Acquisition meter reading abnormity identification method

Abstract

The invention relates to an acquisition and meter reading abnormity identification method, which comprises the following steps of: respectively establishing different abnormal condition models, dynamically configuring states of each terminal, namely a communication identification state, a meter reading identification state, a data identification state and a historical identification state, and dynamically setting a corresponding data acquisition strategy for each terminal. According to the method, meter reading abnormity or communication abnormity can be recognized in time, maximum utilization of data resources and processing resources is guaranteed, meanwhile, the whole system has high processing efficiency, meanwhile, the abnormity can be intelligently diagnosed, the operation and maintenance efficiency of a background system is improved, manual intervention analysis is not needed, and the accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of meter reading application, and in particular to a method for identifying abnormalities in meter reading. Background Art

[0002] With the continuous development of modern marketing, digital transformation has become the development direction of modern marketing, and data collection has become an important part of the process. The collected data not only supports the business development of the profession itself, but also has a growing impact on related professions. Data collection is affected by many factors such as communication channels, collection equipment, and the surrounding environment. At the same time, due to the large number of terminals and metering equipment that need to be operated and maintained, and the complex installation environment, the workload of abnormal collection operation and maintenance is heavy.

[0003] Currently, the diagnosis of collection anomalies is mainly carried out by exporting a list of yesterday's collection failures, manually analyzing the causes of the collection failures one by one, and conducting step-by-step analysis from the main station to the communication channel to the collection terminal and finally to the terminal meter. Ultimately, the cause of the collection failure is preliminarily located to provide guidance for on-site fault elimination.

[0004] In the process of diagnosing abnormal data collection, there is a large amount of data that needs to be processed and judged. The manual processing method has the problems of low efficiency and inaccurate positioning. This workload is large and complicated, and there is inaccurate positioning, which leads to the situation that the main station dispatches late and the on-site fault elimination is slow. For example, the publication number CN110609249A discloses a metering abnormality analysis and processing system based on the information collected by the electric energy meter, which is mainly based on the analysis of the information collected by the electric energy meter to obtain abnormal results, but it cannot verify the reliability of communication. For example, the publication number CN116774135B provides a remote meter reading abnormality monitoring method and system. Through abnormal analysis, each terminal can be evaluated for health, so as to determine whether there is an abnormality, but the health assessment method is not disclosed, so the logic and technical content of the abnormal analysis cannot be known. If the timely discovery and evaluation of the collection abnormality is to be realized, it is necessary to combine multi-dimensional data for analysis, and if the timeliness requirements are to be met at the same time, the restrictions on the amount of data and processing resources are large. The current abnormal recognition technology is difficult to achieve a relatively consistent balance between timeliness and accuracy. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a method for identifying meter reading anomalies, which has high processing efficiency and can perform intelligent diagnosis of anomalies, improve the operation and maintenance efficiency of the background system, and does not require manual intervention analysis, thereby improving accuracy.

[0006] To achieve the above object, the present invention adopts the following technical solution: a method for identifying abnormalities in meter reading, comprising the following steps:

[0007] Step A1: Obtain terminal collection feedback information corresponding to each terminal;

[0008] Step A2: Filter corresponding collection feedback data from the corresponding terminal collection feedback information according to the current state of the terminal. If the terminal state is a communication trigger state, obtain communication feedback data from the terminal feedback information. If the terminal state is a meter reading identification state, obtain meter reading feedback data from the terminal feedback information. If the terminal state is a data identification state, obtain local collection data from the terminal feedback information.

[0009] Step A3: configured with a communication identification strategy, a meter reading identification strategy, a data analysis strategy and a history association strategy, wherein the communication identification strategy obtains communication feedback data to calculate a corresponding communication abnormality value, and when the communication abnormality value is higher than a preset communication verification range, the terminal is given a meter reading identification state; the meter reading identification strategy obtains meter reading feedback data and calculates a corresponding meter reading abnormality value, and when the meter reading abnormality value is lower than a preset meter reading verification range, the meter reading identification state is deleted, and when the meter reading abnormality value is higher than the meter reading verification range, the meter reading identification state is switched to a data identification state; the data analysis strategy obtains local collection data and calculates a corresponding local abnormality value, and when the local abnormality value is lower than a preset local verification range, the data identification state is switched to a meter reading identification state, and when the meter reading abnormality value is higher than the meter reading verification range, the terminal is given a history identification state, and an associated data index is generated; the history association strategy obtains the terminal feedback information obtained by associating the data index to obtain a data index set, and retrieves historical data features from a historical feature database according to the data index set to calculate a historical abnormality value, and when the historical abnormality value is lower than a preset historical verification range, the historical identification state is deleted;

[0010] Step A4: Generate corresponding abnormal analysis sub-results according to the communication abnormal values, meter reading abnormal values, local abnormal values ​​and historical abnormal values ​​generated by the terminal, and bring the abnormal analysis sub-results into the preset abnormal recognition model to output the comprehensive abnormal result.

[0011] In a preferred embodiment: the step A2 is configured with a collection resource configuration strategy and a transmission resource configuration strategy, wherein the collection resource configuration strategy and the transmission resource configuration strategy respectively generate corresponding collection tasks and transmission tasks through the dynamic resource value of the terminal, wherein the dynamic resource value is generated through a preset dynamic mapping function, and Q d =f d (β c (s t -Δs t )), where Q d is the dynamic resource value of the terminal, f d (·) is the dynamic mapping function, β cis the state weight corresponding to the terminal. The state weight has different values ​​according to different terminal states. t is the abnormal value calculated at the last moment of the terminal, Δs t It is a preset benchmark abnormal value corresponding to the terminal state of the terminal.

[0012] In a preferred embodiment: the acquisition resource configuration strategy calculates the periodic acquisition resource value corresponding to each terminal, and determines the acquisition subtask and acquisition time in each acquisition cycle according to the periodic acquisition resource value, and generates the acquisition task according to the determined acquisition subtask and acquisition time; Among them, W d The resource value is collected periodically, r c is the amount of collected data corresponding to the terminal state of the current terminal, Q di The periodic resource collection value corresponding to the i-th terminal, r ci is the amount of collected data corresponding to the i-th terminal, and I is the total number of terminals;

[0013] The transmission resource configuration strategy assigns a corresponding relative priority value to each terminal according to the order of the size of the dynamic resource value. When the relative priority value is greater than the preset upload encapsulation benchmark, a data upload instruction is generated to store the corresponding terminal feedback information, and a data transmission instruction is generated according to the order of the relative priority value in the corresponding acquisition network topology group to configure the order of data transmission. According to the relative priority value, a corresponding data storage instruction is configured for each terminal to configure the space for data storage. The transmission task includes the upload instruction, the data transmission instruction and the data storage instruction.

[0014] In a preferred embodiment: the communication identification strategy is configured with a communication identification algorithm for calculating the communication anomaly value. Among them, s u1 is the communication abnormality value, m is the terminal online status, when the terminal is offline, m is 0, is the preset online status weight, is the preset comprehensive weight of collection, is the preset collection weight for the day. is the preset yesterday's collection weight, is the preset cumulative collection weight. When the terminal is online, m is the preset online value. t is the collection success rate of all electric meters in the area on day t, η t is the acquisition success rate threshold corresponding to day t; a t-1 is the data collection success rate of all meters in the area on day t-1, η t-1 is the acquisition success rate threshold on day t-1, and η t-1 >η t ; T is the cumulative offline duration, η TThe cumulative offline duration threshold.

[0015] In a preferred embodiment: the meter reading identification strategy is configured with a meter reading identification algorithm for calculating meter reading abnormal values,

[0016] Among them, s u2 is the abnormal value of meter reading. is the preset meter reading weight for the day, is the preset weight of yesterday's meter reading, b t is the number of failed data collections of all electric meters in the area on day t, ε t is the collection failure threshold corresponding to day t; b t-1 is the number of failed data collections of all meters in the area on day t-1, ε t-1 It is the threshold of the number of failed collections on day t-1.

[0017] In a preferred embodiment: the data analysis strategy includes an abnormal feature recognition network, the abnormal feature recognition network includes a number of abnormal feature nodes, abnormal feature links are arranged between the abnormal feature nodes, each abnormal feature node is configured with an identification condition, when the acquired local collection data meets the abnormal recognition condition, the corresponding abnormal feature node is activated, and the activated abnormal feature node forms a collection activation link; the data analysis strategy is configured with a data anomaly algorithm to calculate the corresponding local abnormal value,

[0018] Among them, s u3 is the local outlier, v i is the link abnormality weight corresponding to the i-th acquisition activation link, I1 is the total number of acquisition activation links, e ij is the feature similarity value of the jth abnormal feature node corresponding to the i-th acquisition activation link, χ ij is the node abnormal weight of the jth abnormal feature node corresponding to the ith acquisition activation link, and ri is the total number of abnormal feature nodes corresponding to the ith acquisition activation link.

[0019] In a preferred embodiment: the data analysis strategy is also configured with an abnormal index table, the abnormal index table stores a number of associated data indexes, the associated data indexes correspond to the collection activation link, and the corresponding associated data indexes are retrieved according to the collection activation link.

[0020] In a preferred embodiment: the historical correlation strategy is configured with a correlation anomaly algorithm for calculating historical anomaly values.

[0021] Among them, s u4 is a historical outlier, Q α (t) is the power collection function in the terminal feedback information, Q Δ(t) is the power collection function in the historical data characteristics, t n is the nth acquisition period corresponding to the index data set, and N1 is the total number of acquisition periods in the index data set.

[0022] In a preferred embodiment: Step A4 also includes constructing a conclusion association library, which stores a number of abnormal analysis sub-results, and the abnormal analysis sub-results are configured with corresponding trigger conditions. When the communication abnormal value, meter reading abnormal value, local abnormal value and historical abnormal value meet the corresponding trigger conditions, the corresponding abnormal analysis sub-result is output.

[0023] Compared with the prior art, the present invention has the following beneficial effects: through such a design, the present invention respectively establishes different abnormal situation models, dynamically sets corresponding data copying and sampling strategies for each terminal, ensures the maximum utilization of data resources and processing resources, and can satisfy that abnormal situations can be triggered in the first time, and make the whole system have higher processing efficiency, and can perform intelligent diagnosis on abnormalities, improve the operation and maintenance efficiency of the background system, and do not need manual intervention analysis, thereby improving accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A method flow chart of a preferred embodiment of the present invention;

[0025] Figure 2 This is a schematic diagram of the terminal state switching principle of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0026] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0027] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0028] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0029] Reference Figure 1-2 As shown, a method for identifying abnormalities in meter reading collection includes the following steps:

[0030] Step A1, obtain the terminal collection feedback information corresponding to each terminal; first, each terminal will upload the corresponding terminal collection feedback information, which is not only used for collection abnormality detection, but because the present invention configures different terminal states for different situations, the data is not uniformly packaged. First, the power data in the local collection data is packaged, and then packaged in combination with the voltage and current data, and then packaged in combination with the meter reading feedback data, and finally packaged with the collection feedback data, such as Figure 1 As shown in the figure, this encapsulation ensures data security while parsing data to different levels according to different states, thus improving data efficiency.

[0031] Step A2, according to the current state of the terminal, the corresponding collection feedback data is filtered from the corresponding terminal collection feedback information. If the terminal state is a communication trigger state, the communication feedback data is obtained from the terminal feedback information. If the terminal state is a meter reading identification state, the meter reading feedback data is obtained from the terminal feedback information. If the terminal state is a data identification state, the local collection data is obtained from the terminal feedback information. First, different data analysis methods are configured according to different terminal states, so that the minimum analysis resources can be invested in data analysis. At the same time, different terminal states for the terminal also reflect the security level that the terminal needs to be monitored. Through the formulation of this security level element, the accuracy and security will be improved. It can be divided into four levels in turn, including detecting whether the communication is unobstructed, detecting whether the collection instruction can be issued and fed back, detecting whether the data itself is abnormal, and detecting whether the user's use of data is abnormal. At the same time, the accuracy of data collection is also improved. The data of the previous level can provide data support for the next level, further improving reliability.

[0032] In step A2, a collection resource configuration strategy and a transmission resource configuration strategy are configured. The collection resource configuration strategy and the transmission resource configuration strategy generate corresponding collection tasks and transmission tasks respectively through the dynamic resource value of the terminal. The dynamic resource value is generated through a preset dynamic mapping function. d =f d (β c (s t -Δs t )), where Q d is the dynamic resource value of the terminal, f d () is the dynamic mapping function, β c is the state weight corresponding to the terminal. The state weight has different values ​​according to different terminal states. t is the abnormal value calculated at the last moment of the terminal, Δs tThe preset benchmark abnormal value corresponding to the terminal state of the terminal. The purpose of this step is to determine the resource allocation at a moment according to the corresponding abnormal value in each state. For example, if the abnormal value is high, then the resource allocation at the next moment is large, such as increasing the frequency of data collection. In this way, it is necessary to consider the collection resources and collection time, and allocate them in the following way: the collection resource allocation strategy calculates the periodic collection resource value corresponding to each terminal, and determines the collection subtask and collection time in each collection cycle according to the periodic collection resource value, and generates the collection task according to the determined collection subtask and collection time; Among them, W d Collect resource values ​​for the period, r c is the amount of collected data corresponding to the terminal state of the current terminal, Q di The periodic resource collection value corresponding to the i-th terminal, r ci is the amount of collected data corresponding to the ith terminal, and I is the total number of terminals; since the completion of a data collection may require multiple communication transmission actions, each transmission action is defined as a collection subtask, and each collection subtask is pre-configured to have a value. At the same time, the amount of data for each collection subtask can be estimated in advance. Executing the collection subtask will consume the collection cycle resource value, and the value corresponds to a decay factor, and the value will decrease according to the number of collections and other collection actions. In this way, the information with the highest cost performance can be selected for collection according to the order of value / data volume, and arranged in sequence, so that the corresponding collection time can be determined. For example, the five collection subtasks ABCDE have corresponding values ​​of 2 / 3 / 7 / 4 / 10, and the data volume is converted to 1 / 2 / 3 / 2 / 2. In this way, the first collection is E task, and after the collection is completed, the corresponding values ​​are 3 / 4 / 8 / 4 / 2 respectively. The values ​​of the three tasks ABC increase because the collection of task E is completed, and the value of task E decreases because it has been collected once. In this way, the second collection is task A. If other tasks change, then the third collection may be task C. If the value of task A remains unchanged, task A will continue to be collected. In this way, the allocation of collection subtasks is completed, and each task will consume the total collection cycle resource value. After the collection cycle resource value is consumed to 0, the collection action in the cycle is ended.

[0033] The transmission resource configuration strategy assigns a corresponding relative priority value to each terminal according to the order of the size of the dynamic resource value. When the relative priority value is greater than the preset upload encapsulation benchmark, a data upload instruction is generated to store the corresponding terminal feedback information, and a data transmission instruction is generated according to the order of the relative priority value in the corresponding acquisition network topology group to configure the order of data transmission. According to the relative priority value, a corresponding data storage instruction is configured for each terminal to configure the space for data storage. The transmission task includes the upload instruction, data transmission instruction, and data storage instruction. By setting the dynamic resource value, the uploaded resources and the stored resources are configured, so that important data can be uploaded and stored first, and can be directly called when needed without re-requesting.

[0034] Reference Figure 2 As shown, step A3 is configured with a communication identification strategy, a meter reading identification strategy, a data analysis strategy and a history association strategy.

[0035] The communication identification strategy obtains communication feedback data to calculate the corresponding communication anomaly value. When the communication anomaly value is higher than the preset communication verification range, the terminal is assigned a meter reading identification state. The communication verification range is a preset value, but there is a convergence factor. When the anomaly is not triggered, the communication verification range will slowly converge, so as to ensure that there will be a situation where the state needs to be changed. Since the communication anomaly has a relatively small detection burden, it is assigned during the meter reading identification state, that is, the terminal does not not use the communication identification state, but shares the communication identification state and the meter reading identification state.

[0036] The communication identification strategy is configured with a communication identification algorithm for calculating the communication anomaly value. Among them, s u1 is the communication abnormality value, m is the terminal online status, when the terminal is offline, m is 0, is the preset online status weight, is the preset comprehensive weight of collection, is the preset collection weight for the day. is the preset yesterday's collection weight, is the preset cumulative collection weight. When the terminal is online, m is the preset online value. t is the collection success rate of all electric meters in the area on day t, η t is the acquisition success rate threshold corresponding to day t; a t-1 is the data collection success rate of all meters in the area on day t-1, η t-1 is the acquisition success rate threshold on day t-1, and η t-1 >η t ; T is the cumulative offline duration, η T The cumulative offline duration threshold.

[0037] The abnormal value is calculated by calculating whether the acquisition success rate meets the preset conditions and whether there is a disconnection behavior. Preferably, η a is 95%, b is 100%, η T For 2 hours.

[0038] The meter reading identification strategy obtains meter reading feedback data and calculates the corresponding meter reading abnormal value. When the meter reading abnormal value is lower than the preset meter reading verification range, the meter reading identification state is deleted. When the meter reading abnormal value is higher than the meter reading verification range, the meter reading identification state is switched to the data identification state. Here, when the meter reading identification state is higher or lower than the meter reading verification range, it will be identified as an abnormal state, thereby triggering the switching or deletion action. The upper limit value of the meter reading verification range corresponding to the meter reading identification state has an attenuation factor. Over time, the upper limit value will gradually decrease to achieve timed switching to verify the data. There can only be one terminal state corresponding to the meter reading identification state and the data identification state.

[0039] The meter reading recognition strategy is configured with a meter reading recognition algorithm to calculate meter reading abnormal values.

[0040] Among them, s u2 is the abnormal value of meter reading. is the preset meter reading weight for the day, is the preset weight of yesterday's meter reading, b t is the number of failed data collections of all electric meters in the area on day t, ε t-1 b is the collection failure threshold corresponding to day t, which can be set to 5; b t-1 is the number of failed data collections of all meters in the area on day t-1, ε t-1 is the threshold of the number of failed data collection on day t-1, which can be set to 5. By calculating the meter reading abnormal value, if the number of failed data collection increases, the urgency of the analysis needs to be increased, otherwise it is based on the analysis of abnormal values.

[0041] The data analysis strategy obtains the locally collected data and calculates the corresponding local abnormal value. When the local abnormal value is lower than the preset local verification range, the data identification state is switched to the meter reading identification state. When the meter reading abnormal value is higher than the meter reading verification range, the terminal is given a historical identification state and an associated data index is generated. The local abnormal value reflects the abnormal situation of the data collected itself, such as too high data value, abnormal data fluctuation, and the voltage data and current data are often used as the standard for abnormal identification. The data analysis strategy includes an abnormal feature identification network, which includes a number of abnormal feature nodes. There are abnormal feature lines between the abnormal feature nodes. Each abnormal feature node is configured with an identification condition. When the acquired local collected data meets the abnormal identification condition, the corresponding abnormal feature node is activated, and the activated abnormal feature node forms a collection activation link. The data analysis strategy is configured with a data anomaly algorithm to calculate the corresponding local abnormal value.

[0042] Among them, s u3 is the local outlier, v i is the link abnormality weight corresponding to the i-th acquisition activation link, I1 is the total number of acquisition activation links, e ij is the feature similarity value of the jth abnormal feature node corresponding to the i-th acquisition activation link, χ ij is the node abnormal weight of the jth abnormal feature node corresponding to the ith acquisition activation link, and ri is the total number of abnormal feature nodes corresponding to the ith acquisition activation link. Specifically, feature matching is performed on voltage waveform anomalies and current waveform anomalies. For example, if there are 4 anomalies in a waveform, 4 abnormal feature nodes will be activated. If the 4 abnormal nodes can combine 2 groups of acquisition activation links in the abnormal feature recognition network, then I1 is 2, r1 is 2, and r2 is 3. In this way, the corresponding local abnormal value can be calculated. Each acquisition activation link is pre-configured with a corresponding link abnormal weight, and each abnormal feature node is pre-configured with a node abnormal weight. Then, the corresponding feature is calculated according to the similarity between the feature corresponding to the node and the matched waveform feature, and the local abnormal value can be calculated.

[0043] The data analysis strategy is also configured with an abnormal index table, which stores a number of associated data indexes, which correspond to the acquisition activation link, and the corresponding associated data index is retrieved according to the acquisition activation link. Another benefit of using the acquisition activation link is to determine the target waveform information of the data to be acquired next time.

[0044] The historical association strategy obtains the terminal feedback information obtained by associating the data index matching to obtain the data index set, and retrieves the historical data features from the historical feature database according to the data index set to calculate the historical abnormal value. When the historical abnormal value is lower than the preset historical verification range, the historical identification state is deleted. For example, the abnormal value reflects the abnormal situation of the historical data, which corresponds to the matching value of the data index obtained, and the difference between the user's electricity consumption and the past is judged, so as to judge the abnormality;

[0045] The historical correlation strategy is configured with a correlation anomaly algorithm to calculate historical anomaly values.

[0046] Among them, s u4 is a historical outlier, Q α (t) is the power collection function in the terminal feedback information, Q Δ (t) is the power collection function in the historical data characteristics, t n is the nth acquisition period corresponding to the index data set, and N1 is the total number of acquisition periods in the index data set.

[0047] Step A4, generate corresponding abnormal analysis sub-results according to the communication abnormal value, meter reading abnormal value, local abnormal value and historical abnormal value generated by the terminal, and bring the abnormal analysis sub-results into the preset abnormal recognition model to output the comprehensive abnormal result. Step A4 also includes constructing a conclusion association library, the conclusion association library stores a number of abnormal analysis sub-results, and the abnormal analysis sub-results are configured with corresponding trigger conditions. When the communication abnormal value, meter reading abnormal value, local abnormal value and historical abnormal value meet the corresponding trigger conditions, the corresponding abnormal analysis sub-result is output. The abnormal conditions of the four abnormal dimensions within a cycle are comprehensively analyzed through the conclusion association library to obtain a more accurate abnormal conclusion for the terminal, thereby improving the reliability of the results.

[0048] The anomaly recognition model includes a number of recognition trigger nodes corresponding to a terminal configuration, and the topological relationship of the recognition trigger node corresponds to the topological relationship configuration of the terminal. The anomaly recognition model marks the corresponding recognition trigger node according to the corresponding anomaly analysis sub-result, and generates a comparison constellation diagram according to the marking result, and compares the comparison constellation diagram with the constellation feature diagram in the pre-constructed constellation feature diagram library to determine the closest constellation feature diagram, thereby obtaining the comprehensive anomaly result corresponding to the constellation feature diagram. The constellation feature diagram is pre-constructed, and each constellation feature diagram will have a comprehensive anomaly result, so that the systemic anomaly can be judged according to the similarity of the constellation feature diagram.

[0049] Of course, the above are only typical examples of the present invention. In addition, the present invention may also have many other specific implementations. All technical solutions formed by equivalent replacement or equivalent transformation fall within the scope of protection required by the present invention.

Claims

1. A method for identifying abnormalities in meter reading, characterized in that: The following steps are involved: Step A1: Obtain terminal collection feedback information corresponding to each terminal; Step A2: Filter corresponding collection feedback data from the corresponding terminal collection feedback information according to the current state of the terminal. If the terminal state is a communication trigger state, obtain communication feedback data from the terminal feedback information. If the terminal state is a meter reading identification state, obtain meter reading feedback data from the terminal feedback information. If the terminal state is a data identification state, obtain local collection data from the terminal feedback information. Step A3: configured with a communication identification strategy, a meter reading identification strategy, a data analysis strategy and a history association strategy, wherein the communication identification strategy obtains communication feedback data to calculate a corresponding communication abnormality value, and when the communication abnormality value is higher than a preset communication verification range, the terminal is given a meter reading identification state; the meter reading identification strategy obtains meter reading feedback data and calculates a corresponding meter reading abnormality value, and when the meter reading abnormality value is lower than a preset meter reading verification range, the meter reading identification state is deleted, and when the meter reading abnormality value is higher than the meter reading verification range, the meter reading identification state is switched to a data identification state; the data analysis strategy obtains local collection data and calculates a corresponding local abnormality value, and when the local abnormality value is lower than a preset local verification range, the data identification state is switched to a meter reading identification state, and when the meter reading abnormality value is higher than the meter reading verification range, the terminal is given a history identification state, and an associated data index is generated; the history association strategy obtains the terminal feedback information obtained by associating the data index to obtain a data index set, and retrieves historical data features from a historical feature database according to the data index set to calculate a historical abnormality value, and when the historical abnormality value is lower than a preset historical verification range, the historical identification state is deleted; Step A4: Generate corresponding abnormal analysis sub-results according to the communication abnormal values, meter reading abnormal values, local abnormal values ​​and historical abnormal values ​​generated by the terminal, and bring the abnormal analysis sub-results into the preset abnormal recognition model to output the comprehensive abnormal result.

2. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: In step A2, a collection resource configuration strategy and a transmission resource configuration strategy are configured. The collection resource configuration strategy and the transmission resource configuration strategy generate corresponding collection tasks and transmission tasks respectively through the dynamic resource value of the terminal. The dynamic resource value is generated through a preset dynamic mapping function. d =f d (β c (s t -Δs t )), where Q d is the dynamic resource value of the terminal, f d (·) is the dynamic mapping function, β c is the state weight corresponding to the terminal. The state weight has different values ​​according to different terminal states. t is the abnormal value calculated at the last moment of the terminal, Δs t It is a preset benchmark abnormal value corresponding to the terminal state of the terminal.

3. The method for identifying abnormalities in meter reading according to claim 2, characterized in that: The acquisition resource configuration strategy calculates the periodic acquisition resource value corresponding to each terminal, determines the acquisition subtask and acquisition time in each acquisition cycle according to the periodic acquisition resource value, and generates the acquisition task according to the determined acquisition subtask and acquisition time; Among them, W d The resource value is collected periodically, r c is the amount of collected data corresponding to the terminal state of the current terminal, Q di The periodic resource collection value corresponding to the i-th terminal, r ci is the amount of collected data corresponding to the i-th terminal, and I is the total number of terminals; The transmission resource configuration strategy assigns a corresponding relative priority value to each terminal according to the order of the size of the dynamic resource value. When the relative priority value is greater than the preset upload encapsulation benchmark, a data upload instruction is generated to store the corresponding terminal feedback information, and a data transmission instruction is generated according to the order of the relative priority value in the corresponding acquisition network topology group to configure the order of data transmission. According to the relative priority value, a corresponding data storage instruction is configured for each terminal to configure the space for data storage. The transmission task includes the upload instruction, the data transmission instruction and the data storage instruction.

4. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: The communication identification strategy is configured with a communication identification algorithm for calculating the communication anomaly value. Among them, s u1 is the communication abnormality value, m is the terminal online status, when the terminal is offline, m is 0, is the preset online status weight, is the preset comprehensive weight of collection, is the preset collection weight for the day. is the preset yesterday's collection weight, is the preset cumulative collection weight. When the terminal is online, m is the preset online value. t is the collection success rate of all electric meters in the area on day t, η t is the acquisition success rate threshold corresponding to day t; a t-1 is the data collection success rate of all meters in the area on day t-1, η t-1 is the acquisition success rate threshold on day t-1, and η t-1 >η t ; T is the cumulative offline duration, η T The cumulative offline duration threshold.

5. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: The meter reading recognition strategy is configured with a meter reading recognition algorithm to calculate meter reading abnormal values. Among them, s u2 is the abnormal value of meter reading. is the preset meter reading weight for the day, is the preset weight of yesterday's meter reading, b t is the number of failed data collections of all electric meters in the area on day t, ε t is the collection failure threshold corresponding to day t; b t-1 is the number of failed data collections of all meters in the area on day t-1, ε t-1 It is the threshold of the number of failed collections on day t-1.

6. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: The data analysis strategy includes an abnormal feature recognition network, which includes a number of abnormal feature nodes, abnormal feature lines between the abnormal feature nodes, and each abnormal feature node is configured with an identification condition. When the acquired local collection data meets the abnormal recognition condition, the corresponding abnormal feature node is activated, and the activated abnormal feature node forms a collection activation link; the data analysis strategy is configured with a data anomaly algorithm to calculate the corresponding local abnormal value, Among them, s u3 is the local outlier, v i is the link abnormality weight corresponding to the i-th acquisition activation link, I1 is the total number of acquisition activation links, e ij is the feature similarity value of the jth abnormal feature node corresponding to the i-th acquisition activation link, χ ij is the node abnormal weight of the jth abnormal feature node corresponding to the ith acquisition activation link, and ri is the total number of abnormal feature nodes corresponding to the ith acquisition activation link.

7. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: The data analysis strategy is further configured with an abnormal index table, which stores a number of associated data indexes. The associated data indexes correspond to the collection activation links, and the corresponding associated data indexes are retrieved according to the collection activation links.

8. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: The historical correlation strategy is configured with a correlation anomaly algorithm to calculate historical anomaly values. Among them, s u4 is a historical outlier, Q α (t) is the power collection function in the terminal feedback information, Q Δ (t) is the power collection function in the historical data characteristics, t n is the nth acquisition period corresponding to the index data set, and N1 is the total number of acquisition periods in the index data set.

9. The method for identifying abnormalities in meter reading according to claim 1, characterized in that: Step A4 also includes constructing a conclusion association library, which stores a number of abnormal analysis sub-results. The abnormal analysis sub-results are configured with corresponding trigger conditions. When the communication abnormal value, meter reading abnormal value, local abnormal value and historical abnormal value meet the corresponding trigger conditions, the corresponding abnormal analysis sub-result is output.

Images