Method for real-time monitoring of operating state of metering box based on multi-sensor device

Abstract

This application relates to the field of data processing technology, specifically to a method for real-time monitoring of the operating status of a metering box based on multi-sensor equipment. The method includes: collecting parameter data and fault types of the metering box through multiple sensors; determining initial outliers based on parameter data from the same time on different historical days and surrounding times for each time; obtaining the regularity of change based on parameter similarity and fluctuations across different days; dividing adjacent parameter differences into subsequences and assigning similarity anomalies to them to obtain similar outliers, and then combining this with the regularity of change to obtain trend outliers; further combining the initial outliers to obtain fault outliers; determining the importance of fault judgment based on the proportion of different fault outliers; and then combining the fault outliers at each time moment to determine the fault value of the target fault at each time moment; thereby monitoring the operating status of the metering box. This application improves monitoring reliability.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a method for real-time monitoring of the operating status of a metering box based on a multi-sensor device. Background Technology

[0002] With the rapid development of my country's power system and the continuous advancement of smart grid construction, the scale of distribution networks continues to expand, and their structure is becoming increasingly complex. As a key link connecting high-voltage transmission systems and low-voltage power terminals, metering boxes play a crucial role in electricity metering, user electricity management, and distribution network operation monitoring. However, due to long-term exposure to the outdoor environment, load fluctuations, equipment aging, and human factors, metering boxes are prone to safety hazards such as overheating of connections, insulation deterioration, moisture intrusion, abnormal vibration, and even unauthorized opening. If these hazards are not detected and addressed in a timely manner, they can not only lead to inaccurate electricity metering and equipment damage, but may also cause electrical fires and power outages, seriously threatening the safe and stable operation of the power grid and the safety of users' lives and property.

[0003] Traditional metering box maintenance relies primarily on regular manual inspections and emergency repairs after malfunctions, which suffers from long inspection cycles, limited coverage, delayed response, and high labor costs. Especially in remote areas or complex terrain, manual inspections are difficult and inefficient, making it challenging to achieve comprehensive, real-time control over the metering box's operational status. Furthermore, existing monitoring methods are often limited to collecting single parameters (such as temperature), lacking multi-dimensional information fusion and analysis capabilities, and thus unable to accurately assess the overall health and potential risks of the equipment. Therefore, constructing a multi-source sensing network can achieve comprehensive and continuous monitoring of the metering box's internal environment and electrical parameters. This can also improve the intelligence level of the power distribution network and ensure power supply reliability. Summary of the Invention

[0004] To address the technical problem of inaccurate status monitoring, this application provides a method for real-time monitoring of the operating status of a metering box based on multi-sensor devices. The specific technical solution adopted is as follows:

[0005] This application proposes a method for real-time monitoring of the operating status of a metering box based on a multi-sensor device, which includes the following steps:

[0006] The metering box collects parameter data of different parameters through multiple sensors and obtains the fault type of the metering box.

[0007] For any type of parameter, anomalies in the parameter data at each moment are monitored based on the parameter data of the same moment with different numbers of days in history and the parameter data around each moment to obtain the initial abnormal value of each type of parameter at each moment; the change regularity of each type of parameter is obtained based on the similarity between the parameter data of all days in history and the fluctuation of the parameter data of all days.

[0008] For each type of parameter, the daily parameter data are clustered into difference subsequences after taking the difference between adjacent values; anomaly detection is performed on the similarity between each time step and the difference subsequences of the same historical time step to obtain similar outliers for each time step; based on the parameter change regularity and the similar outliers for each time step, the change trend outliers for each time step are obtained; the change trend outliers are combined with the initial outliers to obtain the fault outliers for each time step.

[0009] The fault judgment importance of each type of parameter is obtained by comparing the fault anomaly value of all time parameters for each type of fault with the fault anomaly value of all time parameters for all types of faults. The fault value of the target fault at each time is obtained by comparing the fault anomaly value at each time and the fault judgment importance of each type of parameter.

[0010] The operating status of the metering box is monitored by comparing the initial outlier value and the fault value at each time point with the response threshold.

[0011] In the above scheme, this application introduces the Local Anomaly Factor (LOF) algorithm to detect anomalies in multi-dimensional parameters, effectively distinguishing normal parameter fluctuations from real fault signals, and avoiding the false alarms and missed alarms caused by the fixed threshold in the traditional threshold method. Furthermore, it constructs parameters with regularity of parameter changes and uses a similarity algorithm to evaluate the similarity and stability of multi-day monitoring curves, accurately identifying periodic changes strongly correlated with user electricity usage habits or time, avoiding misjudging regular fluctuations as anomalies. A parameter importance analysis model is constructed, quantifying the sensitivity of different monitoring parameters to various faults based on historical fault data, providing a weighting basis for fault diagnosis. Finally, it comprehensively considers the degree of parameter anomaly and its importance in specific faults to achieve one-to-one matching and judgment for each type of fault. This provides reliable monitoring of the metering box status when parameter monitoring is abnormal, improving the accuracy of status monitoring.

[0012] In one embodiment, the parameters include temperature, humidity, current, active power, reactive power, and power; the fault types include poor contact, insulation degradation, and condensation inside the enclosure.

[0013] In one embodiment, the method for monitoring anomalies in the parameter data at each moment based on parameter data from the same moment across different historical days and the parameter data surrounding each moment to obtain the initial anomaly value for each type of parameter at each moment is as follows:

[0014] Each moment is designated as the target moment, and the moments in history that coincide with the target moment are designated as historical target moments. All historical target moments and the parameter data of the target moments are used as input to the LOF algorithm to obtain the first outlier of the target moment.

[0015] The parameter data of the time intervals around the target time and the parameter data of the target time are used as inputs to the LOF algorithm to obtain the second outlier at the target time;

[0016] The average of the first and second outliers is recorded as the initial outlier.

[0017] In one embodiment, the regularity of change is positively correlated with the similarity among all curves and negatively correlated with the stability of change; the curves are fitted from the parameter data of each day, and the stability of change is the variance of all parameter data of each day.

[0018] In one embodiment, the method for clustering and dividing all daily parameter data into difference subsequences after taking the difference between adjacent values ​​is as follows:

[0019] For each type of parameter, the parameter data of each day are subtracted to obtain a parameter difference sequence. All times in the parameter difference sequence are clustered, and the cluster distance is the Euclidean distance between the time and the parameter difference. Adjacent times in the same cluster are formed into a difference subsequence.

[0020] In one embodiment, the outliers in the trend are positively correlated with the regularity of change and similar outliers, respectively.

[0021] In one embodiment, the fault anomaly value is positively correlated with the trend anomaly value and the initial anomaly value, respectively.

[0022] In one embodiment, the method for obtaining the fault judgment importance of each type of parameter based on the ratio of the fault anomaly value of all time parameters for each type of fault to the fault anomaly value of all time parameters for all types of faults is as follows:

[0023] , This represents the mean of the fault anomaly values ​​of parameter I at all times corresponding to the target fault p. Let represent the mean of the fault anomaly values ​​of the i-th type of parameter at all times corresponding to the target fault p. Indicates the type of parameter. The parameter I represents the importance of the fault determination for the target fault p; the target fault can be any type of fault.

[0024] In one embodiment, the fault value of the target fault occurring at each time moment is positively correlated with the fault anomaly value of all parameters at each time moment and the fault judgment importance of each parameter to the target fault.

[0025] In one embodiment, the method for monitoring the operating status of the metering box by comparing the initial outlier value and the fault value at each time step with a response threshold is as follows:

[0026] Obtain all times for each type of fault in history, and obtain the fault value of each type of fault at each time. Use the minimum fault value corresponding to each type of fault as the fault threshold for each type of fault.

[0027] When the initial outlier value at the current moment is greater than the initial outlier threshold, the metering box exhibits an abnormal phenomenon.

[0028] When the initial abnormal value at the current moment is less than or equal to the initial abnormal value, the metering box is operating normally;

[0029] When an abnormality occurs in the meter box at the current moment, the fault value of each type of fault is obtained. If the fault value is greater than the fault threshold of each type of fault, the meter box has experienced that type of fault; otherwise, the fault has not occurred.

[0030] The beneficial effects of this application are as follows:

[0031] This application introduces the Local Anomaly Factor (LOF) algorithm to detect anomalies in multidimensional parameters, effectively distinguishing normal parameter fluctuations from genuine fault signals and avoiding false alarms and missed alarms caused by fixed thresholds in traditional threshold methods. Furthermore, it constructs parameters with regularity in parameter changes and uses a similarity algorithm to evaluate the similarity and stability of multi-day monitoring curves, accurately identifying periodic changes strongly correlated with user electricity usage habits or time, avoiding misjudging regular fluctuations as anomalies. A parameter importance analysis model is constructed, quantifying the sensitivity of different monitoring parameters to various faults based on historical fault data, providing a weighting basis for fault diagnosis. Finally, it comprehensively considers the degree of parameter anomaly and its importance in specific faults to achieve one-to-one matching and judgment for each type of fault. This provides reliable monitoring of the metering box status when parameter monitoring is abnormal, improving the accuracy of status monitoring. Attached Figure Description

[0032] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart of a method for real-time monitoring of the operating status of a metering box based on a multi-sensor device, provided in one embodiment of this application. Detailed Implementation

[0034] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the real-time monitoring method for the operating status of a metering box based on a multi-sensor device proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0036] Example of a method for real-time monitoring of metering box operating status based on multi-sensor devices:

[0037] The following description, in conjunction with the accompanying drawings, details the specific scheme of the real-time monitoring method for the operating status of a metering box based on a multi-sensor device provided in this application.

[0038] Please see Figure 1 The diagram illustrates a flowchart of a real-time monitoring method for the operating status of a metering box based on a multi-sensor device, according to an embodiment of this application. The method includes the following steps:

[0039] Step S001: Collect parameter data of the metering box and obtain its fault type through multiple sensors.

[0040] Data on various parameters of the metering box are collected using multiple different sensors. These parameters include temperature, humidity, current, active power, reactive power, and power. In this embodiment, the parameter data is collected at 5-minute intervals.

[0041] All faults in the metering box were categorized as follows: loose screws, oxidation and corrosion, and aging of connectors leading to increased contact resistance and exacerbated Joule heating, which may result in a significant increase in local or overall temperature and a decrease in current, were categorized as poor contact; long-term high temperature, humidity, dirt, and aging leading to decreased insulation performance, which may result in persistently high or sudden humidity, slight increase in local temperature (increased dielectric loss), leakage leading to increased current, and decrease in insulation resistance, were categorized as insulation degradation; and poor sealing and rain seepage leading to moisture accumulation and increased humidity were categorized as condensation inside the box. This method yielded all fault types for the metering box.

[0042] At this point, all fault types and parameter data of the metering box have been obtained.

[0043] Step S002: Determine initial outliers based on parameter data from the same time on different historical days and the surrounding times of each time; obtain the regularity of change based on parameter similarity and parameter fluctuation on different days.

[0044] In the process of monitoring the status of the metering box, it is first necessary to perform anomaly analysis on various parameters. Only when parameter anomalies occur can problems with the status of the metering box be caused. Therefore, this application performs anomaly analysis on the obtained parameters. The following is an analysis of any one of the parameters.

[0045] The time periods of a preset number of days are used to form the first time period. In this embodiment, the length of the first time period is 30 days. Within the first time period before the current time, the time periods that are the same as the current time are marked as the current historical time. For any parameter, the parameter data of the current historical time within the first time period and the parameter data of the current time are used as inputs to the LOF algorithm to obtain the first outlier value of the parameter at the current time.

[0046] Simultaneously, a preset time period is used to form a second time period, which in this embodiment is 1 hour in length. The parameter data of all times within the second time period before the current time and the parameter data of the current time are used as inputs to the LOF algorithm to obtain the second outlier value of the parameter at the current time.

[0047] The average of the first and second outliers is calculated and recorded as the initial outlier. If the initial outlier is greater than 1, it indicates an anomaly has occurred; if the initial outlier is less than or equal to 1, it indicates that the current metering box is operating normally. Further analysis is conducted when an anomaly occurs in the metering box at the current moment.

[0048] Considering that abnormal metering box parameters may indicate a malfunction in the metering box operation, and that a malfunction can lead to a series of abnormal parameters or abnormal trends in those parameters, it is necessary to analyze different malfunctions by combining different parameter data. By acquiring historical metering box status monitoring results and performing anomaly analysis on the monitoring data based on these results, the importance of each monitoring data point can be analyzed. Finally, based on the obtained importance results of each parameter data, a comprehensive judgment of the metering box status is made.

[0049] In order to better monitor the status of the metering box, it is necessary to perform anomaly analysis on the parameters. Considering that the parameters monitored by the metering box may be related to the user's usage habits, that is, time-related, the anomaly analysis can first be performed based on the time correlation of the acquired parameters, thereby increasing the reliability of the anomaly analysis of the monitoring parameters.

[0050] For any parameter, with a daily monitoring period, all parameter data for each day are fitted into a curve. For the first time period, the similarity between the curves of any two days is calculated. Then, the stability of change is obtained based on the fluctuation of the parameter data. If the curve similarity is greater, the stability of change is worse, which reflects the strong time correlation of the parameter.

[0051] Therefore, the regularity of change is obtained based on the similarity and stability of change among all curves. The regularity of change reflects the correlation with user habits.

[0052] The regularity of the changes is positively correlated with the similarity among all curves, and negatively correlated with the stability of all changes.

[0053] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship is determined by the actual application, and this application does not impose any special restrictions.

[0054] It should be noted that negative correlation means that when one variable increases, the other variable decreases accordingly, and the two variables change in opposite directions. When one variable changes from large to small or from small to large, the other variable also changes from small to large or from large to small. The specific relationship is determined by practical application, and this application does not impose any special restrictions.

[0055] Preferably, in this embodiment, for any curve, the variance value of the parameter data is calculated, and the variance value is used as the stability of the curve; the reciprocal of the DTW distance between two curves is used as the similarity between the two curves; within the first time period, all the stability values ​​are calculated and recorded as the mean stability value; the mean of all similarities is calculated and recorded as the similarity mean; the normalized result of the ratio of the similarity mean to the stability mean is used as the regularity of change.

[0056] The greater the similarity between curves and the smaller the curve's stability, the greater the similarity among all daily current curves even when the corresponding parameter varies significantly each day. This indicates a stronger regularity in the curve's changes and suggests a greater likelihood that the changes are related to user habits or time.

[0057] Thus, the variation patterns of each type of parameter have been obtained.

[0058] Step S003: Based on the similarity of parameters and parameter fluctuations over different days, obtain the regularity of change; divide the adjacent differences of parameters into subsequences, and obtain similar outlier values ​​by assigning them to anomalies of similarity, and then obtain trend outlier values ​​by combining the regularity of change; and finally obtain fault outlier values ​​by combining the initial outlier values.

[0059] For each type of parameter, its change sequence is obtained to analyze whether there are any anomalies in the trend of parameter data. Specifically, for the curve corresponding to each type of parameter each day, the parameter data of adjacent moments in the curve are subtracted to obtain the parameter difference sequence. All moments in the parameter difference sequence are clustered, and the cluster distance is the Euclidean distance between the moment and the parameter difference, where the moment and the parameter difference are normalized values. In this way, all moments are clustered, and adjacent moments in the same cluster are formed into a subsequence. This completes the division of the parameter difference sequence into several difference subsequences.

[0060] The difference subsequence at the current time is denoted as the target subsequence, and the difference subsequence at the corresponding time in the history is denoted as the historical subsequence. The anomalous properties of the current and historical subsequences are obtained by comparing their similarity. In this embodiment, the similarity between the target and historical subsequences is calculated by using the inverse normalized value of the DTW distance between them as the similarity score.

[0061] The similarity between the target subsequence and all historical subsequences was processed using the LOF algorithm to obtain the LOF value corresponding to each similarity. The mean of all LOF values ​​was used as the similarity outlier. If the similarity between the curve obtained at the current time and the curve obtained on other days is larger than the outlier, it indicates that the curve obtained under the current condition is more abnormal, that is, the trend of the obtained parameter curve is more abnormal.

[0062] Outliers in the parameter change trend are obtained based on the regularity of parameter changes and the corresponding similar outliers.

[0063] The outliers in the trend are positively correlated with the regularity of change and similar outliers, respectively.

[0064] Preferably, in this embodiment, the outlier of the change trend is the product of the regularity of the change and the similar outlier. The stronger the regularity of the change and the larger the corresponding similar outlier, the larger the outlier of the parameter's change trend.

[0065] The fault anomaly value of the parameter is obtained by combining the trend anomaly value with the initial anomaly value.

[0066] The fault anomalies are positively correlated with the trend anomalies and the initial anomalies, respectively.

[0067] Preferably, in this embodiment, the fault anomaly value is the product of the trend anomaly value and the initial anomaly value.

[0068] At this point, the fault and abnormal values ​​for each parameter have been obtained.

[0069] Step S004: Determine the importance of fault judgment based on the proportion of fault anomalies of different faults; then combine the fault anomalies at each time to determine the fault value at each time when the target fault occurs.

[0070] When making different fault judgments, different parameters have different sensitivities to faults. Therefore, when analyzing each fault individually, it is necessary to combine historical data to analyze the sensitivity of each parameter, and when analyzing faults in real time, the fault status should be judged comprehensively based on the parameter sensitivity and the corresponding parameter monitoring value.

[0071] Any type of fault is designated as the target fault. In the historical data, the time when the target fault occurred is recorded as the fault time. Using the method described above, the fault anomaly value corresponding to each parameter at each time point is obtained. In this embodiment, the historical data is taken from the first time period.

[0072] The fault judgment importance of each parameter is determined by the ratio of the fault anomaly value of each type of parameter at the target fault location to the fault anomaly value of all types of parameters at the target fault location.

[0073] Preferably, in this embodiment, the expression for the importance of fault determination is:

[0074] , This represents the mean of the fault anomaly values ​​of parameter I at all times corresponding to the target fault p. Let represent the mean of the fault anomaly values ​​of the i-th type of parameter at all times corresponding to the target fault p. Indicates the type of parameter. This indicates the importance of parameter I in the fault diagnosis of target fault p.

[0075] Among them, when a target fault p occurs in the history, if the proportion of the abnormal value of one of the parameters in the total abnormal value of all parameters is larger, it means that the parameter has higher importance when the target fault p occurs, and the more anomaly judgment can be made based on the parameter.

[0076] For each time moment, the fault value of the target fault at each time moment is obtained based on the fault anomaly values ​​of all parameters at each time moment and the fault judgment importance of each parameter to the target fault; the larger the fault value, the more likely the target fault is to occur at that time moment.

[0077] The fault value of the target fault at each time moment is positively correlated with the fault anomaly value of all parameters at each time moment and the fault judgment importance of each parameter to the target fault.

[0078] Preferably, in this embodiment, the expression for the fault value of the target fault occurring at each time moment is:

[0079] , This indicates the importance of the i-th type of parameter for the fault judgment of the target fault p. Let represent the fault anomaly value of the i-th type of parameter at each time step corresponding to the target fault p. Indicates the type of parameter. This represents the fault value at each time step where the target fault p occurs.

[0080] At this point, the fault value of the target fault p at each time step has been obtained.

[0081] Step S005: Monitor the operating status of the metering box based on the initial abnormal value and the fault value of each fault at each time.

[0082] Count all times when the target fault occurs, calculate the fault value of the target fault at that time, and take the minimum value as the fault threshold.

[0083] If the initial anomaly value at the current moment is greater than 1, it indicates an anomaly has occurred. If the initial anomaly value is less than or equal to 1, it indicates that the current metering box is operating normally. If an anomaly occurs, the fault value of the target fault at the current moment is calculated and compared with the fault threshold. If the fault value is greater than the fault threshold, it indicates that the fault has occurred at the current moment, and an alarm is triggered. Otherwise, the fault has not occurred.

[0084] All fault types are judged using the above method to achieve real-time monitoring of the metering box's operating status.

[0085] It should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0086] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for real-time monitoring of the operating status of a metering box based on multi-sensor equipment, characterized in that, The method includes the following steps: The metering box collects parameter data of different parameters through multiple sensors and obtains the fault type of the metering box. For any type of parameter, anomalies in the parameter data at each moment are monitored based on the parameter data of the same moment with different numbers of days in history and the parameter data around each moment to obtain the initial abnormal value of each type of parameter at each moment; the change regularity of each type of parameter is obtained based on the similarity between parameter data of all days in history and the fluctuation of parameter data of all days. For each type of parameter, the daily parameter data are clustered into difference subsequences after taking the difference between adjacent values; anomaly detection is performed on the similarity between each time step and the difference subsequences of the same historical time step to obtain similar outliers for each time step; based on the parameter change regularity and the similar outliers for each time step, the change trend outliers for each time step are obtained; the change trend outliers are combined with the initial outliers to obtain the fault outliers for each time step. The fault judgment importance of each type of parameter is obtained by comparing the fault anomaly value of all time parameters for each type of fault with the fault anomaly value of all time parameters for all types of faults. The fault value of the target fault at each time is obtained by comparing the fault anomaly value at each time and the fault judgment importance of each type of parameter. The operating status of the metering box is monitored by comparing the initial outlier value and the fault value at each time point with the response threshold. The method for monitoring anomalies in parameter data at each time point and obtaining the initial anomaly value for each type of parameter at each time point based on parameter data from the same time point at different historical days and the parameter data surrounding each time point is as follows: Each moment is recorded as the target moment, and the moments in history that are the same as the target moment are recorded as historical target moments; all historical target moments and the parameter data of the target moments are used as input to the LOF algorithm to obtain the first outlier of the target moment; The parameter data of the time intervals around the target time and the parameter data of the target time are used as inputs to the LOF algorithm to obtain the second outlier at the target time; The average of the first and second outliers is recorded as the initial outlier.

2. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The parameters include temperature, humidity, current, active power, reactive power, and power; the fault types include poor contact, insulation degradation, and condensation inside the enclosure.

3. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The regularity of change is positively correlated with the similarity among all curves and negatively correlated with the stability of change; the curves are fitted from the parameter data of each day, and the stability of change is the variance of all parameter data of each day.

4. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The method for clustering and dividing all daily parameter data into difference subsequences by taking the difference between adjacent values ​​is as follows: For each type of parameter, the parameter data of each day are subtracted to obtain a parameter difference sequence. All times in the parameter difference sequence are clustered, and the cluster distance is the Euclidean distance between the time and the parameter difference. Adjacent times in the same cluster are formed into a difference subsequence.

5. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The outliers in the trend are positively correlated with the regularity of change and similar outliers, respectively.

6. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The fault anomalies are positively correlated with the trend anomalies and the initial anomalies, respectively.

7. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The method for determining the importance of each type of parameter based on the ratio of the fault anomaly value of all time parameters for each type of fault to the fault anomaly value of all time parameters for all types of faults is as follows: , This represents the mean of the fault anomaly values ​​of parameter I at all times corresponding to the target fault p. Let represent the mean of the fault anomaly values ​​of the i-th type of parameter at all times corresponding to the target fault p. Indicates the type of parameter. This indicates the importance of parameter I for the fault diagnosis of target fault p; The target fault can be any type of fault.

8. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 7, characterized in that, The fault value of the target fault at each time moment is positively correlated with the fault anomaly value of all parameters at each time moment and the fault judgment importance of each parameter to the target fault.

9. The method for real-time monitoring of the operating status of a metering box based on a multi-sensor device as described in claim 1, characterized in that, The method for monitoring the operating status of the metering box by comparing the initial outlier value and the fault value at each time step with the response threshold is as follows: Obtain all times for each type of fault in history, and obtain the fault value of each type of fault at each time. Use the minimum fault value corresponding to each type of fault as the fault threshold for each type of fault. When the initial outlier value at the current moment is greater than the initial outlier threshold, the metering box exhibits an abnormal phenomenon. When the initial abnormal value at the current moment is less than or equal to the initial abnormal value, the metering box is operating normally; When an abnormality occurs in the meter box at the current moment, the fault value of each type of fault is obtained at the current moment. When the fault value is greater than the fault threshold of each type of fault, the meter box has experienced that type of fault. Conversely, no such fault occurs.

Images