Distributed concentrator state monitoring method and system

Abstract

The invention relates to the technical field of concentrator state monitoring, and particularly discloses a distributed concentrator state monitoring method and system, and the method comprises the steps: querying a sensor in a concentrator, and constructing a data matrix according to the position of the sensor; acquiring operation data of each sensor at each moment based on the data matrix; identifying the operation data, determining the operation state of each concentrator, and synchronously determining the relevancy between the concentrators; wherein the operation state is a label set; adjusting the data acquisition frequency of each concentrator according to the correlation degree; according to the method, the data of the concentrator is subjected to matrix processing, the data of each sensor in the concentrator within a period of time is converted into a three-dimensional matrix, then the convolution identification model of the image is expanded and applied to the three-dimensional matrix, and the matching target features are traversed in the three-dimensional matrix, so that various operation states are identified, and the accuracy of the operation states is improved. And the state evaluation content is greatly enriched.

Description

Technical Field

[0001] The present invention relates to the technical field of concentrator status monitoring, and in particular to a distributed concentrator status monitoring method and system. Background Art

[0002] A concentrator is a device or system used for data aggregation and management, and is widely used in communications, electricity, computer networks and other fields. Its main function is to collect data or signals from multiple sources, and then process, forward or manage them. In layman's terms, it is a smart meter data statistics device used to obtain data within the scene and upload it to the platform. Its working status is related to the data collection process of the scene and is extremely important. Therefore, its status monitoring is required.

[0003] Most of the existing status monitoring solutions adopt conventional warning monitoring solutions. For example, when certain parameters reach a certain threshold, an alarm message is generated. Although this status monitoring method has low cost, the evaluation breadth is very poor, that is, only a limited number of evaluations can be performed on the concentrator. In fact, the working state of the concentrator itself is extremely complex. How to refine the status monitoring process of the concentrator and enrich the status monitoring results is the technical problem that the technical solution of the present invention wants to solve. Summary of the invention

[0004] The object of the present invention is to provide a distributed concentrator status monitoring method and system to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A distributed concentrator status monitoring method and system, the method comprising: Query the sensors in the concentrator and build a data matrix based on the location of the sensors; Based on the data matrix, the operation data of each sensor at each moment is obtained; wherein the operation data is obtained in a sampling manner; Identify the operation data, determine the operation status of each concentrator, and simultaneously determine the correlation between the concentrators; wherein the operation status is a tag set; The data acquisition frequency of each concentrator is adjusted according to the correlation.

[0006] As a further solution of the present invention: the step of acquiring the operating data of each sensor at each moment based on the data matrix includes: Read the data acquisition frequency of each concentrator, and generate an acquisition instruction based on the data acquisition frequency; After generating the acquisition instruction, a sensor is randomly selected; Receive the sensor value with time tag uploaded by the selected sensor; Query the row and column positions corresponding to the sensor in the data matrix, and fill the sensor values ​​into the row and column positions; Perform data extension on the data matrix containing the sensor values ​​to obtain the data matrix at that moment, and count the data matrices of all concentrators at the same moment as the operation data at that moment; The process of data extension is to simulate the sensing values ​​of other row and column positions based on the row and column positions of the existing sensing values.

[0007] As a further solution of the present invention: after the acquisition instruction is generated, the step of randomly selecting a sensor includes: Query the historical data of each sensor, query the distance between the sensor and the selected sensor, and determine the selection probability based on the historical data and distance; Randomly select sensors based on the selection probability; The loop is executed until the total number of selected sensors reaches the preset number.

[0008] As a further solution of the present invention: the step of performing data extension on the data matrix containing the sensor values ​​to obtain the data matrix at the moment, and counting the data matrices of all concentrators at the same moment as the operation data at the moment includes: For any row and column position that is not filled with the sensing value in the data matrix, query the previous filled data at the row and column position as the first data, and simultaneously obtain the time tag of the previous filled data; Querying the row and column positions of the filled sensor values ​​within the preset radius, and querying the sensor value of the row and column positions of the filled sensor values ​​that is closest to the time tag in the time domain as the second data; Taking the filled sensing values ​​as the third data, and determining the predicted data at the row and column positions of the unfilled sensing values ​​according to the first data, the second data and the third data; The loop is executed until all the rows and columns that are not filled with sensor values ​​are inserted with predicted data to obtain the data matrix at that moment. The data matrices of all concentrators at the same moment are counted as the operating data at that moment. The calculation process of the predicted data is: ; In the formula, To predict data, Represents the first data, represents the second data, Indicates the third data.

[0009] As a further solution of the present invention: the steps of identifying the operation data, determining the operation status of each concentrator, and synchronously determining the correlation between the concentrators include: Read all the operation data within a preset time period and sort them in chronological order to obtain a three-dimensional matrix; Read the abnormal features in the preset feature table, traverse the three-dimensional matrix, and calculate the matching degree; When the matching degree reaches a preset matching degree threshold, the operation state corresponding to the abnormal feature is read; wherein the abnormal feature is a three-dimensional feature, and the feature table contains abnormal feature items and operation state items; The three-dimensional matrices of each concentrator in the same time period are compared and the similarity is calculated as the correlation between the concentrators.

[0010] As a further solution of the present invention: the step of adjusting the data acquisition frequency of each concentrator according to the correlation comprises: clustering the concentrators based on the correlation; For each type of concentrator, query the operating status of the concentrator; Determining the basic frequency of the same type of concentrators according to the operating status; Query the total number of concentrators of this type and determine the additional frequency range; For any concentrator in this type of concentrator, an additional frequency is randomly selected within the additional frequency range and combined with the basic frequency as the final data acquisition frequency.

[0011] The technical solution of the present invention also provides a distributed concentrator status monitoring system, the system comprising: A data matrix building module is used to query the sensors in the concentrator and build a data matrix based on the location of the sensors; An operation data acquisition module, used to acquire the operation data of each sensor at each moment based on the data matrix; wherein the operation data acquisition process is a sampling acquisition process; An operation status identification module, used to identify the operation data, determine the operation status of each concentrator, and synchronously determine the correlation between the concentrators; wherein the operation status is a tag set; The data acquisition adjustment module is used to adjust the data acquisition frequency of each concentrator according to the correlation.

[0012] As a further solution of the present invention: the operation data acquisition module includes: An instruction generating unit, used for reading the data acquisition frequency of each concentrator, and generating an acquisition instruction based on the data acquisition frequency; A sensor selection unit, used to randomly select a sensor after generating an acquisition instruction; A value receiving unit, used to receive the sensor value with a time tag uploaded by the selected sensor; A value filling unit is used to query the row and column positions corresponding to the sensor in the data matrix and fill the sensor values ​​into the row and column positions; The data extension unit is used to extend the data matrix containing the sensor values ​​to obtain the data matrix at the moment, and to count the data matrices of all concentrators at the same moment as the operation data at the moment; The process of data extension is to simulate the sensing values ​​of other row and column positions based on the row and column positions of the existing sensing values.

[0013] As a further solution of the present invention: the sensor selection unit includes: The probability determination subunit is used to query the historical data of each sensor, query the distance between the sensor and the selected sensor, and determine the selection probability based on the historical data and the distance; A selection execution subunit is used to randomly select a sensor based on a selection probability; The loop execution subunit is used for loop execution until the total number of selected sensors reaches a preset number.

[0014] As a further solution of the present invention: the data extension unit includes: A first query subunit is used to query the previous filled data at any row and column position that is not filled with the sensing value in the data matrix as the first data, and simultaneously obtain the time tag of the previous filled data; A second query subunit is used to query the row and column positions of the filled sensor values ​​within a preset radius, and query the sensor value of the row and column positions of the filled sensor values ​​that is closest to the time tag in the time domain as the second data; A prediction execution subunit, configured to use the filled sensor value as the third data, and determine the prediction data at the row and column position of the unfilled sensor value according to the first data, the second data and the third data; The output subunit is used for loop execution until all the row and column positions that are not filled with sensor values ​​are inserted with prediction data, so as to obtain the data matrix at that moment, and to count the data matrices of all concentrators at the same moment as the operation data at that moment; The calculation process of the predicted data is: ; In the formula, To predict data, Represents the first data, represents the second data, Indicates the third data.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention performs matrix processing on the data of the concentrator, converts the data of each sensor in the concentrator within a period of time into a three-dimensional matrix, and then extends the convolution recognition model of the image to be applied to the three-dimensional matrix, traverses and matches the target features in the three-dimensional matrix, and then identifies a variety of operating states, which greatly enriches the state evaluation content. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, 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 some embodiments of the present invention.

[0017] Figure 1 The overall flow chart of the distributed concentrator status monitoring method is shown.

[0018] Figure 2 Shown is a structural diagram of a distributed concentrator status monitoring system. DETAILED DESCRIPTION

[0019] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] Figure 1 The following is a general flow chart of a distributed concentrator status monitoring method and system. In an embodiment of the present invention, a distributed concentrator status monitoring method includes: Step S100: query the sensors in the concentrator and construct a data matrix according to the positions of the sensors; The main function of a concentrator is to collect data or signals from multiple sources, and then process, forward or manage them. It is also a device itself, which is equipped with a large number of sensors to obtain the parameters of the concentrator during operation, query the position of the sensor in the concentrator, and arrange them according to a preset direction. A two-dimensional data matrix can be constructed, and each row and column position in the data matrix corresponds to a sensor; specifically, regarding the correspondence between the row and column positions and the sensors in the data matrix, the sensors can be read in a preset direction, such as from the upper left end to the lower right end of the concentrator, and then to the right and downward in turn, and correspond to the row and column positions on the right and below in turn. It is worth mentioning that the row and column positions in the data matrix form a rectangle. In other words, the number of columns is the same for each row, but the distribution of sensors is not the same. Sensors in different rows may be different, which means that some row and column positions in the matrix do not have corresponding sensors. At this time, a preset default value is inserted at the row and column position to indicate that there is no sensor.

[0021] In addition, it should be noted that the concentrators in the present application are of the same model, have exactly the same structure, and have the same built-in sensors, so the corresponding data matrices are also the same data matrices.

[0022] Step S200: acquiring the operating data of each sensor at each moment based on the data matrix; wherein the operation data acquisition process is a sampling acquisition process; The execution subject of this method can obtain the monitoring data of each sensor. The monitoring data obtained by the sensor contains a time tag, which indicates the time when the data was obtained. The monitoring data of each sensor is obtained to obtain the operating data at each moment. The operating data can be understood as the monitoring data of all sensors at the same moment. The operation data acquisition process is a sampling acquisition process, that is, all the data extracted from some sensors are called operation data. Because there are many sensors, the sampling data acquisition process can greatly reduce energy consumption. In addition, the sampling process of different concentrators is not the same, which also guarantees the breadth of the data acquisition process to a certain extent, that is, there will be no data missing caused by obtaining only certain data.

[0023] Step S300: Identify the operation data, determine the operation status of each concentrator, and simultaneously determine the correlation between the concentrators; wherein the operation status is a tag set; By identifying the acquired operation data, the operation status of each concentrator can be determined; wherein, the operation data is the operation data at a certain moment. If the operation data at a certain moment is used as an independent variable to determine the operation status of the concentrator, there will be great randomness. Therefore, the process of identifying the operation data is generally to identify the operation data within a period of time, and then determine the operation status; at the same time, the present application will also compare the operation data to determine the correlation between each concentrator; in actual applications, different concentrators often have a correlation, they may be a series architecture, or be affected by a certain line at the same time. If only theoretical analysis is used, although the correlation of each concentrator can be determined, theoretical analysis is very difficult, especially it is difficult to consider comprehensive environmental factors. Therefore, the correlation determination process based on actual data in the present application is more in line with implementation and has higher accuracy.

[0024] In addition, the running status is a label set, and can also be set as a string set, which is used to represent the evaluation content corresponding to the running data.

[0025] Step S400: adjusting the data acquisition frequency of each concentrator according to the correlation; After calculating the correlation between the concentrators, the concentrators can be clustered into different categories according to the correlation. At this time, the clustering results can be used to adjust the data acquisition frequency of each concentrator, which affects step S200, which is used to determine how often to obtain operating data. Step S200 is applied to each concentrator. When executing step S200, each concentrator will perform an operation and obtain sensor data once.

[0026] It is worth mentioning that regarding the sensors in the concentrator, in fact, some intelligent modules built into the concentrator can also be used as "sensors" to obtain data from the concentrator and then evaluate its operating status. Therefore, the sensors in the above content should be more accurately described as monitors, which are electronic devices with specific data acquisition and data transmission functions.

[0027] Regarding step S200, the step of acquiring the operating data of each sensor at each moment based on the data matrix includes: Read the data acquisition frequency of each concentrator, and generate an acquisition instruction based on the data acquisition frequency; After generating the acquisition instruction, a sensor is randomly selected; Receive the sensor value with time tag uploaded by the selected sensor; Query the row and column positions corresponding to the sensor in the data matrix, and fill the sensor values ​​into the row and column positions; Perform data extension on the data matrix containing the sensor values ​​to obtain the data matrix at that moment, and count the data matrices of all concentrators at the same moment as the operation data at that moment; The process of data extension is to simulate the sensing values ​​of other row and column positions based on the row and column positions of the existing sensing values.

[0028] In an example of the technical solution of the present invention, for any concentrator, the data acquisition frequency of the concentrator is read, and an acquisition instruction is generated based on the data acquisition frequency; after the acquisition instruction is generated, a sensor is randomly selected from all sensors in the concentrator, and the sensor value containing the time tag uploaded by the selected sensor is received, and then the corresponding row and column position of the sensor in the data matrix is ​​queried, and the sensor value is filled into the row and column position; since the random selection process selects a part of the sensors, only a part of the data in the data matrix is ​​the actually acquired data. For the data that is not acquired, the mean at the corresponding row and column position can be directly used, or a prediction can be made. The above content adopts a prediction scheme, and the data matrix containing the sensor value is extended (based on the actually acquired sensor data, other unacquired data is predicted), and the data matrix at that moment is obtained, and the data matrix of all concentrators at the same moment is counted as the operation data at that moment.

[0029] It should be noted that in the process of counting the data matrices of all concentrators at the same time, since each concentrator has an independent data acquisition frequency, not all concentrators have data matrices at any time. To count the data matrices of all concentrators at the same time, generally, the data matrix obtained by a certain concentrator is received first, and then its time tag is queried to check whether there are other concentrators that have obtained data matrices with the same time tag; the same time tag is not the same in an absolute sense, as long as the time difference is less than a preset threshold, the time tags of the data matrices of the two concentrators are considered to be the same.

[0030] As a preferred embodiment of the technical solution of the present invention, after generating the acquisition instruction, the step of randomly selecting a sensor includes: Query the historical data of each sensor, query the distance between the sensor and the selected sensor, and determine the selection probability based on the historical data and distance; Randomly select sensors based on the selection probability; The loop is executed until the total number of selected sensors reaches the preset number.

[0031] In an example of the technical solution of the present invention, the sensor selection process in each concentrator is described, and the historical data of each sensor is queried, and the distance between the sensor and the selected sensor is queried. The former is a time condition, and the latter is a space condition. Combining these two queried parameters, a selection probability is jointly determined, and then a sensor is randomly selected based on the selection probability; the loop is executed until the total number of selected sensors reaches a preset number.

[0032] It is worth mentioning that regarding the preset number, it can also be random, or more precisely, it can be dynamic, querying the historical monitoring results of the concentrator, that is, the previous operating status, determining the risk level of the concentrator based on the operating status, and then adjusting the number. The higher the risk level, the more sensors need to be selected; in addition, the operating status has been limited to a tag set, and the staff will conduct a risk assessment on each tag in advance to determine the corresponding risk value. When determining the risk level, query all tags within a preset time period (such as a week), calculate the sum of the risk values ​​corresponding to each tag, and obtain the final risk value as the risk level.

[0033] As a preferred embodiment of the technical solution of the present invention, the step of performing data extension on the data matrix containing the sensor values ​​to obtain the data matrix at the moment, and counting the data matrices of all concentrators at the same moment as the operation data at the moment includes: For any row and column position that is not filled with the sensing value in the data matrix, query the previous filled data at the row and column position as the first data, and simultaneously obtain the time tag of the previous filled data; Querying the row and column positions of the filled sensor values ​​within the preset radius, and querying the sensor value of the row and column positions of the filled sensor values ​​that is closest to the time tag in the time domain as the second data; Taking the filled sensing values ​​as the third data, and determining the predicted data at the row and column positions of the unfilled sensing values ​​according to the first data, the second data and the third data; The loop is executed until all the rows and columns that are not filled with sensor values ​​are inserted with predicted data to obtain the data matrix at that moment. The data matrices of all concentrators at the same moment are counted as the operating data at that moment. The calculation process of the predicted data is: ; In the formula, To predict data, Represents the first data, represents the second data, Indicates the third data.

[0034] The above content specifically limits the prediction process. For any row and column position of an unfilled sensor value in the data matrix, the previous filled data at the row and column position is queried as the first data, and the time tag of the previous filled data is obtained at the same time. The meaning of this step is that, assuming that a prediction needs to be made at a certain position, its nearest data is queried as the first data; then, the row and column positions of the filled sensor values ​​within a preset radius are queried, and the sensor value closest to the time tag in the time domain of the row and column position of the filled sensor values ​​is queried as the second data. The meaning of this step is that, based on the time of the first data, the data closest to the time at the row and column positions of the nearby (nearby in space) filled data is queried as the second data; then, the filled sensor value is used as the third data, and the predicted data at the row and column position of the unfilled sensor value is determined based on the first data, the second data and the third data.

[0035] The above content is a scheme for predicting the row and column positions of any unfilled sensor values, which is executed in a loop until all row and column positions of all unfilled sensor values ​​have a predicted data. At this time, the data extension is completed and the data matrix at the current moment is obtained.

[0036] Finally, the data matrix of all concentrators at the same time is counted as the operation data at that time.

[0037] As a preferred embodiment of the technical solution of the present invention, the steps of identifying the operation data, determining the operation status of each concentrator, and synchronously determining the correlation between the concentrators include: Read all the operation data within a preset time period and sort them in chronological order to obtain a three-dimensional matrix; Read the abnormal features in the preset feature table, traverse the three-dimensional matrix, and calculate the matching degree; When the matching degree reaches a preset matching degree threshold, the operation state corresponding to the abnormal feature is read; wherein the abnormal feature is a three-dimensional feature, and the feature table contains abnormal feature items and operation state items; The three-dimensional matrices of each concentrator in the same time period are compared and the similarity is calculated as the correlation between the concentrators.

[0038] In an example of the technical solution of the present invention, a specific description is given of the data application process. All operating data within a preset time period is read and sorted in chronological order. At this time, the sensing values ​​of each sensor in the concentrator within a period of time constitute a three-dimensional matrix. Then, the abnormal features are read in the preset feature table, the three-dimensional matrix is ​​traversed, and the matching degree is calculated. This process is equivalent to the convolution recognition process of the image. The difference is that the image is two-dimensional, and the present application is three-dimensional. This is an innovative application of the prior art. The abnormal feature is a preset small three-dimensional matrix, such as 3*3*3 or 5*5*5. Of course, a larger size is also feasible. It indicates that under a certain operating state, the sensing values ​​of sensors within a range are what they are within a period of time; the traversal process is a comparison process, and the abnormal features are compared with small areas of the same size in the three-dimensional matrix. The comparison result is the matching degree, which indicates the similarity between the abnormal features and the corresponding small areas; when the matching degree is high enough, it means that the corresponding abnormal features appear in the three-dimensional matrix, and the operating state corresponding to the abnormal features can be read.

[0039] Furthermore, on the basis of taking the three-dimensional matrix as the feature of the concentrator, the three-dimensional matrices of each concentrator in the same time period are compared, and the similarity is calculated, which can be used as the correlation between the concentrators. The higher the correlation, the more similar the sensing values ​​of each sensor in a period of time, and the higher the correlation between the two.

[0040] It is worth mentioning that, considering the problem of negative correlation, we can first subtract the data matrices at adjacent moments, and then calculate the absolute value to obtain the changes at adjacent moments. Every two time points correspond to a change (a difference matrix). After statistically analyzing the changes, we get the change matrix, compare the change matrices, and calculate the similarity as the correlation. This method is more accurate. For example, if there are 10 data matrices in the three-dimensional matrix, then the calculated change matrix is ​​9 matrices. Each matrix is ​​the difference matrix of the corresponding two data matrices. Each value in the difference matrix is ​​processed with the absolute value to ensure that the difference is positive.

[0041] As a preferred embodiment of the technical solution of the present invention, the step of adjusting the data acquisition frequency of each concentrator according to the correlation comprises: clustering the concentrators based on the correlation; For each type of concentrator, query the operating status of the concentrator; Determining the basic frequency of the same type of concentrators according to the operating status; Query the total number of concentrators of this type and determine the additional frequency range; For any concentrator in this type of concentrator, an additional frequency is randomly selected within the additional frequency range and combined with the basic frequency as the final data acquisition frequency.

[0042] In an example of the technical solution of the present invention, the concentrators are clustered based on the correlation. The correlation and distance are a dual set of data. The higher the correlation, the smaller the "distance" in the clustering process. Therefore, most of the existing clustering schemes can be applied based on the correlation, such as the K-means clustering scheme, etc.; for each type of concentrator, the operating status of the concentrator of this type is queried. The operating status of concentrators of the same type are almost similar. The union of the operating status of concentrators of the same type is calculated, and the basic frequency can be further determined. A feasible way is to assign a risk value to each label in the operating status, accumulate all risk values, and determine the basic frequency according to the inverse ratio of the accumulated risk value.

[0043] On the basis of the above content, after determining the basic frequency of the same type of concentrators according to the operating status, an additional frequency range can be determined according to the total number of the same type of concentrators, and for any concentrator in this type of concentrator, an additional frequency is randomly selected within the additional frequency range and combined with the basic frequency as the final data acquisition frequency, thereby performing differentiated processing on the data acquisition frequency of the concentrator.

[0044] Among them, regarding the additional frequency range, it has two symbols, including positive and negative. The meanings of the two are different and are preset by the staff. If it is a positive sign, it means that the more concentrators of the same type, the more important their data is. Correspondingly, the data acquisition frequency of this type of concentrator will be higher; if it is a negative sign, it means that the more concentrators of the same type, the operating status of each concentrator can be roughly inferred from the concentrators of the same type. Correspondingly, the data acquisition frequency of each concentrator can be reduced, thereby simplifying the data acquisition process; generally, more negative signs are used, which means that if there are more concentrators of the same type, the data acquisition pressure of each concentrator can be smaller.

[0045] Figure 2 The structure diagram of the distributed concentrator status monitoring system is shown. In a preferred embodiment of the technical solution of the present invention, a distributed concentrator status monitoring system is also provided. The system 10 includes: A data matrix building module 11 is used to query the sensors in the concentrator and build a data matrix according to the positions of the sensors; An operation data acquisition module 12 is used to acquire the operation data of each sensor at each moment based on the data matrix; wherein the operation data acquisition process is a sampling acquisition process; The operation status identification module 13 is used to identify the operation data, determine the operation status of each concentrator, and simultaneously determine the correlation between the concentrators; wherein the operation status is a tag set; The data acquisition adjustment module 14 is used to adjust the data acquisition frequency of each concentrator according to the correlation.

[0046] Furthermore, the operation data acquisition module 12 includes: An instruction generating unit, used for reading the data acquisition frequency of each concentrator, and generating an acquisition instruction based on the data acquisition frequency; A sensor selection unit, used to randomly select a sensor after generating an acquisition instruction; A value receiving unit, used to receive the sensor value with a time tag uploaded by the selected sensor; A value filling unit is used to query the row and column positions corresponding to the sensor in the data matrix and fill the sensor values ​​into the row and column positions; The data extension unit is used to extend the data matrix containing the sensor values ​​to obtain the data matrix at the moment, and to count the data matrices of all concentrators at the same moment as the operation data at the moment; The process of data extension is to simulate the sensing values ​​of other row and column positions based on the row and column positions of the existing sensing values.

[0047] Specifically, the sensor selection unit includes: The probability determination subunit is used to query the historical data of each sensor, query the distance between the sensor and the selected sensor, and determine the selection probability based on the historical data and the distance; A selection execution subunit is used to randomly select a sensor based on a selection probability; The loop execution subunit is used for loop execution until the total number of selected sensors reaches a preset number.

[0048] Furthermore, the data extension unit includes: A first query subunit is used to query the previous filled data at any row and column position that is not filled with the sensing value in the data matrix as the first data, and simultaneously obtain the time tag of the previous filled data; A second query subunit is used to query the row and column positions of the filled sensor values ​​within a preset radius, and query the sensor value of the row and column positions of the filled sensor values ​​that is closest to the time tag in the time domain as the second data; A prediction execution subunit, configured to use the filled sensor value as the third data, and determine the prediction data at the row and column position of the unfilled sensor value according to the first data, the second data and the third data; The output subunit is used for loop execution until all the row and column positions that are not filled with sensor values ​​are inserted with prediction data, so as to obtain the data matrix at that moment, and to count the data matrices of all concentrators at the same moment as the operation data at that moment; The calculation process of the predicted data is: ; In the formula, To predict data, Represents the first data, represents the second data, Indicates the third data.

[0049] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A distributed concentrator status monitoring method, characterized in that: The method comprises: Query the sensors in the concentrator and build a data matrix based on the location of the sensors; Acquiring the operating data of each sensor at each moment based on the data matrix; wherein the acquisition process of the operating data is a sampling acquisition process; The operation data is identified, the operation status of each concentrator is determined, and the correlation between each concentrator is determined synchronously; including: reading all the operation data within a preset time period, and sorting them in chronological order to obtain a three-dimensional matrix; reading abnormal features in a preset feature table, traversing the three-dimensional matrix, and calculating the matching degree; when the matching degree reaches a preset matching degree threshold, reading the operation status corresponding to the abnormal feature; wherein the abnormal feature is a three-dimensional feature, and the feature table contains abnormal feature items and operation status items; comparing the three-dimensional matrices of each concentrator in the same time period, calculating the similarity as the correlation between each concentrator; wherein the operation status is a label set; The data acquisition frequency of each concentrator is adjusted according to the correlation.

2. The distributed concentrator status monitoring method according to claim 1, characterized in that: The step of acquiring the operating data of each sensor at each moment based on the data matrix comprises: Read the data acquisition frequency of each concentrator, and generate an acquisition instruction based on the data acquisition frequency; After generating the acquisition instruction, a sensor is randomly selected; Receive the sensor value with time tag uploaded by the selected sensor; Query the row and column positions corresponding to the sensor in the data matrix, and fill the sensor values ​​into the row and column positions; Perform data extension on the data matrix containing the sensor values ​​to obtain the data matrix at that moment, and count the data matrices of all concentrators at the same moment as the operation data at that moment; The process of data extension is to simulate the sensing values ​​of other row and column positions based on the row and column positions of the existing sensing values.

3. The distributed concentrator status monitoring method according to claim 2, characterized in that: After the acquisition instruction is generated, the step of randomly selecting a sensor includes: Query the historical data of each sensor, query the distance between the sensor and the selected sensor, and determine the selection probability based on the historical data and distance; Randomly select sensors based on the selection probability; The loop is executed until the total number of selected sensors reaches the preset number.

4. The distributed concentrator status monitoring method according to claim 3, characterized in that: The step of performing data extension on the data matrix containing the sensor values ​​to obtain the data matrix at the moment, and counting the data matrices of all concentrators at the same moment as the operation data at the moment includes: For any row and column position that is not filled with the sensing value in the data matrix, query the previous filled data at the row and column position as the first data, and simultaneously obtain the time tag of the previous filled data; Querying the row and column positions of the filled sensor values ​​within the preset radius, and querying the sensor value of the row and column positions of the filled sensor values ​​that is closest to the time tag in the time domain as the second data; Taking the filled sensing values ​​as the third data, and determining the predicted data at the row and column positions of the unfilled sensing values ​​according to the first data, the second data and the third data; The loop is executed until all the rows and columns that are not filled with sensor values ​​are inserted with predicted data to obtain the data matrix at that moment. The data matrices of all concentrators at the same moment are counted as the operating data at that moment. The calculation process of the predicted data is: ; In the formula, To predict data, Represents the first data, represents the second data, Indicates the third data.

5. The distributed concentrator status monitoring method according to claim 1, characterized in that: The step of adjusting the data acquisition frequency of each concentrator according to the correlation comprises: clustering the concentrators based on the correlation; For each type of concentrator, query the operating status of the concentrator; Determining the basic frequency of the same type of concentrators according to the operating status; Query the total number of concentrators of this type and determine the additional frequency range; For any concentrator in this type of concentrator, an additional frequency is randomly selected within the additional frequency range and combined with the basic frequency as the final data acquisition frequency.

6. A distributed concentrator status monitoring system, characterized in that: The system comprises: A data matrix building module is used to query the sensors in the concentrator and build a data matrix based on the location of the sensors; An operation data acquisition module, used to acquire the operation data of each sensor at each moment based on the data matrix; wherein the operation data acquisition process is a sampling acquisition process; The operation status identification module is used to identify the operation data, determine the operation status of each concentrator, and synchronously determine the correlation between each concentrator; including: reading all the operation data within a preset time period, and sorting them in chronological order to obtain a three-dimensional matrix; reading abnormal features in a preset feature table, traversing the three-dimensional matrix, and calculating the matching degree; when the matching degree reaches a preset matching degree threshold, reading the operation status corresponding to the abnormal feature; wherein the abnormal feature is a three-dimensional feature, and the feature table contains abnormal feature items and operation status items; comparing the three-dimensional matrices of each concentrator in the same time period, calculating the similarity as the correlation between each concentrator; wherein the operation status is a label set; The data acquisition adjustment module is used to adjust the data acquisition frequency of each concentrator according to the correlation.

7. The distributed concentrator status monitoring system according to claim 6, characterized in that: The operation data acquisition module comprises: An instruction generating unit, used for reading the data acquisition frequency of each concentrator, and generating an acquisition instruction based on the data acquisition frequency; A sensor selection unit, used to randomly select a sensor after generating an acquisition instruction; A value receiving unit, used to receive the sensor value with a time tag uploaded by the selected sensor; A value filling unit is used to query the row and column positions corresponding to the sensor in the data matrix and fill the sensor values ​​into the row and column positions; The data extension unit is used to extend the data matrix containing the sensor values ​​to obtain the data matrix at the moment, and to count the data matrices of all concentrators at the same moment as the operation data at the moment; The process of data extension is to simulate the sensing values ​​of other row and column positions based on the row and column positions of the existing sensing values.

8. The distributed concentrator status monitoring system according to claim 7, characterized in that: The sensor selection unit comprises: The probability determination subunit is used to query the historical data of each sensor, query the distance between the sensor and the selected sensor, and determine the selection probability based on the historical data and the distance; A selection execution subunit is used to randomly select a sensor based on a selection probability; The loop execution subunit is used for loop execution until the total number of selected sensors reaches a preset number.

9. The distributed concentrator status monitoring system according to claim 8, characterized in that: The data extension unit comprises: A first query subunit is used to query the previous filled data at any row and column position that is not filled with the sensing value in the data matrix as the first data, and simultaneously obtain the time tag of the previous filled data; A second query subunit is used to query the row and column positions of the filled sensor values ​​within a preset radius, and query the sensor value of the row and column positions of the filled sensor values ​​that is closest to the time tag in the time domain as the second data; A prediction execution subunit, configured to use the filled sensor value as the third data, and determine the prediction data at the row and column position of the unfilled sensor value according to the first data, the second data and the third data; The output subunit is used for loop execution until all the row and column positions that are not filled with sensor values ​​are inserted with predicted data, so as to obtain the data matrix at that moment, and to count the data matrices of all concentrators at the same moment as the operation data at that moment; The calculation process of the predicted data is: ; In the formula, To predict data, Represents the first data, represents the second data, Indicates the third data.

Images