based on the specific adjustment instruction and monitoring the feedback changes of the abnormal fluctuation index in real time to obtain the adjusted setting value of the associated configuration parameter set includes:

[0146] Based on the specific adjustment instruction, obtain the parameter identifier, adjustment direction identifier, and adjustment range identifier of the associated configuration parameter set;

[0147] Based on the dependencies between configuration parameters in the associated configuration parameter set, determine the sequential adjustment order of the associated configuration parameter set;

[0148] The current parameter to be adjusted in the associated configuration parameter set is selected according to the serial adjustment order, and the basic adjustment step size of the computer is determined according to the degree of deviation between the current abnormal characteristic value of the computer and the normal operation characteristic of the computer.

[0149] An incremental adjustment instruction containing the parameter identifier and the basic step size is sent to the computer to perform a single modification and record the intermediate setting value after modification.

[0150] The average rate of change of the abnormal fluctuation index is continuously collected, and the average rate of change is compared with the trend of change of the abnormal features to obtain the abnormal change comparison result of the computer.

[0151] Based on the comparison results of the abnormal changes, the intermediate setting value is iteratively optimized to obtain the adjusted setting value of the associated configuration parameter set.

[0152] The unique parameter identifier corresponding to each configuration parameter in the associated configuration parameter set is extracted one by one from the encoding information of the specific adjustment instructions. At the same time, the adjustment direction identifier used to indicate whether the parameter value is increased or decreased, and the adjustment range identifier that specifies the specific value of the parameter adjustment are extracted to ensure that the three identifiers of each configuration parameter are completely consistent and without deviation.

[0153] The configuration logic and operational relationships of all configuration parameters in the associated configuration parameter set are analyzed, and the sequential dependencies between each configuration parameter are clarified. That is, the parameter association rules that require the completion of the preceding adjustment before the subsequent adjustment can be carried out are determined. Based on the rules, all configuration parameters are sorted to form a unique serial adjustment order.

[0154] Following the established sequential adjustment order, the configuration parameters to be adjusted are selected as the current parameters to be adjusted. The specific numerical differences between the current abnormal characteristic value and the normal operating characteristic of the computer are compared. Based on the actual size of the difference, the appropriate adjustment base step size is determined. This adjustment base step size is the numerical unit of a single modification of the current parameter to be adjusted.

[0155] The parameter identifier containing the parameter to be adjusted and the incremental adjustment command of the determined adjustment base step size are sent to the parameter configuration execution module of the computer. The module then performs a single value modification operation on the parameter setting value of the parameter to be adjusted according to the command. After the modification is completed, the updated value of the parameter is recorded as an intermediate setting value.

[0156] After a single modification of the parameter setting, the system continuously collects data on the changes in the computer's abnormal fluctuation index within a set time period, calculates the average rate of change of the abnormal fluctuation index within that time period, and compares the calculated average rate of change with the original trend of the abnormal characteristics in terms of both value and direction, thus forming the computer's abnormal change comparison results.

[0157] The effect of a single parameter modification on the abnormal fluctuation index is judged based on the comparison results of abnormal changes. If the abnormal fluctuation index does not reach the normal range, the parameter modification operation is continued to be performed and the intermediate setting value is updated based on the current intermediate setting value, according to the adjustment direction indicator and adjustment base step size. This optimization process is repeated until the abnormal fluctuation index returns to the normal range. At this time, the final intermediate setting value is the adjusted setting value of the associated configuration parameter set.

[0158] The beneficial effects include: accurately extracting core identifiers of parameter adjustments from specific adjustment instructions, providing a clear basis for parameter adjustments; determining the sequential adjustment order by analyzing parameter dependencies, avoiding system operation conflicts caused by disordered parameter adjustments; determining the basic adjustment step size by combining the deviation between abnormal and normal characteristic values, making parameter adjustments more closely aligned with actual abnormal situations; achieving fine-grained parameter adjustments through single incremental modifications and recording intermediate setpoints; continuously collecting the average rate of change of abnormal fluctuation indicators and comparing it with abnormal characteristic trends, providing real-time feedback on the improvement effect of parameter adjustments; iteratively optimizing intermediate setpoints based on comparison results, gradually adapting parameter setpoints to system operation requirements; and finally, obtaining adjusted setpoints that accurately solve the problem of abnormal fluctuations in the computer, ensuring the effectiveness and accuracy of computer parameter self-calibration.

[0159] S6. The abnormal feature, the parameter setting value, and the adjusted setting value are associated and recorded as a successful calibration case of the computer.

[0160] In this embodiment of the invention, the step of associating and recording the abnormal feature, the parameter setting value, and the adjusted setting value as a successful calibration case for the computer includes:

[0161] The abnormal features, the parameter settings, and the adjusted settings are integrated into the computer's original case data.

[0162] The original case data is structured and encoded to obtain the standard case record format of the computer;

[0163] The standard case record format is compared item by item with the computer's historical successful cases to obtain similar cases of the computer. The confidence of the similar cases is increased and the timestamps of the similar cases are updated to obtain the case record of the computer.

[0164] Based on the abnormal indicator types and adjusted parameter types in the case records, multi-level index entries are constructed, and the storage location pointers of the case records are respectively attached to the corresponding nodes of the abnormal indicator type index table and the parameter type index table to obtain the successful calibration cases of the computer.

[0165] The abnormal features generated during the computer's parameter self-calibration process, the parameter settings before adjustment, and the adjusted settings after adjustment are all collected. These three types of core data are classified and integrated according to data attributes to form the computer's original case data, which includes abnormal characteristics, parameter status before adjustment, and parameter status after adjustment. This ensures that the original case data fully covers the key information of the calibration process without any data omissions.

[0166] According to the preset structured coding rules, the abnormal features, parameter settings, and adjusted settings in the original case data are defined and formatted. Fixed field identifiers and data storage formats are assigned to each type of data, transforming the unstructured original case data into a computer standard case record format with unified format specifications, ensuring the readability and retrieval of the case data.

[0167] The standardized case record format is compared field by field and item by item with the historical successful cases stored in the computer. Similar cases in the computer that are consistent with the current case in terms of abnormal feature type and parameter adjustment type are identified. The confidence score of the identified similar cases is increased by a preset value to enhance their reference weight. At the same time, the timestamp of the similar cases is updated to the current system time. The updated similar cases and the current standard case record format are integrated to form the case record of the computer.

[0168] Based on the abnormal indicator types and adjusted parameter types in the case records, a multi-level index entry is constructed, including a first-level abnormal indicator type index and a second-level parameter type index. Corresponding index nodes are created for different types of abnormal indicators and parameter types. The storage location pointers of the case records in the storage medium are respectively associated and mounted to the corresponding nodes in the abnormal indicator type index table and the parameter type index table. The case records associated through the index are the successful calibration cases of the computer, realizing the rapid retrieval and retrieval of cases.

[0169] The beneficial effects include: integrating anomaly features, parameter settings, and adjusted settings to form original case data, fully preserving the core process information of parameter calibration; structuring and encoding the original case data to unify the case record format, improving the standardization and management efficiency of case data; comparing with historical successful cases and updating the confidence level and timestamps of similar cases to strengthen the weight of high-reference-value cases, ensuring the timeliness and practicality of the case library; and constructing multi-level indexes based on anomaly indicator types and parameter types and attaching storage pointers to achieve accurate and rapid retrieval of successful calibration cases. This provides directly referable successful case data for subsequent parameter self-calibration of similar computer anomalies, improving the efficiency and accuracy of parameter self-calibration.

[0170] like Figure 2 The diagram shown is a functional block diagram of a computer parameter self-calibration system based on real-time index feedback provided in an embodiment of the present invention.

[0171] The computer parameter self-calibration system 100 based on real-time indicator feedback described in this invention can be installed in an electronic device. Depending on the functions implemented, the computer parameter self-calibration system 100 may include a data collection module 101, an anomaly comparison module 102, a configuration parameter association module 103, an adjustment instruction generation module 104, a parameter adjustment module 105, and a case storage module 106. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0172] In this embodiment, the functions of each module / unit are as follows:

[0173] The data aggregation module 101 is used to aggregate the timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events into the computer's performance index data, and to perform time-series storage on the performance index data to obtain the computer's real-time index change curve.

[0174] The anomaly comparison module 102 is used to extract the rate of change, fluctuation amplitude and trend direction of the real-time indicator change curve within a preset time window, obtain the dynamic features of the real-time indicator change curve, and compare the dynamic features with the normal operation features of the computer to obtain the abnormal fluctuation index and abnormal features of the computer.

[0175] The configuration parameter association module 103 is used to traverse the currently effective configuration parameter set of the computer using the abnormal fluctuation index as an index, and obtain the associated configuration parameter set of the computer.

[0176] The adjustment instruction generation module 104 is used to determine the parameter adjustment direction and adjustment range of the abnormal feature based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, and generate a specific adjustment instruction for the associated configuration parameter set.

[0177] The parameter adjustment module 105 is used to incrementally adjust the parameter setting value based on the specific adjustment instruction, and monitor the feedback changes of the abnormal fluctuation index in real time to obtain the adjusted setting value of the associated configuration parameter set.

[0178] The case storage module 106 is used to associate and record the abnormal features, the parameter setting values, and the adjusted setting values ​​as successful calibration cases of the computer.

[0179] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0180] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0181] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0182] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0183] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0184] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A computer parameter self-calibration method based on real-time indicator feedback, characterized in that, The method includes: S1. The timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events are collected into the computer's performance index data, and the performance index data is stored in a time-series format to obtain the real-time index change curve of the computer. S2. Within a preset time window, extract the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve to obtain the dynamic characteristics of the real-time indicator change curve, and compare the similarity of the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer. S3. Using the abnormal fluctuation index as an index, traverse the currently effective set of configuration parameters of the computer to obtain the associated set of configuration parameters of the computer; S4. Based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, determine the parameter adjustment direction and adjustment range of the abnormal feature, and generate a specific adjustment instruction for the associated configuration parameter set; S5. Based on the specific adjustment instruction, the parameter setting value is incrementally adjusted, and the feedback change of the abnormal fluctuation index is monitored in real time to obtain the adjusted setting value of the associated configuration parameter set. S6. The abnormal feature, the parameter setting value, and the adjusted setting value are associated and recorded as a successful calibration case of the computer.

2. The computer parameter self-calibration method based on real-time index feedback as described in claim 1, characterized in that, The process of aggregating the timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission / reception events into performance index data, and storing the performance index data in a time-series format to obtain the computer's real-time index change curves, includes: Raw event records are captured from the central processing unit scheduler, memory management unit, block device layer and network protocol stack in the computer. The raw event records include a high-precision timestamp of the time when the event occurred and an event type identifier. Based on the event type identifier, extract the context switching count of the CPU scheduling event, the page fault type of the memory access event, the request size and priority of the disk input / output event, and the message length and transmission protocol type of the network data transmission and reception event in the computer to obtain the key attributes of the computer. According to the high-precision timestamp, the key attributes are filled into the computer's circular buffer, and an interrupt is triggered when the circular buffer overflows to notify the computer's user-mode program to read the data; Remove incomplete records from the circular buffer and correct any misalignments in the records within the circular buffer to obtain the computer's performance metrics data. Based on the event type identifier, the performance index data is mapped to a two-dimensional coordinate system with time as the horizontal axis and index value as the vertical axis, and adjacent index data points are connected by straight line segments to obtain the real-time index change curve of the computer.

3. The computer parameter self-calibration method based on real-time index feedback as described in claim 1, characterized in that, Within a preset time window, the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve are extracted to obtain the dynamic characteristics of the real-time indicator change curve. These dynamic characteristics are then compared with the normal operating characteristics of the computer to obtain abnormal fluctuation indicators and abnormal characteristics of the computer, including: Using a preset time window as a sliding window, the real-time indicator change curve is divided into curve segments, and the curve segments are discretized and sampled to obtain the first sequence of the computer. The first sequence is analyzed for trend to obtain the trend direction identifier of the curve segment; Historical indicator segments corresponding to the preset time window are extracted from the computer's operation log, and the historical indicator segments are discretized and sampled to obtain the computer's second sequence; Calculate the dynamic characteristic deviation between the first sequence and the second sequence to obtain the abnormal fluctuation index of the computer; The abnormal fluctuation index is compared and analyzed with the fluctuation tolerance threshold of the computer to obtain the abnormal segments of the computer, and the abnormal segments and the trend direction identifiers corresponding to the abnormal segments are recorded as the abnormal features of the computer.

4. The computer parameter self-calibration method based on real-time index feedback as described in claim 3, characterized in that, The formula for calculating the abnormal fluctuation index is as follows: ； in, This indicates the abnormal fluctuation index. This represents the total number of discrete sampling points within the preset time window. Indicates the first sequence of the first sequence The time series value of each sampling point Indicates the second sequence of the first... The time series value of each sampling point Indicates the second sequence of the first... Historical standard deviation of the sampling points This represents a pre-defined non-zero minimum constant. Indicates the first sequence of the first sequence The instantaneous rate of change of each sampling point This represents the historical average rate of change of the second sequence.

5. The computer parameter self-calibration method based on real-time index feedback as described in claim 1, characterized in that, The step of using the abnormal fluctuation index as an index to traverse the currently effective set of configuration parameters of the computer to obtain the associated set of configuration parameters of the computer includes: The abnormal fluctuation index is decomposed to obtain the abnormal feature components of the computer and the abnormal intensity corresponding to the abnormal feature components. Obtain the currently effective set of configuration parameters for the computer, wherein the set of configuration parameters is pre-configured with a set of parameter feature tags; For the configuration parameter set, the parameter feature label set is matched with the abnormal feature components, and the abnormal intensity is weighted and fused to obtain the initial relevance of the configuration parameter set; The initial relevance is passed to the upper-level and lower-level configuration parameters of the configuration parameter set to obtain the extended relevance of the configuration parameter set; Based on the extended relevance, the configuration parameter set is sorted and filtered to obtain the initial associated configuration parameter set of the computer; Redundancy is removed from the initial set of associated configuration parameters to obtain the set of associated configuration parameters for the computer.

6. The computer parameter self-calibration method based on real-time index feedback as described in claim 5, characterized in that, The step of removing redundancy from the initial associated configuration parameter set to obtain the associated configuration parameter set of the computer includes: Obtain the parameter type identifier and parameter scope of the configuration parameters in the initial associated configuration parameter set; Based on the parameter type identifier, cluster analysis is performed on the associated configuration parameter set to obtain the parameter group of the computer; Configuration parameters with overlapping scopes in the parameter group are marked as redundant parameter pairs for the computer; The computer's associated configuration parameter set is obtained by removing the configuration parameters with low contribution from the redundant parameter pairs in the initial associated configuration parameter set.

7. The computer parameter self-calibration method based on real-time index feedback as described in claim 1, characterized in that, The step of determining the parameter adjustment direction and adjustment range of the abnormal feature based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, and generating a specific adjustment instruction for the associated configuration parameter set, includes: Extract the anomaly type identifier and anomaly degree of the aforementioned anomaly features; The associated configuration parameter set is structured and parsed to obtain the parameter type identifier, current setting value, adjustable range, and adjustment step size of the associated configuration parameter set; A correlation analysis is performed on the anomaly type identifier and the parameter type identifier to obtain the matching relationship between the anomaly type identifier and the parameter type identifier, and the adjustment direction of the associated configuration parameter set is determined based on the matching relationship; The degree of abnormality is mapped to a level to obtain the adjustment range level of the associated configuration parameter set; The adjustment range of the associated configuration parameter set is determined based on the adjustment range level and the adjustment step size; The adjustable range and the adjustment magnitude are combined and encoded to obtain the specific adjustment instructions for the associated configuration parameter set.

8. The computer parameter self-calibration method based on real-time index feedback as described in claim 1, characterized in that, The step of incrementally adjusting the parameter setting value based on the specific adjustment instruction and monitoring the feedback changes of the abnormal fluctuation index in real time to obtain the adjusted setting value of the associated configuration parameter set includes: Based on the specific adjustment instruction, obtain the parameter identifier, adjustment direction identifier, and adjustment range identifier of the associated configuration parameter set; Based on the dependencies between configuration parameters in the associated configuration parameter set, determine the sequential adjustment order of the associated configuration parameter set; The current parameter to be adjusted in the associated configuration parameter set is selected according to the serial adjustment order, and the basic adjustment step size of the computer is determined according to the degree of deviation between the current abnormal characteristic value of the computer and the normal operation characteristic of the computer. An incremental adjustment instruction containing the parameter identifier and the basic step size is sent to the computer to perform a single modification and record the intermediate setting value after modification. The average rate of change of the abnormal fluctuation index is continuously collected, and the average rate of change is compared with the trend of change of the abnormal features to obtain the abnormal change comparison result of the computer. Based on the comparison results of the abnormal changes, the intermediate setting value is iteratively optimized to obtain the adjusted setting value of the associated configuration parameter set.

9. The computer parameter self-calibration method based on real-time index feedback as described in claim 1, characterized in that, The step of associating and recording the abnormal characteristics, the parameter settings, and the adjusted settings as a successful calibration case for the computer includes: The abnormal features, the parameter settings, and the adjusted settings are integrated into the computer's original case data. The original case data is structured and encoded to obtain the standard case record format of the computer; The standard case record format is compared item by item with the computer's historical successful cases to obtain similar cases of the computer. The confidence of the similar cases is increased and the timestamps of the similar cases are updated to obtain the case record of the computer. Based on the abnormal indicator types and adjusted parameter types in the case records, multi-level index entries are constructed, and the storage location pointers of the case records are respectively attached to the corresponding nodes of the abnormal indicator type index table and the parameter type index table to obtain the successful calibration cases of the computer.

10. A computer parameter self-calibration system based on real-time index feedback, characterized in that, The system for implementing the computer parameter self-calibration method based on real-time index feedback as described in claim 1 includes: The data aggregation module is used to aggregate the timestamps and attributes of the computer's central processing unit scheduling, memory access, disk input / output, and network data transmission and reception events into the computer's performance index data, and to store the performance index data in a time-series format to obtain the computer's real-time index change curve. The anomaly comparison module is used to extract the rate of change, fluctuation amplitude, and trend direction of the real-time indicator change curve within a preset time window, obtain the dynamic characteristics of the real-time indicator change curve, and compare the dynamic characteristics with the normal operation characteristics of the computer to obtain the abnormal fluctuation indicators and abnormal characteristics of the computer. The configuration parameter association module is used to traverse the currently effective configuration parameter set of the computer using the abnormal fluctuation index as an index, and obtain the associated configuration parameter set of the computer. The adjustment instruction generation module is used to determine the parameter adjustment direction and adjustment range of the abnormal feature based on the correspondence between the abnormal feature and the parameter setting value in the associated configuration parameter set, and generate a specific adjustment instruction for the associated configuration parameter set. The parameter adjustment module is used to incrementally adjust the parameter setting value based on the specific adjustment instruction, and monitor the feedback changes of the abnormal fluctuation index in real time to obtain the adjusted setting value of the associated configuration parameter set. The case storage module is used to associate and record the abnormal features, the parameter settings, and the adjusted settings as successful calibration cases of the computer.

Images