Fault detection method and system for file cabinet

Abstract

The invention relates to the technical field of file cabinet management, in particular to a fault detection method and system for a file cabinet. The invention aims to solve the problem that the potential fault problem is easy to omit in the fault detection of the existing file cabinet to cause greater loss. According to the method, the normal reference models corresponding to different use modes are established by monitoring and collecting the daily use operation and internal operation state data of the file cabinet, the continuously monitored file cabinet operation data are compared with the normal reference models, and the potential fault data deviating from the normal operation mode in the file cabinet operation data are identified; and analyzing the potential fault data, and identifying a fault type or behavior corresponding to the potential fault data and an influence level of the corresponding fault or behavior on files in the file cabinet. Through the operation, the operation reliability and safety of the file cabinet are greatly improved, and the labor cost and potential risk of fault detection of the file cabinet are effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of filing cabinet management, and in particular to a fault detection method and system for filing cabinets. Background Art

[0002] Among many organizations and enterprises, most rely on filing cabinets to store and manage important documents and information. The security and integrity of files directly affect the operation and decision-making of organizations or enterprises. Therefore, it is necessary to ensure the status of filing cabinets in use to ensure timely access and prevent data loss.

[0003] A failure in a filing cabinet may result in an inability to view stored files or cause the loss of files. Traditional filing cabinet fault detection often relies on regular manual inspections, which is not only time-consuming and labor-intensive, but also prone to missing potential problems. When a failure occurs, it often causes greater losses because the potential problem was not discovered in time.

[0004] Therefore, there is an urgent need to develop a fault detection method and system for filing cabinets to overcome the shortcomings of the prior art. Summary of the invention

[0005] In order to solve the problem that potential faults are easily missed in the prior art and cause greater losses, the present invention provides a fault detection method for a filing cabinet, comprising the following steps: S1: In the early stage of using the filing cabinet, use sensors to monitor and collect the operating data of the filing cabinet, wherein the operating data of the filing cabinet includes the daily operation of the filing cabinet and the internal operating status data of the filing cabinet, and perform fault detection at the same time; S2: Analyze and identify normal operating data in the file cabinet operation data, and establish normal benchmark models corresponding to different usage modes; S3: Continuously monitor the operation data of the filing cabinet, identify the current usage mode, and compare the operation data of the filing cabinet with the normal reference model, identify data in the operation data of the filing cabinet that deviates from the normal operation mode, and mark the data as potential fault data; S4: Analyze the potential fault data, identify the fault type or behavior corresponding to the potential fault data, and determine the impact level on the files inside the filing cabinet; S5: After the fault is diagnosed, a detailed proactive maintenance suggestion plan is generated, and the fault information and proactive maintenance suggestion plan are pushed to the user and maintenance personnel; S6: Record all maintenance information, and use the maintenance information to optimize the judgment conditions in the fault detection method.

[0006] Furthermore, the establishment of normal reference models corresponding to different usage modes includes: S21: performing cluster analysis on the daily usage operation data to identify different usage patterns, where the usage patterns include normal office use, high-frequency use, and low-frequency use; S22: identifying key characteristic parameters from the file cabinet operation data, where the key characteristic parameters are used to describe the normal operation characteristics of the file cabinet under different usage modes; S23: For different usage modes, use a machine learning algorithm to construct normal benchmark models corresponding to different usage modes, and store the normal benchmark models in a benchmark model library.

[0007] Furthermore, the normal benchmark model is constructed using a support vector machine, and the specific calculation formula is: in, To achieve the overall goal of filing cabinet fault detection, Minimize to obtain the normal benchmark model corresponding to different usage modes; is the weight vector; is the bias term, representing the baseline of normal operating conditions; is a penalty parameter used to balance the sensitivity of the normal baseline model to abnormal situations; is the weight coefficient of different usage modes, which is used to dynamically adjust the importance of samples; It is a relaxation variable, which is used to detect small fluctuations in normal use of the filing cabinet without immediately judging the small fluctuations as potential faults.

[0008] Furthermore, the normal benchmark model also includes a dynamic update strategy: S211: After a preset time period ends or when the accumulated file cabinet operation data reaches a preset storage threshold, re-monitor and collect the latest file cabinet operation data; S212: using the same method as the current normal benchmark model training, preprocessing the collected file cabinet operation data to extract key feature parameters; and performing statistical analysis to identify the changing trend of the usage pattern; S213: Compare the change trend with the current normal benchmark model to determine whether there is a significant deviation or change; S214: When there is a significant deviation or change, retrain the machine learning model using the latest file cabinet operation data and update the corresponding normal benchmark model; S215: Use an independent test set to perform performance evaluation on the updated normal baseline model. When the normal baseline model does not meet the performance standard, retrain the normal baseline model until it meets the performance evaluation standard.

[0009] Furthermore, the comparing the file cabinet operation data with the normal benchmark model includes: S31: continuously monitoring and collecting the operation data of the filing cabinet, and preprocessing the collected operation data of the filing cabinet; S32: based on the pre-processed file cabinet operation data, extract key feature parameters in real time within a set sliding time window; S33: using a pre-trained clustering model, performing pattern recognition on the key feature parameters in the current time window, and matching the pattern recognition result with a predefined usage pattern; S34: selecting a corresponding normal benchmark model from the benchmark model library according to the matching result, and comparing the file cabinet operation data in the current time window with the selected normal benchmark model; S35: If the degree of deviation between the filing cabinet operation data and the normal reference model exceeds a preset threshold, the deviated portion of the filing cabinet operation data is marked as potential fault data.

[0010] Furthermore, the setting of the sliding time window includes: S321: Predefine sliding time windows of different time lengths based on different usage patterns, including short-term, medium-term and long-term windows, wherein the short-term window is used to quickly detect frequent and sudden events, the medium-term window is used to detect changes in daily usage patterns, and the long-term window is used to capture trend changes or slowly developing faults over a long period of time; S322: Selecting a suitable sliding time window type according to the current usage mode; S323: For the selected sliding time window, dynamically adjust the window size according to the actual usage frequency and operation data characteristics of the filing cabinet: If the number of times the filing cabinet is accessed exceeds a preset threshold within the short-term window, the length of the sliding time window is shortened to capture anomalies more quickly; Within the limited time of the medium-term window, if the fluctuation range of the file cabinet usage data continues to be lower than the preset range, the length of the sliding window is expanded to observe longer trend data fluctuations; Within the time limit of the long-term window, if there is no abnormality in the previous file cabinet operation data, the length of the sliding time window is expanded; S324: setting an initial anomaly detection threshold for each sliding time window, and evaluating the deviation of data within the sliding time window period from the normal benchmark in real time through mean square error or mean absolute error; and according to the evaluation result, when anomalies are continuously monitored within the sliding time window period and the degree of deviation exceeds a preset threshold, reducing the threshold value according to a preset adjustment amount; S325: Record all sliding time windows and threshold adjustment operation histories and test results, regularly evaluate the effectiveness of test results and window adjustments, and optimize and adjust the sliding time windows and threshold ranges. Evaluation indicators include detection accuracy, recall rate, and F1 score.

[0011] Furthermore, the analyzing of the potential fault data specifically includes: S41: segmenting the time series marked as the potential fault data, and extracting statistical features and frequency domain features of each segment; S42: Analyze the extracted statistical features and frequency domain features using a pre-trained multi-layer fault classification model to identify a specific type of fault or a specific pattern of abnormal behavior; S43: Based on the recognition result, combined with the structural information of the filing cabinet and the sensitivity level of the stored files, a quantitative assessment is made on the impact degree of the files inside the filing cabinet; S44: Based on the quantitative assessment results, the impact level is divided into three levels: low, medium and high, and the corresponding processing priority and corresponding strategy are defined for each level.

[0012] Furthermore, the multi-layer fault classification model includes a primary classification layer, an advanced classification layer and a classification fusion layer; wherein the primary classification layer is used to identify simple patterns or abnormal behaviors in potential fault data by preliminarily classifying the filing cabinet operation data marked as potential fault data; the advanced classification layer is used to perform fine classification of the filing cabinet operation data marked as potential fault data that cannot be identified based on the output of the primary classification layer, and identify more specific fault types or behaviors; the classification fusion layer is used to determine the final fault type by combining the results of the primary classification layer and the advanced classification layer through weighted voting.

[0013] Furthermore, the fault detection method for a filing cabinet also includes: establishing a three-dimensional virtual model of the filing cabinet, and visually displaying the fault point, fault type and impact level at the corresponding position of the three-dimensional virtual model.

[0014] Furthermore, a fault detection system for a filing cabinet includes: A data acquisition module for recording the operation data of the filing cabinet using sensors; A data processing module, used for preprocessing and feature extraction of the file cabinet operation data; A usage mode management module is used to analyze the pre-processed operation data of the filing cabinet, identify the current usage mode of the filing cabinet, and manage and update the normal reference models corresponding to different usage modes; A fault detection module, for comparing the file cabinet operation data with the normal reference model corresponding to the usage mode, identifying potential fault data, and analyzing the potential fault data to identify corresponding fault types and behaviors; The visualization module is used to provide the user with a visualization display of the fault information and fault repair information of the filing cabinet.

[0015] Beneficial effects: 1. The present invention can dynamically identify abnormal situations in daily use and understand the current movement state of the filing cabinet by establishing normal benchmark models corresponding to different usage modes and comparing the operation data of the filing cabinet with the normal benchmark model in real time. This not only significantly improves the accuracy and efficiency of fault detection, but also ensures that the present invention can quickly adapt to different usage conditions by dynamically adjusting the model and detection threshold; 2. The present invention can capture anomalies of different time scales and types by adopting a sliding time window and a multi-level fault classification model. It can not only respond quickly to sudden failures, but also identify slowly developing potential problems, greatly improving the operational reliability and safety of the filing cabinet. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of the method of the present invention; Figure 2 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0017] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0018] This embodiment discloses a fault detection method for a filing cabinet, as shown in the attached Figure 1 As shown, it includes steps S1 to S7: S1: In the early stage of using the filing cabinet, use sensors to record the filing cabinet operation data, where the filing cabinet operation data includes the daily operation data of the filing cabinet and the internal operation status data of the filing cabinet, and perform fault detection at the same time.

[0019] Specifically, sensors for monitoring the daily operation of the filing cabinet by the user and sensors for monitoring the internal operation status of the filing cabinet are installed at key positions of the filing cabinet. The sensors for monitoring the daily operation of the filing cabinet by the user include but are not limited to cameras and pressure sensors, and the sensors for monitoring the internal operation status of the filing cabinet include but are not limited to weight sensors, temperature and humidity sensors, and vibration sensors. The daily operation data include but are not limited to the file access time, the identity of the access personnel, the access file information, the operation duration, and the operation frequency, etc.; the internal operation status data include but are not limited to the temperature and humidity, current, voltage, and vibration intensity felt when the filing cabinet is in operation.

[0020] In the early stage of collecting filing cabinet operation data, preliminary fault detection is also performed. In this early stage, any captured filing cabinet operation data that deviates from the normal operating mode will be immediately marked as potential fault data for rapid diagnosis and processing. At the same time, the normal operating data within this stage is used to establish a normal baseline model corresponding to different usage modes.

[0021] S2: Analyze and identify the normal operation data in the file cabinet operation data, and establish a normal benchmark model corresponding to different usage modes.

[0022] Specifically, in the early stage of the use of the filing cabinet, the normal operation data obtained by monitoring will be used to establish a normal baseline model corresponding to different usage modes, including: S21: Use K-means clustering to perform cluster analysis on the key features of the daily use operation data monitored and collected in the initial stage of filing cabinet use, including but not limited to access frequency, access time period, access users and access item information, to identify different usage patterns, including but not limited to normal office work, high-frequency use and low-frequency use, where normal office work can be the access frequency during normal working hours and the access users are relatively regular, high-frequency use can be the access frequency that is significantly higher than the average level, and low-frequency use can be the access frequency that is significantly lower than the average level.

[0023] S22: Identify key characteristic parameters from daily usage operation data and internal operating status data, where the key characteristic parameters are used to describe the normal operating characteristics of the filing cabinet under different usage modes, and the key characteristic parameters include but are not limited to switching frequency, internal temperature and humidity changes, vibration intensity and specific gas concentration; for different usage modes, the internal operating status data of the filing cabinet will also be different, so the filing cabinet operation data under different modes needs to be fused, and the key characteristic parameters that can effectively distinguish different usage modes are selected from the fused data.

[0024] S23: For different usage modes, use a machine learning algorithm to construct normal benchmark models corresponding to different usage modes, and store the normal benchmark models in a benchmark model library.

[0025] Specifically, the machine learning algorithm used in the construction of the normal benchmark model is the support vector machine, and the specific calculation formula is: in, To achieve the overall goal of filing cabinet fault detection, Minimize and obtain normal benchmark models corresponding to different usage modes; is the weight vector; is the bias term, representing the baseline of normal operating conditions; is a penalty parameter used to balance the sensitivity of the normal baseline model to abnormal conditions; The weighted loss term is adjusted to control the importance scores of different features in different usage modes and optimize the detection accuracy. is the weight coefficient of different usage modes, which is used to dynamically adjust the importance of samples; It is a relaxation variable, which is used to detect small fluctuations in normal use of the filing cabinet without immediately judging the small fluctuations as potential faults.

[0026] After the normal benchmark model is established and stored in the benchmark model library, it is necessary to regularly update the stored normal benchmark model to adapt the normal benchmark model to the changing usage environment and usage habits and reduce the error rate during fault detection. The dynamic update strategy of the normal benchmark model is as follows: S211: After the preset time period ends or when the accumulated file cabinet operation data reaches the preset storage threshold, restart monitoring and collect the latest file cabinet operation data. In dynamic update, because real-time data collection focuses more on immediate status and short-term anomalies, dynamic update needs to analyze data over a longer time range to identify long-term trends and pattern changes, therefore, dynamic update requires additional operations to aggregate and process long-term data.

[0027] S212: Using the same training method as the current benchmark model, pre-process the collected filing cabinet operation data to extract key feature parameters; and perform statistical analysis to identify changing trends in usage patterns.

[0028] S213: Use Euclidean distance to compare the change trend with the current normal benchmark model to identify whether the long-term change in the filing cabinet operation data is still within the normal benchmark range or has deviated from the preset normal mode.

[0029] S214: When there is a significant deviation or change, the machine learning model is retrained using the latest filing cabinet operation data to update the corresponding normal baseline model.

[0030] S215: Use an independent test set to perform performance evaluation on the updated normal baseline model. When the performance standard is not met, retrain the normal baseline model until it meets the performance evaluation standard.

[0031] S3: Continuously monitor the operating data of the filing cabinet, identify the current usage mode, and compare the operating data of the filing cabinet with the normal baseline model, identify data in the operating data of the filing cabinet that deviates from the normal operating mode, and mark the data as potential fault data.

[0032] In this embodiment, the file cabinet operation data is compared with the normal benchmark model, including: S31: Continuously monitor and collect the operation data of the filing cabinet, and pre-process the collected original operation data of the filing cabinet; the pre-processing operations include but are not limited to denoising, filtering and spectrum analysis of vibration data, outlier removal and smoothing of temperature data and humidity data, statistical analysis of usage frequency data, audio feature extraction of abnormal sound data, etc.

[0033] S32: Based on the pre-processed filing cabinet operation data, key feature parameters are extracted in real time within the set sliding time window, wherein the sliding time window is initially of a default length and is subsequently dynamically adjusted according to actual conditions; wherein the key feature parameters include but are not limited to extracting vibration amplitude, vibration frequency and frequency distribution from vibration data, extracting temperature change trend, humidity change trend and temperature and humidity fluctuation range from temperature data and humidity data, extracting the number of uses per unit time, usage time interval and number of uses during non-working hours from usage frequency data, and extracting sound frequency and sound pattern from abnormal sound data.

[0034] S33: Use a pre-trained K-means clustering model to perform pattern recognition on key feature parameters in the current time window, and match the pattern recognition results with a predefined usage pattern.

[0035] S34: According to the matching result, a corresponding normal benchmark model is selected from the benchmark model library, and the file cabinet operation data within the current sliding time window is compared with the selected normal benchmark model.

[0036] S35: If the degree of deviation between the filing cabinet operation data and the normal reference model exceeds a preset threshold, the filing cabinet operation data exceeding the preset threshold is marked as potential fault data.

[0037] Specifically, for potential abnormalities in the internal operating status data of the filing cabinet, the potential fault data is directly analyzed to identify the fault type and behavior corresponding to the potential fault data; For potential abnormalities in the daily operation data of the filing cabinet, an inquiry notification is sent to the user to confirm whether the operation is performed by the user himself or the user is aware of it, and based on the user's feedback: If the user confirms the operation, the daily use operation data is marked as normal operation data, and the subsequent steps are not performed; If the user denies performing the operation, or no feedback is received from the user within the preset inquiry time, the daily usage operation data is marked as abnormal daily usage operation data, and then the potential fault data is analyzed to identify the fault type and behavior corresponding to the potential fault data.

[0038] Specifically, the specific implementation steps of the sliding time window set in step S2 are: S321: Predefine sliding time windows of different lengths based on different usage patterns, including short-term, medium-term and long-term windows; the short-term window is used to quickly detect frequent and sudden events and can be set within minutes; the medium-term window is used to detect changes in daily usage patterns and can be set within hours; the long-term window is used to capture trend changes or slowly developing faults over a long period of time and can be set within days to weeks; S322: Selecting a suitable sliding time window type according to the current usage mode; S323: For the selected sliding time window, dynamically adjust the window size according to the actual usage frequency and operation data characteristics of the filing cabinet: If the number of times the filing cabinet is accessed exceeds the preset threshold within the short-term window limit, the length of the sliding time window is shortened to capture anomalies more quickly; In the limited time of the medium-term window, the fluctuation range of the filing cabinet usage data continues to be lower than the preset range, and the length of the sliding time window is expanded to observe longer trend data fluctuations; Within the long-term window limit, if there is no abnormality in the previous file cabinet operation data, the length of the sliding time window will be expanded; S324: setting an initial anomaly detection threshold for each sliding time window, and evaluating the deviation of data within the sliding time window period from the normal benchmark in real time through mean square error or mean absolute error; and according to the evaluation result, when anomalies are continuously monitored within the sliding time window period and the degree of deviation exceeds the preset threshold, reducing the threshold value according to the preset step size; S325: Record all sliding time windows and threshold adjustment operation histories and test results, regularly evaluate the effectiveness of test results and window adjustments, and optimize and adjust the sliding time windows and threshold ranges. Evaluation indicators include detection accuracy, recall rate, and F1 score.

[0039] S4: Analyze the potential fault data, identify the fault type or behavior corresponding to the potential fault data, and determine the impact level on the files inside the filing cabinet.

[0040] In this embodiment, the analyzing of potential fault data includes: S41: Segment the time series marked as potential fault data, and extract the statistical features and frequency domain features of each segment. When segmenting, first initialize a minimum segment length and a maximum segment length; calculate the local error at each possible segment point; when the local error exceeds the preset error, segment at that point.

[0041] S42: Use a pre-trained multi-layer fault classification model to analyze the extracted statistical features and frequency domain features to identify the specific type of fault or the specific pattern of abnormal behavior.

[0042] Specifically, the multi-layer fault classification model includes a primary classification layer, an advanced classification layer and a classification fusion layer; wherein, the primary classification layer may adopt a support vector machine to preliminarily classify the data marked as potential faults, identify simple patterns or abnormal behaviors in the potential fault data, and output a rough fault category; the advanced classification layer may adopt a random forest algorithm to finely classify the data marked as potential faults that cannot be identified based on the output of the primary classification layer, and identify more specific fault types or behaviors; the classification fusion layer is used to determine the final fault type by combining the results of the primary classification layer and the advanced classification layer through weighted voting.

[0043] S43: Based on the recognition results, combined with the structural information of the filing cabinet and the sensitivity level of the stored files, a quantitative assessment is made on the impact on the files inside the filing cabinet.

[0044] S44: Based on the quantitative assessment results, the impact level is divided into three levels: low, medium and high, and the corresponding processing priority and corresponding strategy are defined for each level.

[0045] S5: After the fault or high-risk factor is confirmed, a detailed proactive maintenance suggestion plan is generated, and minor fault information and proactive maintenance suggestions are pushed to the user according to the fault type and impact level, while severe fault information and proactive maintenance suggestions are pushed to the user and maintenance personnel at the same time. Among them, the fault information includes but is not limited to the fault type, fault time, fault location, fault registration and impact range, and the proactive maintenance suggestion plan includes but is not limited to the recommended maintenance measures, required maintenance tools and materials, and maintenance precautions.

[0046] S6: Record all maintenance information, and use the maintenance information to optimize the judgment conditions in the fault detection method.

[0047] S7: Establish a three-dimensional virtual model of the filing cabinet, the three-dimensional virtual model includes the components of the filing cabinet and sensor location information; according to the identified fault type and behavior, associate the fault information with the three-dimensional virtual model, and visualize the fault point, fault type and impact level at the corresponding position of the three-dimensional virtual model; according to the impact of the fault on the archives, identify the affected archive area on the three-dimensional virtual model.

[0048] In this embodiment, a fault detection system for a filing cabinet is also included, as shown in the attached Figure 2 As shown, a data acquisition module, a data processing module, a usage mode management module, a fault detection module and a visualization module are provided: A data acquisition module for recording the operation data of the filing cabinet using sensors; A data processing module, used for preprocessing and feature extraction of the file cabinet operation data; A usage mode management module is used to analyze the pre-processed file cabinet operation data, identify the current usage mode of the file cabinet, and manage and update the normal reference model corresponding to different usage modes; A fault detection module, for comparing the operation data of the filing cabinet with a normal reference model of the corresponding usage mode, identifying potential fault data, and analyzing the potential fault data to identify corresponding fault types and behaviors; The visualization module is used to provide users with a visual display of the filing cabinet's fault information and fault repair information.

[0049] The above-mentioned embodiments only express the preferred implementation modes of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person skilled in the art, several modifications, improvements and substitutions can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the attached claims.

Claims

1. A fault detection method for a filing cabinet, characterized in that: The following steps are included: S1: In the early stage of using the filing cabinet, use sensors to monitor and collect the operating data of the filing cabinet, wherein the operating data of the filing cabinet includes the daily operation data of the filing cabinet and the internal operating status data of the filing cabinet, and perform fault detection at the same time; S2: Analyze and identify normal operating data in the file cabinet operation data, and establish normal benchmark models corresponding to different usage modes; S3: Continuously monitor the operation data of the filing cabinet, identify the current usage mode, and compare the operation data of the filing cabinet with the normal reference model, identify data in the operation data of the filing cabinet that deviates from the normal operation mode, and mark the data as potential fault data; S4: Analyze the potential fault data, identify the fault type or behavior corresponding to the potential fault data, and determine the impact level on the files inside the filing cabinet; S5: After the fault is diagnosed, a detailed proactive maintenance suggestion plan is generated, and the fault information and proactive maintenance suggestion plan are pushed to the user and maintenance personnel; S6: Record all maintenance information, and use the maintenance information to optimize the judgment conditions in the fault detection method.

2. A fault detection method for a filing cabinet according to claim 1, characterized in that: The establishment of normal benchmark models corresponding to different usage modes includes: S21: performing cluster analysis on the daily usage operation data to identify different usage patterns, where the usage patterns include normal office use, high-frequency use, and low-frequency use; S22: identifying key characteristic parameters from the file cabinet operation data, where the key characteristic parameters are used to describe the normal operation characteristics of the file cabinet under different usage modes; S23: For different usage modes, use a machine learning algorithm to construct normal benchmark models corresponding to different usage modes, and store the normal benchmark models in a benchmark model library.

3. A fault detection method for a filing cabinet according to claim 2, characterized in that: The normal benchmark model is constructed using a support vector machine, and the specific calculation formula is: in, To achieve the overall goal of filing cabinet fault detection, Minimize to obtain the normal benchmark model corresponding to different usage modes; is the weight vector; is the bias term, representing the baseline of normal operating conditions; is a penalty parameter used to balance the sensitivity of the normal baseline model to abnormal situations; is the weight coefficient of different usage modes, which is used to dynamically adjust the importance of samples; It is a relaxation variable, which is used to detect small fluctuations in normal use of the filing cabinet without immediately judging the small fluctuations as potential faults.

4. A method for fault detection of a filing cabinet according to claim 3, characterized in that: The normal benchmark model also includes a dynamic update strategy: S211: After a preset time period ends or when the accumulated file cabinet operation data reaches a preset storage threshold, re-monitor and collect the latest file cabinet operation data; S212: using the same method as the current normal benchmark model training, preprocessing the collected file cabinet operation data to extract key feature parameters; and conduct statistical analysis to identify trends in usage patterns; S213: Compare the change trend with the current normal benchmark model to determine whether there is a significant deviation or change; S214: When there is a significant deviation or change, retrain the machine learning model using the latest file cabinet operation data and update the corresponding normal benchmark model; S215: Use an independent test set to perform performance evaluation on the updated normal baseline model. When the normal baseline model does not meet the performance standard, retrain the normal baseline model until it meets the performance evaluation standard.

5. A fault detection method for a filing cabinet according to claim 1, characterized in that: The comparing the filing cabinet operation data with the normal benchmark model includes: S31: continuously monitoring and collecting the operation data of the filing cabinet, and preprocessing the collected operation data of the filing cabinet; S32: based on the pre-processed file cabinet operation data, extract key feature parameters in real time within a set sliding time window; S33: using a pre-trained clustering model, performing pattern recognition on the key feature parameters in the current time window, and matching the pattern recognition result with a predefined usage pattern; S34: selecting a corresponding normal benchmark model from the benchmark model library according to the matching result, and comparing the file cabinet operation data in the current time window with the selected normal benchmark model; S35: If the degree of deviation between the filing cabinet operation data and the normal reference model exceeds a preset threshold, the deviated portion of the filing cabinet operation data is marked as potential fault data.

6. A method for fault detection of a filing cabinet according to claim 5, characterized in that: The setting of the sliding time window includes: S321: Predefine sliding time windows of different time lengths based on different usage patterns, including short-term, medium-term and long-term windows, wherein the short-term window is used to quickly detect frequent and sudden events, the medium-term window is used to detect changes in daily usage patterns, and the long-term window is used to capture trend changes or slowly developing faults over a long period of time; S322: Selecting a suitable sliding time window type according to the current usage mode; S323: For the selected sliding time window, dynamically adjust the window size according to the actual usage frequency and operation data characteristics of the filing cabinet: If the number of times the filing cabinet is accessed exceeds a preset threshold within the short-term window, the length of the sliding time window is shortened to capture anomalies more quickly; Within the limited time of the medium-term window, if the fluctuation range of the file cabinet usage data continues to be lower than the preset range, the length of the sliding window is expanded to observe longer trend data fluctuations; Within the limited time of the long-term window, if there is no abnormality in the previous operation data of the filing cabinet, the length of the sliding time window is expanded; S324: setting an initial anomaly detection threshold for each sliding time window, and evaluating the deviation of data within the sliding time window period from the normal benchmark in real time through mean square error or mean absolute error; and according to the evaluation result, when anomalies are continuously monitored within the sliding time window period and the degree of deviation exceeds a preset threshold, reducing the threshold value according to a preset adjustment amount; S325: Record all sliding time windows and threshold adjustment operation histories and test results, regularly evaluate the effectiveness of test results and window adjustments, and optimize and adjust the sliding time windows and threshold ranges. Evaluation indicators include detection accuracy, recall rate, and F1 score.

7. A fault detection method for a filing cabinet according to claim 1, characterized in that: The potential fault data is analyzed, specifically including: S41: segmenting the time series marked as the potential fault data, and extracting statistical features and frequency domain features of each segment; S42: Analyze the extracted statistical features and frequency domain features using a pre-trained multi-layer fault classification model to identify a specific type of fault or a specific pattern of abnormal behavior; S43: Based on the recognition result, combined with the structural information of the filing cabinet and the sensitivity level of the stored files, a quantitative assessment is made on the impact degree of the files inside the filing cabinet; S44: Based on the quantitative assessment results, the impact level is divided into three levels: low, medium and high, and the corresponding processing priority and corresponding strategy are defined for each level.

8. A fault detection method for a filing cabinet according to claim 7, characterized in that: The multi-layer fault classification model includes a primary classification layer, an advanced classification layer and a classification fusion layer; wherein the primary classification layer is used to identify simple patterns or abnormal behaviors in potential fault data by preliminarily classifying the filing cabinet operation data marked as potential fault data; the advanced classification layer is used to perform fine classification of the filing cabinet operation data marked as potential fault data that cannot be identified based on the output of the primary classification layer, and identify more specific fault types or behaviors; the classification fusion layer is used to determine the final fault type by combining the results of the primary classification layer and the advanced classification layer through weighted voting.

9. A fault detection method for a filing cabinet according to claim 1, characterized in that: The fault detection method for a filing cabinet also includes: establishing a three-dimensional virtual model of the filing cabinet, and visually displaying the fault point, fault type and impact level at the corresponding position of the three-dimensional virtual model.

10. A fault detection system for a filing cabinet, characterized in that: Included are: A data acquisition module for recording the operation data of the filing cabinet using sensors; A data processing module, used for preprocessing and feature extraction of the file cabinet operation data; A usage mode management module, used to analyze the pre-processed operation data of the filing cabinet and identify the current usage mode of the filing cabinet; and manage and update normal baseline models corresponding to different usage patterns; A fault detection module, for comparing the file cabinet operation data with the normal reference model corresponding to the usage mode, identifying potential fault data, and analyzing the potential fault data to identify corresponding fault types and behaviors; The visualization module is used to provide the user with a visualization display of the fault information and fault repair information of the filing cabinet.

Images