Information verification method and storage medium based on keyboard leather case life cycle management system

Abstract

The present invention relates to the field of information verification technology, and specifically to an information verification method and storage medium based on a keyboard case life cycle management system, comprising the following steps: S1: collecting multi-dimensional information of a keyboard case; S2: setting an adaptive dynamic threshold for real-time adjustment of the data fluctuation range; S3: using a fast response algorithm to detect short-term data fluctuations, and identifying anomalies in long-term trends through trend analysis, thereby distinguishing instantaneous fluctuations from gradually formed trend anomalies; S4: performing cross-validation to determine whether the anomaly actually exists; S5: when S4 is confirmed as abnormal data, an abnormality report will be generated; S6: using a self-learning algorithm to update the dynamic threshold and detection strategy. The present invention, through the application of multi-dimensional data cross-validation and self-learning algorithms, realizes dynamic threshold adjustment and anomaly detection optimization, effectively improving the accuracy of keyboard case information verification and the automatic management capability of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of information verification, and in particular to an information verification method and a storage medium based on a keyboard leather case life cycle management system. Background Art

[0002] With the popularization of electronic devices, keyboard leather cases, as a widely used protective accessory, undertake multiple functions such as protection, support and input assistance. However, in the process of its life cycle management, a large amount of data collection and analysis is involved, including real-time monitoring of multi-dimensional data such as frequency of use, key pressure, ambient temperature and humidity, and physical wear. These data are not only used to judge the usage status of the keyboard leather case, but also need to be processed through algorithms to ensure the accuracy and completeness of the information. However, the existing keyboard leather case management system lacks comprehensive analysis of multi-dimensional data and effective anomaly detection mechanism, which leads to false alarms or omissions during information verification, reducing the reliability of the system.

[0003] When facing the multi-dimensional data generated during the use of the keyboard leather case, the existing technology often relies on a single dimension or a fixed threshold to perform anomaly detection, which cannot fully consider the dynamic changes of the data and the correlation between different dimensions. This method is not only not flexible enough in complex usage environments, but also easily leads to inaccurate threshold settings, increasing the probability of false alarms. Therefore, there is an urgent need for a method that can use a self-learning algorithm and combine the correlation of multi-dimensional data to perform dynamic threshold adjustment and anomaly detection optimization, so as to improve the accuracy and reliability of information verification, thereby solving the technical difficulties existing in the existing technology. Summary of the invention

[0004] Based on the above purpose, the present invention provides an information verification method and storage medium based on a keyboard leather case life cycle management system.

[0005] The information verification method based on the keyboard leather case life cycle management system includes the following steps:

[0006] S1: Collect the usage frequency, pressure data, ambient temperature and humidity, and physical wear information of the keyboard leather case, and perform denoising and time synchronization processing on the collected data;

[0007] S2: Based on historical data and data collected by S1, the machine learning algorithm is used to set adaptive dynamic thresholds to adjust the data fluctuation range in real time to adapt to different usage scenarios and user behaviors;

[0008] S3: After the dynamic threshold is set, the fast response algorithm is used to detect short-term data fluctuations, and the anomalies in the long-term trend are identified through trend analysis, thereby distinguishing between instantaneous fluctuations and gradually formed trend anomalies;

[0009] S4: When an anomaly is detected, cross-validation is performed based on multi-dimensional data such as usage frequency, pressure, ambient temperature and humidity, and physical wear data to determine whether the anomaly actually exists;

[0010] S5: When S4 confirms abnormal data, an abnormality report will be generated and the system administrator will be notified for manual confirmation or processing;

[0011] S6: Provide feedback on confirmed abnormal data and use self-learning algorithms to update dynamic thresholds and detection strategies.

[0012] Optionally, the S1 specifically includes:

[0013] S11: The usage frequency of each key is recorded in real time by a micro counting sensor embedded under the keyboard key. The sensor is installed at the bottom of the keyboard leather cover and is in direct contact with each key to record the number of times each key is pressed;

[0014] S12: Use the pressure sensor embedded in the bottom of the keyboard leather cover to monitor the pressure data generated by each key press in real time;

[0015] S13: Collecting ambient temperature and humidity data in real time through a temperature and humidity sensor installed inside the keyboard leather cover;

[0016] S14: using a resistance sensor embedded in the surface of the keyboard leather cover to monitor the wear condition thereof, wherein the resistance sensor determines the wear degree by detecting the change of the resistance value of the leather cover surface material;

[0017] S15: De-noising all collected data, using a denoising algorithm based on wavelet transform to perform multi-scale decomposition on the signal, and thresholding the wavelet coefficients using a preset threshold, thereby filtering out random noise in the signal;

[0018] S16: Use the synchronized clock signal to timestamp the data from different sensors to ensure the time consistency of each data point.

[0019] Optionally, the S2 specifically includes:

[0020] S21: Based on the real-time data collected by S1 and the historical data stored in the system, the use frequency, pressure data, ambient temperature and humidity, and physical wear information of the keyboard leather case are preprocessed; the preprocessing includes data denoising, missing value filling, and unit standardization, which is used to unify data from different sources and time into a unified scale;

[0021] S22: extracting predetermined features from historical data and real-time data, including the changing trend of key usage frequency, the fluctuation range of key pressure, the changing rate of ambient temperature and humidity, and the physical wear rate, to provide effective input data for the machine learning algorithm;

[0022] S23: constructing a training data set based on the extracted features, wherein the training data set is composed of historical data samples, the samples cover typical data in different usage scenarios, and annotate corresponding threshold ranges;

[0023] S24: Apply a machine learning algorithm based on random forest to learn the constructed training data set. The training process automatically identifies the complex relationships and feature weights between different features by constructing multiple decision trees, so that the final constructed model can predict dynamic thresholds in real time based on the input feature data and adapt to data fluctuations and changes in usage scenarios;

[0024] S25: Generate and continuously adjust dynamic thresholds based on real-time input data through the trained random forest model.

[0025] Optionally, the S24 specifically includes:

[0026] S241: In the random forest algorithm, multiple decision trees are constructed to learn the relationship between input features and thresholds. Each decision tree splits by recursively splitting the input data set. In each splitting process, the optimal feature node is selected based on the criterion of maximizing the Gini coefficient or information gain.

[0027] S242: Calculate the importance of features. The contribution of each feature to splitting in different decision trees is measured by the reduction of information gain or Gini coefficient.

[0028] S243: automatically identifying the importance of each feature by constructing multiple decision trees, wherein the final weight of the feature is the average value of the importance of the feature in each decision tree;

[0029] S244: Through the constructed random forest model, after the feature data is input in real time, each decision tree independently predicts the threshold; the output results of the decision tree are calculated by majority voting method or average value to generate the final dynamic threshold.

[0030] Optionally, the S3 specifically includes:

[0031] S31: After the dynamic threshold is set, the real-time data is detected using a fast response algorithm. The fast response algorithm is based on a sliding window mechanism and sets a fixed-length time window. , analyze the data in each time period; for each new data point , and compare it with the average value in the previous time window Compare, if satisfied ,in, If it is a dynamic threshold, it is judged as short-term fluctuation abnormality;

[0032] S32: Identify trend changes in long-term data through trend analysis algorithms; long-term trends are identified by calculating the weighted average of data in multiple time periods, and the formula is: ,in, For the moment The long-term trend value of For the moment data, is the weighting coefficient; by analyzing the trend value When the trend value exceeds the dynamic threshold continuously, When , it is judged as a trend abnormality;

[0033] S33: If the short-term fluctuation amplitude exceeds the set threshold but the fluctuation time is short and the long-term trend has not changed significantly, it is judged as an instantaneous fluctuation; conversely, if the fluctuation duration is long and is accompanied by continuous changes in the long-term trend, it is judged as a trend anomaly.

[0034] Optionally, the S4 specifically includes:

[0035] S41: Based on the threshold range of each dimension in the historical data, single-dimensional verification is performed on the usage frequency, pressure, ambient temperature and humidity, and physical wear data one by one to determine whether the data of each dimension is within a reasonable range. If the data of a certain dimension exceeds its reasonable range, it is marked as a possible abnormality.

[0036] S42: Based on the single-dimensional verification, the correlation analysis of multi-dimensional data is performed by calculating the correlation coefficient between different dimensional data. , to determine whether there is an abnormal correlation between the data of each dimension; if the correlation coefficients of multiple dimensions deviate significantly from the historical normal values, it may be a real anomaly;

[0037] S43: Based on the results of single-dimensional verification and multi-dimensional data correlation analysis, cross-validate data of different dimensions, and judge whether the data anomalies of each dimension are consistent by comprehensively considering the fluctuations of frequency of use, key pressure, ambient temperature and humidity, and physical wear data; if the data anomalies of multiple dimensions show strong correlation and consistent trends, the system determines that the anomaly is a true anomaly; otherwise, it is a false alarm.

[0038] Optionally, the S43 specifically includes:

[0039] S431: standardizing data of different dimensions including usage frequency, key pressure, ambient temperature and humidity, and physical wear;

[0040] S432: For each standardized data, calculating its fluctuation range within a predetermined time period;

[0041] S433: Calculate the fluctuation of all dimensional data and analyze whether the fluctuations of various dimensions are consistent;

[0042] S434: Based on the fluctuation consistency analysis, combined with the correlation between the various dimensions calculated in S42 Perform weighted processing. If the correlation between some dimensions is strong, a higher weight will be given to the dimension;

[0043] S435: After analyzing the fluctuation range and correlation of multi-dimensional data, a comprehensive judgment is made to determine whether there are multi-dimensional consistent anomalies. If the fluctuations of multiple dimensions are highly consistent and the correlation is strong, it will be determined as a true anomaly; otherwise, it will be regarded as a false alarm.

[0044] Optionally, the S5 specifically includes:

[0045] S51: When S4 confirms that a certain data is abnormal data, an abnormal report will be automatically generated, which includes the abnormal data source, trigger conditions, abnormal analysis results of multi-dimensional data such as usage frequency, key pressure, ambient temperature and humidity, and physical wear. The abnormal report will also record the time when the abnormality occurred, the duration of the abnormality, and the relevant historical data involved;

[0046] S52: The generated exception report will be automatically stored in a preset exception record database and associated with historical data and processing logs related to the corresponding exception, wherein each exception record will be accompanied by a unique identifier;

[0047] S53: Push the generated exception report to the system administrator in real time;

[0048] S54: After receiving the exception notification, the system administrator views the details of the exception report through the provided interactive interface and performs manual confirmation based on the data analysis results in the report. If the administrator confirms that the exception is a false alarm, the alarm is canceled through the system; if the exception is a real exception, appropriate processing measures are taken according to the suggestions provided in the report.

[0049] Optionally, the S6 specifically includes:

[0050] S61: The abnormal data confirmed by the system administrator will be automatically stored in the abnormal database. Each abnormal data includes the trigger condition, abnormal type, related multi-dimensional data, and context data when the threshold is set, which serves as the basic sample for model update;

[0051] S62: constructing a training data set for the self-learning algorithm based on the stored abnormal data and the corresponding historical normal data, wherein each training sample consists of multi-dimensional data and its corresponding abnormal determination result;

[0052] S63: A self-learning algorithm based on a gradient boosting decision tree is used to train the constructed training data set. The self-learning algorithm analyzes the multi-dimensional features in the abnormal data and optimizes the decision tree structure in the model, allowing the system to gradually learn the changes in the abnormal data pattern through multiple rounds of training, and automatically adjust the dynamic threshold and detection strategy according to the training results.

[0053] The present application also provides a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the information verification method described above are implemented.

[0054] Beneficial effects of the present invention:

[0055] The present invention, by introducing cross-validation and self-learning algorithms of multi-dimensional data, can realize a comprehensive analysis of data such as usage frequency, key pressure, ambient temperature and humidity, and physical wear during the life cycle management of the keyboard cover. The method can dynamically adjust the detection threshold according to changes in the actual use environment, ensuring that the system can capture abnormal data in a timely and accurate manner, effectively reducing the probability of false alarms and missed alarms.

[0056] The present invention, through the application of self-learning algorithm, enables the system to automatically learn the changes in abnormal data patterns and gradually optimize the detection model. The feedback data after each abnormality confirmation is used to continuously improve the model, so that the system can maintain efficient abnormality detection performance in long-term operation. This mechanism not only improves the accuracy of keyboard leather case information verification, but also enhances the stability and automatic processing capabilities of the system, thereby greatly extending the effective life cycle management of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 A schematic diagram of the information verification method flow in an embodiment of the present invention;

[0059] Figure 2 The cross-validation process diagram of the embodiment of the present invention is shown in FIG. DETAILED DESCRIPTION

[0060] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0061] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0062] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0063] like Figure 1-Figure 2 As shown, the information verification method based on the keyboard leather case life cycle management system includes the following steps:

[0064] S1: Collect the usage frequency, pressure data, ambient temperature and humidity, and physical wear information of the keyboard leather case, and perform denoising and time synchronization on the collected data to ensure the continuity and consistency of the data;

[0065] S2: Based on historical data and data collected by S1, the machine learning algorithm is used to set adaptive dynamic thresholds to adjust the data fluctuation range in real time to adapt to different usage scenarios and user behaviors, and reduce false alarms caused by fixed thresholds;

[0066] S3: After the dynamic threshold is set, the fast response algorithm is used to detect short-term data fluctuations, and the anomalies in the long-term trend are identified through trend analysis, thereby distinguishing between instantaneous fluctuations and gradually formed trend anomalies;

[0067] S4: When an anomaly is detected, cross-validation is performed based on multi-dimensional data such as usage frequency, pressure, ambient temperature and humidity, and physical wear data to determine whether the anomaly actually exists. The correlation analysis of different data sources is used to reduce the occurrence of false alarms.

[0068] S5: When S4 confirms abnormal data, a detailed abnormality report will be generated and the system administrator will be notified for manual confirmation or processing;

[0069] S6: Provide feedback on confirmed abnormal data and use self-learning algorithms to update dynamic thresholds and detection strategies, so that anomalies can be identified more accurately in future verification processes and the false alarm rate can be continuously optimized.

[0070] S1 specifically includes:

[0071] S11: The frequency of use of each key is recorded in real time through a micro counting sensor embedded under the keyboard keys. The sensor is installed at the bottom of the keyboard leather cover and is in direct contact with each key to record the number of times each key is pressed. The data is transmitted to the processing unit in the form of a digital signal and matched one by one with the corresponding key number to ensure accurate statistics of the key usage frequency.

[0072] S12: Use the pressure sensor embedded in the bottom of the keyboard leather cover to monitor the pressure data generated each time a key is pressed in real time. The pressure sensor is made of piezoelectric material and can generate an electrical signal when the key is subjected to external force. The strength of the electrical signal is proportional to the pressure.

[0073] S13: Real-time collection of ambient temperature and humidity data through a temperature and humidity sensor installed on the inner side of the keyboard leather cover; the sensor is located at the center of the keyboard leather cover to ensure uniform monitoring of ambient temperature and humidity changes, and the sensor converts temperature and humidity signals into corresponding electrical signals;

[0074] S14: A resistance sensor embedded in the surface of the keyboard leather cover is used to monitor its wear. The resistance sensor determines the degree of wear by detecting the change in the resistance value of the leather cover surface material. Specifically, when the leather cover surface is worn, the resistance value will change. Thickness of leather case Inversely proportional, ,in is the resistivity of the material, is the length of the sensor; by continuously monitoring the change of the resistance value, the wear condition of the leather case can be determined in real time;

[0075] S15: De-noise all collected data by using a wavelet transform-based denoising algorithm to perform multi-scale decomposition on the signal. Suppose the collected signal is , through the wavelet basis function Transform the signal to obtain the wavelet coefficients ,in, is the scale factor, is the translation factor, and the wavelet coefficients are thresholded by a preset threshold, thereby filtering out random noise in the signal and retaining the main features of the signal;

[0076] S16: Use the synchronous clock signal to timestamp the data from different sensors to ensure the time consistency of each data point. Time synchronization is generated by a high-precision clock source, and the timestamp is embedded in the data frame of each sensor to ensure that all data can be analyzed and processed under the same time reference. The above steps can ensure the accurate collection of the keyboard leather case's usage frequency, pressure data, ambient temperature and humidity, and physical wear information. At the same time, combined with the principle of resistance sensors and wavelet transform denoising technology, the accuracy, real-time and stability of the data are improved, providing a reliable data basis for subsequent anomaly detection and effectively reducing noise interference in the collected data.

[0077] The adaptive dynamic threshold setting in S2 specifically includes:

[0078] S21: Based on the real-time data collected by S1 and the historical data stored in the system, the keyboard leather case usage frequency, pressure data, ambient temperature and humidity, and physical wear information are preprocessed; the preprocessing includes data denoising, missing value filling, and unit standardization, which is used to unify data from different sources and time into a unified scale;

[0079] S22: extracting predetermined features from historical data and real-time data, including the changing trend of key usage frequency, the fluctuation range of key pressure, the changing rate of ambient temperature and humidity, and the physical wear rate, to provide effective input data for the machine learning algorithm;

[0080] S23: Construct a training data set based on the extracted features. The training data set consists of historical data samples. The samples cover typical data in different usage scenarios and are annotated with corresponding threshold ranges. Through these diverse samples, we ensure that the machine learning model can be generalized to more practical scenarios.

[0081] S24: Apply a machine learning algorithm based on random forest to learn the constructed training data set. The training process automatically identifies the complex relationships and feature weights between different features by constructing multiple decision trees, so that the final constructed model can predict dynamic thresholds in real time based on the input feature data and adapt to data fluctuations and changes in usage scenarios;

[0082] S25: Generate and continuously adjust dynamic thresholds based on real-time input data through the trained random forest model. The model automatically adapts to the current operating status according to the real-time environment, pressure changes, usage frequency and physical wear information to ensure that the threshold is reasonable and effective, and dynamically adapts to different user behaviors and environmental changes. The above steps use the dynamic thresholds generated by the random forest algorithm through clear feature extraction and data set construction, which can effectively cope with the changes in the use of the keyboard cover in different scenarios and realize real-time and accurate threshold adjustment. This method can not only improve the system's adaptability, but also improve the data monitoring accuracy of the keyboard cover in a changing environment.

[0083] S24 specifically includes:

[0084] S241: In the random forest algorithm, multiple decision trees are constructed to learn the relationship between input features and thresholds. Each decision tree splits by recursively splitting the input data set. In each splitting process, the optimal feature node is selected based on the criterion of maximizing the Gini coefficient or information gain. The Gini coefficient is , the calculation formula is: ,in, is the dataset of the current node, For the Class samples in the dataset The proportion of The total number of categories in the data set is determined by minimizing the Gini coefficient to ensure that the sample category of each node is as pure as possible;

[0085] S242: Calculate the importance of features. The contribution of each feature to the split in different decision trees is measured by the information gain or the reduction of the Gini coefficient. It is calculated as the sum of the reductions in the Gini coefficient when the feature is used as a split node in all decision trees. The specific formula is:

[0086] ,in, Features The importance value of is the total number of decision trees, For the The root node data set of a decision tree, For this decision tree The data set of child nodes, is the number of samples in the child node data set, is the number of samples in the root node data set;

[0087] S243: Automatically identify the importance of each feature and the final weight of the feature by building multiple decision trees is the feature importance in each decision tree The average value is calculated as:

[0088] ,in, Features The combined weight among all decision trees, is the total number of decision trees, Features Importance value in each decision tree;

[0089] S244: Through the constructed random forest model, after inputting feature data in real time, each decision tree Independent prediction threshold; the output of the decision tree is calculated by majority voting or average value to generate the final dynamic threshold, which is expressed as: ,in, is the final generated dynamic threshold, is the total number of decision trees, For the The predicted value of a decision tree; through the above steps, the random forest model can automatically identify the complex relationship between input features and reasonably allocate feature weights by constructing multiple decision trees, and generate dynamic thresholds through the prediction results of each decision tree. The model can respond to data fluctuations and changes in usage scenarios in real time, improving the adaptability and accuracy of the system.

[0090] S3 specifically includes:

[0091] S31: After the dynamic threshold is set, the real-time data is detected using the fast response algorithm. The fast response algorithm is based on a sliding window mechanism and sets a fixed-length time window. , analyze the data in each time period; for each new data point , and compare it with the average value in the previous time window Compare, if satisfied ,in, If it is a dynamic threshold, it is judged as a short-term fluctuation anomaly; the algorithm can quickly identify instantaneous abnormal fluctuations and has a short response time;

[0092] S32: Identify trend changes in long-term data through trend analysis algorithms. Trend analysis algorithms are based on weighted moving average (WMA) to smooth data and set weighting coefficients. To give higher weight to recent data; the identification of long-term trends is determined by calculating the weighted average of data in multiple time periods. The formula is: ,in, For the moment The long-term trend value of For the moment data, is the weighting coefficient; by analyzing the trend value When the trend value exceeds the dynamic threshold continuously, When , it is judged as a trend abnormality;

[0093] S33: In order to distinguish short-term fluctuations from long-term trend anomalies, first calculate the frequency and amplitude of short-term fluctuations. If the short-term fluctuation amplitude exceeds the set threshold but the fluctuation time is short, and the long-term trend has not changed significantly, it is judged as an instantaneous fluctuation; conversely, if the fluctuation lasts for a long time and is accompanied by continuous changes in the long-term trend, it is judged as a trend anomaly; for the detected instantaneous fluctuations and trend anomalies, the system processes them separately. For instantaneous fluctuations, they are marked and continued to be monitored to avoid false alarms; for trend anomalies, cross-validation is performed; through the above steps, the rapid response algorithm and trend analysis algorithm can be effectively utilized to accurately identify short-term fluctuations and long-term trends in the keyboard leather case usage data, and distinguish instantaneous fluctuations from trend anomalies. This method not only improves the efficiency of anomaly detection, but also reduces the false alarm rate, providing a guarantee for the reliability of the system.

[0094] S4 specifically includes:

[0095] S41: Based on the threshold range of each dimension in the historical data, single-dimensional verification is performed on the usage frequency, pressure, ambient temperature and humidity, and physical wear data one by one to determine whether the data of each dimension is within a reasonable range. If the data of a certain dimension exceeds its reasonable range, it is marked as a possible abnormality.

[0096] S42: Based on the single-dimensional verification, the correlation analysis of multi-dimensional data is performed by calculating the correlation coefficient between different dimensional data. , to determine whether there is an abnormal correlation between the data in each dimension. The formula is: ,in, For two dimensions and The correlation coefficient between and They are data of different dimensions. and is the mean of the corresponding dimension; if the correlation coefficients of multiple dimensions deviate significantly from the historical normal value, it may be a real anomaly;

[0097] S43: Based on the results of single-dimensional verification and multi-dimensional data correlation analysis, cross-validate data of different dimensions, and judge whether the data anomalies of each dimension are consistent by comprehensively considering the fluctuations of frequency of use, key pressure, ambient temperature and humidity, and physical wear data; if the data anomalies of multiple dimensions show strong correlation and consistent trends, the system determines that the anomaly is a real anomaly; otherwise, it is a false alarm; through the above steps, use multi-dimensional data cross-validation technology to conduct a comprehensive analysis of trend anomalies from multiple angles such as frequency of use, key pressure, ambient temperature and humidity, and physical wear.

[0098] S43 specifically includes:

[0099] S431: standardizing the data of different dimensions, such as usage frequency, key pressure, ambient temperature and humidity, and physical wear, so that the data of different dimensions are adjusted to the same scale range for effective comparison in subsequent analysis;

[0100] S432: For each standardized data, calculate its fluctuation range within a predetermined time period, such fluctuation range can reflect the change of the data within the time period and help the system identify abnormal fluctuations;

[0101] S433: Calculate the fluctuation of all dimensional data and analyze whether the fluctuations of various dimensions are consistent;

[0102] S434: Based on the fluctuation consistency analysis, combined with the correlation between the various dimensions calculated in S42 Perform weighted processing. If the correlation between some dimensions is strong, a higher weight will be given to the dimension, so that the dimension with strong correlation plays a greater role in consistency judgment, thereby improving the accuracy of anomaly detection;

[0103] S435: After analyzing the fluctuation range and correlation of multi-dimensional data, a comprehensive judgment is made to determine whether there are multi-dimensional consistent anomalies. If the fluctuations of multiple dimensions are highly consistent and the correlation is strong, it will be determined as a real anomaly; otherwise, it will be regarded as a false alarm. Through the above steps, the standardized processing, fluctuation range analysis and correlation weighting methods can be effectively utilized to realize the consistent anomaly judgment of multi-dimensional data. This method enhances the accuracy of anomaly detection and reduces false alarms.

[0104] The steps for cross validation are as follows:

[0105] Standardization processing: Assume that the usage frequency data is , key pressure data is 、Ambient temperature and humidity data are , physical wear data is , the data of each dimension Normalized to , the formula is: ,in, For the The original data of the dimension, is the historical mean of this dimension, is the standard deviation of the dimension. Through standardization, the data of all dimensions are converted to the same scale range for easy comparison;

[0106] Calculation of data fluctuation range: For each standardized dimension data , calculate its fluctuation range within a certain time window. The fluctuation range can be calculated by calculating the standard deviation of the data within the time window To characterize: ,in, For Dimension In the window The fluctuation range within For the window Data at a moment, is the average value of the data in the dimension window, is the number of data points in the time window;

[0107] Calculation of the fluctuation consistency function: After calculating the fluctuation range of each dimension, the system constructs a consistency function , to determine whether the fluctuations in each dimension are consistent; this function is calculated based on the sum of the squares of the fluctuations in each dimension, and the expression is: ,in, represents the fluctuation consistency of all dimensions, if The value exceeds the preset threshold , it indicates that the data fluctuations in each dimension are significant and consistent, which may be a real anomaly;

[0108] Weighted processing of correlation coefficient: In order to enhance the accuracy of consistency judgment, the multi-dimensional correlation coefficient in step S42 is combined , weight the consistency function. If the correlation between two dimensions is strong, give it a higher weight. ; The updated consistency function expression is: ,in, For Dimension The weight of the dimension reflects the correlation between the dimension and other dimensions. Through weighted processing, the system can more accurately determine whether the abnormal data is consistent.

[0109] Final judgment of abnormal consistency: When the weighted consistency function The value exceeds the threshold , and the fluctuations of the dimensions with strong correlation are consistent, the system determines that the anomaly is a real anomaly. Otherwise, if the consistency function value is low or the correlation is insufficient, the system considers it to be a possible false alarm.

[0110] S5 specifically includes:

[0111] S51: When S4 confirms that a certain data is abnormal data, it will automatically generate an abnormal report, which includes the abnormal data source, trigger conditions, abnormal analysis results of multi-dimensional data such as usage frequency, key pressure, ambient temperature and humidity, and physical wear. The abnormal report will also record the time when the abnormality occurred, the duration of the abnormality, and the relevant historical data involved, so that the system administrator can fully understand the abnormal situation;

[0112] S52: The generated exception report will be automatically stored in a preset exception record database and associated with the historical data and processing log related to the corresponding exception, wherein each exception record will be accompanied by a unique identifier to facilitate subsequent retrieval and comparison;

[0113] S53: Push the generated exception report to the system administrator in real time, and the notification method includes email, system notification or mobile application. When the exception report is pushed, the administrator will receive a prompt immediately, so that he can manually confirm or process the abnormal data;

[0114] S54: After receiving the exception notification, the system administrator can view the details of the exception report through the provided interactive interface and manually confirm it based on the data analysis results in the report. If the administrator confirms that the exception is a false alarm, the alarm will be canceled through the system; if the exception is a real exception, appropriate processing measures will be taken according to the suggestions provided in the report, and the cause of the exception will be further analyzed or the system threshold settings will be adjusted to ensure the accuracy of subsequent data monitoring; through the above steps, the system can automatically generate a detailed exception report after confirming the exception, and promptly notify the administrator for manual confirmation or processing. This solution realizes the organic combination of automated exception detection and manual intervention to ensure that abnormal data is processed in a timely manner.

[0115] S6 specifically includes:

[0116] S61: The abnormal data confirmed by the system administrator will be automatically stored in the abnormal database. Each abnormal data includes the trigger condition, abnormal type, related multi-dimensional data (frequency of use, key pressure, ambient temperature and humidity, and physical wear information), and context data when the threshold is set, which serves as the basic sample for model update;

[0117] S62: constructing a training data set for the self-learning algorithm based on the stored abnormal data and the corresponding historical normal data, wherein each training sample consists of multi-dimensional data and its corresponding abnormal determination result, so as to ensure that the model can learn the difference between abnormal and normal data when updating, thereby enhancing the generalization ability of the model;

[0118] S63: A self-learning algorithm based on gradient boosted decision tree (GBDT) is used to train the constructed training data set. The self-learning algorithm analyzes the multi-dimensional features in the abnormal data and optimizes the decision tree structure in the model, so that the system can gradually learn the changes in the abnormal data pattern through multiple rounds of training, and automatically adjust the dynamic threshold and detection strategy according to the training results to improve the ability to identify abnormal data. Through the above steps, the self-learning algorithm can be used to continuously optimize the detection model based on the confirmed abnormal data, so that the system can adapt to the changes in the abnormal data pattern and improve its recognition accuracy and automatic processing capabilities for abnormalities.

[0119] The present application also provides a storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned information verification method are implemented.

[0120] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0121] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An information verification method based on a keyboard case life cycle management system, characterized in that: The following steps are involved: S1: Collect the usage frequency, pressure data, ambient temperature and humidity, and physical wear information of the keyboard leather case, and perform denoising and time synchronization processing on the collected data; The S1 specifically includes: S11: The usage frequency of each key is recorded in real time by a micro counting sensor embedded under the keyboard key. The sensor is installed at the bottom of the keyboard leather cover and is in direct contact with each key to record the number of times each key is pressed; S12: Use the pressure sensor embedded in the bottom of the keyboard leather cover to monitor the pressure data generated by each key press in real time; S13: Collecting ambient temperature and humidity data in real time through a temperature and humidity sensor installed inside the keyboard leather cover; S14: using a resistance sensor embedded in the surface of the keyboard leather cover to monitor the wear condition thereof, wherein the resistance sensor determines the wear degree by detecting the change of the resistance value of the leather cover surface material; S15: De-noising all collected data, using a denoising algorithm based on wavelet transform to perform multi-scale decomposition on the signal, and thresholding the wavelet coefficients using a preset threshold, thereby filtering out random noise in the signal; S16: Use a synchronized clock signal to timestamp the data from different sensors to ensure the time consistency of each data point; S2: Based on historical data and data collected by S1, the machine learning algorithm is used to set adaptive dynamic thresholds to adjust the data fluctuation range in real time to adapt to different usage scenarios and user behaviors; The S2 specifically includes: S21: Based on the real-time data collected by S1 and the historical data stored in the system, the use frequency, pressure data, ambient temperature and humidity, and physical wear information of the keyboard leather case are preprocessed; the preprocessing includes data denoising, missing value filling, and unit standardization, which is used to unify data from different sources and time into a unified scale; S22: extracting predetermined features from historical data and real-time data, including the changing trend of key usage frequency, the fluctuation range of key pressure, the changing rate of ambient temperature and humidity, and the physical wear rate, to provide effective input data for the machine learning algorithm; S23: constructing a training data set based on the extracted features, wherein the training data set is composed of historical data samples, the samples cover typical data in different usage scenarios, and annotate corresponding threshold ranges; S24: Apply a machine learning algorithm based on random forest to learn the constructed training data set. The training process automatically identifies the complex relationships and feature weights between different features by constructing multiple decision trees, so that the final constructed model can predict dynamic thresholds in real time based on the input feature data and adapt to data fluctuations and changes in usage scenarios; S25: Generate and continuously adjust dynamic thresholds based on real-time input data through the trained random forest model; S3: After the dynamic threshold is set, the fast response algorithm is used to detect short-term data fluctuations, and the anomalies in the long-term trend are identified through trend analysis, thereby distinguishing between instantaneous fluctuations and gradually formed trend anomalies; S4: When an anomaly is detected, cross-validation is performed based on multi-dimensional data such as usage frequency, pressure, ambient temperature and humidity, and physical wear data to determine whether the anomaly actually exists; The S4 specifically includes: S41: Based on the threshold range of each dimension in the historical data, single-dimensional verification is performed on the usage frequency, pressure, ambient temperature and humidity, and physical wear data one by one to determine whether the data of each dimension is within a reasonable range. If the data of a certain dimension exceeds its reasonable range, it is marked as a possible abnormality. S42: Based on the single-dimensional verification, the correlation analysis of multi-dimensional data is performed by calculating the correlation coefficient between different dimensional data. , to determine whether there is abnormal correlation between the data of each dimension; S43: Based on the results of single-dimensional verification and multi-dimensional data correlation analysis, cross-validate the data of different dimensions, and judge whether the data anomalies of each dimension are consistent by comprehensively considering the fluctuations of usage frequency, key pressure, ambient temperature and humidity, and physical wear data; S5: When S4 confirms abnormal data, an abnormality report will be generated and the system administrator will be notified for manual confirmation or processing; S6: Provide feedback on confirmed abnormal data and use self-learning algorithms to update dynamic thresholds and detection strategies.

2. The information verification method based on the keyboard case life cycle management system according to claim 1 is characterized in that: The S24 specifically includes: S241: In the random forest algorithm, multiple decision trees are constructed to learn the relationship between input features and thresholds. Each decision tree splits by recursively splitting the input data set. In each splitting process, the optimal feature node is selected based on the criterion of maximizing the Gini coefficient or information gain. S242: Calculate the importance of features. The contribution of each feature to splitting in different decision trees is measured by the reduction of information gain or Gini coefficient. S243: automatically identifying the importance of each feature by constructing multiple decision trees, wherein the final weight of the feature is the average value of the importance of the feature in each decision tree; S244: Through the constructed random forest model, after the feature data is input in real time, each decision tree independently predicts the threshold; the output results of the decision tree are calculated by majority voting method or average value to generate the final dynamic threshold.

3. The information verification method based on the keyboard case life cycle management system according to claim 1 is characterized in that: The S43 specifically includes: S431: standardizing data of different dimensions including usage frequency, key pressure, ambient temperature and humidity, and physical wear; S432: For each standardized data, calculating its fluctuation range within a predetermined time period; S433: Calculate the fluctuation of all dimensional data and analyze whether the fluctuations of various dimensions are consistent; S434: Based on the fluctuation consistency analysis, combined with the correlation between the various dimensions calculated in S42 Perform weighted processing and assign weights according to the strength of the correlation between dimensions; S435: After analyzing the fluctuation range and correlation of the multi-dimensional data, a comprehensive judgment is made to determine whether there are multi-dimensional consistent anomalies.

4. The information verification method based on the keyboard case life cycle management system according to claim 1 is characterized in that: The S5 specifically includes: S51: When S4 confirms that a certain data is abnormal data, an abnormal report will be automatically generated, which includes the abnormal data source, trigger conditions, abnormal analysis results of multi-dimensional data such as usage frequency, key pressure, ambient temperature and humidity, and physical wear. The abnormal report will also record the time when the abnormality occurred, the duration of the abnormality, and the relevant historical data involved; S52: The generated exception report will be automatically stored in a preset exception record database and associated with historical data and processing logs related to the corresponding exception, wherein each exception record will be accompanied by a unique identifier; S53: Push the generated exception report to the system administrator in real time; S54: After receiving the exception notification, the system administrator views the details of the exception report through the provided interactive interface and performs manual confirmation based on the data analysis results in the report. If the administrator confirms that the exception is a false alarm, the alarm is canceled through the system; if the exception is a real exception, appropriate processing measures are taken according to the suggestions provided in the report.

5. The information verification method based on the keyboard case life cycle management system according to claim 1 is characterized in that: The S6 specifically includes: S61: The abnormal data confirmed by the system administrator will be automatically stored in the abnormal database. Each abnormal data contains the trigger condition, abnormal type, related multi-dimensional data, and context data when the threshold is set, which serves as the basic sample for model update; S62: constructing a training data set for the self-learning algorithm based on the stored abnormal data and the corresponding historical normal data, wherein each training sample consists of multi-dimensional data and its corresponding abnormal determination result; S63: A self-learning algorithm based on a gradient boosting decision tree is used to train the constructed training data set. The self-learning algorithm analyzes the multi-dimensional features in the abnormal data and optimizes the decision tree structure in the model, allowing the system to gradually learn the changes in the abnormal data pattern through multiple rounds of training, and automatically adjust the dynamic threshold and detection strategy according to the training results.

6. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the steps of the information verification method according to any one of claims 1 to 5.

Images