Big data full life cycle state monitoring method and operation self-feedback electric energy meter

Abstract

The invention discloses a big data full life cycle state monitoring method and an operation self-feedback electric energy meter, and the method comprises the following steps: S1, generating an electric energy meter twinborn body set based on the operation data of the electric energy meter; s2, extracting a state parameter vector, a space coordinate and a layer attribute representation vector, and generating a nested state embedding tensor; s3, constructing a state reasoning tensor field; s4, calculating an abnormal trend cumulative score value, and marking a to-be-fed-back state; s5, generating a feedback induction action set in combination with the trend direction, the variation amplitude and the neighborhood state density; s6, the electric energy meter executes data reading cycle adjustment, communication frequency adjustment and state self-marking, and generates a feedback execution record; and S7, reconstructing a state reasoning tensor field, and forming a state monitoring and autonomous feedback closed loop. According to the invention, accurate monitoring and rapid feedback of the state of the electric energy meter are realized, and the system stability and the operation management intelligence level are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment state monitoring, and particularly relates to a big data full life cycle state monitoring method and a running self-feedback electric energy meter. BACKGROUND

[0002] Under the background of continuous deepening of digital management of power distribution terminals, as a key sensing device, the electric energy meter bears the important functions of energy consumption data reading and state monitoring. In the prior art, the electric energy meter generally adopts a fixed cycle reading and a unified frequency reporting strategy, which cannot dynamically adjust the sensing intensity and communication behavior according to the running state. These shortcomings result in low resource utilization efficiency of the device in application scenarios with different stability, lagging feedback response, and difficulty in meeting the fine management needs.

[0003] Now some methods combine GIS maps to build an electric energy meter spatial information model to realize geographic location visualization, but the collaborative modeling capability between spatial attributes and running states is not enough, and a trend reasoning mechanism based on spatial adjacency relationship and state parameter difference cannot be established, and the high-dimensional information fusion capability is also lacking. Most of the existing researches focus on static abnormality judgment of individual state of the electric energy meter, ignoring the interaction between adjacent electric energy meters in the state evolution process, which makes it difficult to accurately identify and warn the local state change trend.

[0004] In addition, the current methods generally lack an abnormal scoring mechanism based on the trend evolution rate. The current threshold comparison based method is difficult to deal with small disturbances and potential abnormal growth paths under complex working conditions, and the state judgment granularity is insufficient. The feedback control mechanism of the electric energy meter also has the problem of single structure, which cannot dynamically generate customized feedback strategies according to the device state change trend and neighborhood environment, and cannot realize intelligent adjustment and closed-loop control of the electric energy meter itself.

[0005] Although the existing digital twin method has made some progress in device modeling, its application in the field of electric energy meters is mostly limited to the static data mapping level, and lacks the ability to fuse state parameters and spatial features in modeling. It has not yet formed a dynamic reasoning and feedback path driven by state evolution trend. In view of the above problems, the present field needs a full life cycle monitoring method that integrates spatial location, state parameters and trend evolution modeling to build an electric energy meter with self-feedback capability, thereby improving the running sensing depth and response intelligence level.

[0006] Therefore, how to provide a big data full life cycle state monitoring method and a running self-feedback electric energy meter is a problem that needs to be solved by those skilled in the art. SUMMARY

[0007] One purpose of the present application is to provide a big data full life cycle state monitoring method and a running self-feedback electric energy meter, the present application uses position perception modeling, trend inference scoring and dynamic feedback generation mechanism, constructs a digital twin system supporting state closed loop update, has the advantages of fine perception, timely response and intelligent feedback.

[0008] A big data full life cycle state monitoring method according to an embodiment of the present application comprises the following steps:

[0009] S1, based on the data of each electric energy meter, a set of electric energy meter twins is constructed;

[0010] S2, the state parameter vector, spatial coordinates and layer attribute representation vector of each electric energy meter twin are extracted and input into a modeling structure for fusing state and spatial features, a position perception attention mechanism is introduced to encode the spatial coordinates, participate in feature fusion calculation, and generate a nested state embedding tensor;

[0011] S3, based on the nested state embedding tensor, the difference between the state parameter vectors of adjacent electric energy meter twins and the spatial distance, an output state inference tensor field is generated;

[0012] S4, the continuous state change rate of each electric energy meter twin is extracted from the state inference tensor field, and the abnormal trend cumulative score value is calculated, when the abnormal trend cumulative score value exceeds the preset threshold value, the electric energy meter twin is automatically marked as a feedback state;

[0013] S5, according to the trend direction, change amplitude and adjacent state parameter vector density of the electric energy meter twin in the state inference tensor field, a feedback induction action set is generated;

[0014] S6, the electric energy meter automatically executes data reading period adjustment, communication frequency adjustment and running state self-marking operation according to the feedback induction action set, records the feedback execution data, and generates a feedback execution record;

[0015] S7, the feedback execution record and the state parameter vector after execution are updated to the electric energy meter twin, the nested state embedding tensor is updated, the state inference tensor field is reconstructed, and the electric energy meter state monitoring and autonomous feedback closed loop are realized.

[0016] Optionally, the S1 specifically comprises: calculating a stability score value according to the running data volatility and communication stability of each electric energy meter in the latest continuous period, the stability score value being obtained by weighted calculation of voltage variance, current fluctuation rate, active power offset rate and communication interruption times, and dividing the electric energy meters into high stability level, medium stability level and low stability level according to the stability score value: the high stability level electric energy meter is the electric energy meter with the stability score value lower than a first threshold value, the sampling period being set to 6 hours; the medium stability level electric energy meter is the electric energy meter with the stability score value between the first threshold value and a second threshold value, the sampling period being set to 1 hour; and the low stability level electric energy meter is the electric energy meter with the stability score value higher than the second threshold value, the sampling period being set to 10 minutes, and the constructed electric energy meter twin comprises a state parameter vector, a spatial coordinate and a layer attribute representation vector.

[0017] Optionally, the S2 specifically comprises:

[0018] S21, extracting the state parameter vector, spatial coordinate, layer attribute representation vector and stability level identifier of each electric energy meter twin, the state parameter vector comprising voltage, current, active power, reactive power, power factor and sampling time identifier, the spatial coordinate being the longitude and latitude of the electric energy meter, the layer attribute representation vector comprising power supply path level, physical installation type and topology structure identifier, and the stability level identifier indicating that the electric energy meter belongs to high stability level, medium stability level or low stability level;

[0019] S22, performing location coding based on the spatial coordinate, generating a location coding vector by using an improved multi-scale sine-cosine function to respectively encode the longitude and latitude, and splicing the location coding vector and the layer attribute representation vector to form a spatial structure representation vector;

[0020] S23, splicing the state parameter vector, spatial structure representation vector and stability level identifier into a fusion input vector set, and inputting the fusion input vector set into a location-aware attention mechanism, the location-aware attention mechanism generating attention weights by calculating the spatial correlation between the fusion input vectors;

[0021] S24, performing weighted fusion on the fusion input vector set based on the attention weights to generate a nested state embedding tensor of each electric energy meter twin.

[0022] Optionally, the process of using the improved multi-scale sine-cosine function to respectively encode the longitude and latitude in the step S22 comprises:

[0023] S221, setting the longitude of each electric energy meter twin as x, the latitude as y, the coding dimension as d, the coding serial number of each dimension as i, and the perturbation parameter sequence corresponding to the i-th dimension as i the frequency perturbation parameter as φ i the phase perturbation parameter as ψ i, the tensor weighting coefficient corresponding to the stability level is ω s , the longitude encoding value e i and the combination of the latitude encoding value l i The nested expression is:

[0024]

[0025] Wherein, ω s is the weighting coefficient corresponding to the stability level of the electric energy meter, the high stability level is 1.0, the medium stability level is 1.3, and the low stability level is 1.7;

[0026] S222, the longitude encoding value e1, e2, …, e d and the latitude encoding value l1, l2, …, l d are spliced into a position encoding vector with a length of 2d.

[0027] Optionally, the position-aware attention mechanism in the step S23 specifically includes: the processing process of the position-aware attention mechanism includes: based on the state parameter vector and the position encoding vector, respectively performing normalization operation to constitute the state input sequence and the position input sequence, splitting the position input sequence according to the longitude and latitude dimensions to generate a sensitive guide sequence for controlling the attention distribution direction, combining the Euclidean distance of the spatial coordinates between the electric energy meter twins, the grouping difference of the stability level and the change amplitude of the state parameter vector, constructing a multi-factor guide matrix for measuring the degree of correlation as the position guide weight in the attention mechanism, in the attention calculation process, according to the sensitive guide sequence and the position guide weight, the weight distribution relationship between the state input sequences is jointly adjusted, and the state input sequences are dynamically weighted before fusion according to the weight adjustment coefficient corresponding to the high stability, medium stability and low stability level.

[0028] Optionally, the S4 specifically includes:

[0029] S41, extract the state parameter vector of each electric energy meter twin in the state reasoning tensor field at the continuous sampling time step, and construct the time sequence state trajectory;

[0030] S42, performing time difference operation on each state trajectory to obtain the state change rate between each time step, and marking the time step with a change rate greater than the fluctuation threshold corresponding to the stability level as an abnormal fluctuation point;

[0031] S43, for each electric energy meter twin, the number of continuous abnormal fluctuation points in a fixed sampling time window is counted, and the trend disturbance score is accumulated and generated in combination with the change rate value and the change direction consistency coefficient;

[0032] S44, performing a time window integration processing according to the trend disturbance score and the duration of the abnormal fluctuation point, to form an abnormal trend cumulative score value;

[0033] S45, comparing the abnormal trend cumulative score value with a threshold value of a preset stable level grading division, if the score value exceeds the upper threshold value corresponding to the stable level to which the electric energy meter twin belongs, the electric energy meter twin is automatically marked as a feedback required state.

[0034] Optionally, the S5 specifically comprises:

[0035] In the electric energy meter twin marked as the feedback required state, the state parameter vector of the electric energy meter twin in the state reasoning tensor field for three consecutive time steps is extracted, the state evolution trend vector is formed according to the change direction of the state parameter vector, the state parameter vector density distribution of the adjacent electric energy meter twin within the spatial neighborhood of the electric energy meter twin is calculated, the density gradient change in each direction is counted, the density direction vector is constructed, the trend consistency degree is measured by the cosine value between the state evolution trend vector and the density direction vector, the consistency degree, the average change amplitude and the density intensity are input as joint features, the rule mapping is performed, and the optimal feedback direction and the feedback action type of the electric energy meter twin in the current state are judged.

[0036] The strategy of the stable level feedback regulation comprises the following contents: when the stable level is a high level, the data reading cycle adjustment step is ten percent of the present cycle, the communication frequency adjustment factor is the basic frequency minus twenty seconds, and the state self-marking type is state retention; when the stable level is a medium level, the data reading cycle adjustment step is thirty percent of the present cycle, the communication frequency adjustment factor is the basic frequency minus ten seconds, and the state self-marking type is dynamic observation; when the stable level is a low level, the data reading cycle adjustment step is fifty percent of the present cycle, the communication frequency adjustment factor is not adjusted, and the state self-marking type is high-frequency monitoring.

[0037] An electric energy meter according to an embodiment of the application comprises the following modules:

[0038] A data reading module is configured to read original measurement data and construct an electric energy meter state parameter vector, and generate an electric energy meter twin set.

[0039] A spatial fusion module is configured to extract the state parameter vector, the spatial coordinates and the layer attribute representation vector, and perform fusion feature modeling.

[0040] A position coding module is configured to perform position perception coding on the spatial coordinates, and generate a nested state embedding tensor.

[0041] A state reasoning module is configured to calculate a state difference value and a spatial distance, and output a state reasoning tensor field.

[0042] an anomaly detection module configured to extract a continuous state change rate, calculate an anomaly trend cumulative score value, and mark a state requiring feedback;

[0043] a feedback generation module configured to generate a feedback inducing action set according to a trend direction, a change amplitude, and a density;

[0044] a self-feedback execution module configured to perform a sampling period adjustment, a communication frequency adjustment, and a state self-marking operation, and generate a feedback execution record;

[0045] a state updating module configured to update a state parameter vector and a nested state embedding tensor of an electric energy meter twin, and reconstruct a state reasoning tensor field.

[0046] The present application has the following beneficial effects:

[0047] The present application improves the accuracy and efficiency of electric energy meter state monitoring. By constructing a nested state embedding tensor and a state reasoning tensor field, the present application realizes joint modeling of electric energy meters in spatial correlation and state evolution dimensions, accurately extracts a continuous state change rate and performs trend identification, and improves the accuracy and response foresight of anomaly detection.

[0048] The present application realizes active feedback and dynamic regulation of electric energy meters. By joint analysis based on a trend direction, a change amplitude, and a neighborhood state density, the present application generates a differentiated feedback inducing action set, so that the electric energy meter has the ability to adaptively adjust a sampling period, a communication frequency, and a state self-marking, and completes a closed-loop control process from state identification to feedback execution.

[0049] The present application constructs a hierarchical perception mechanism to improve resource utilization efficiency. By dividing electric energy meter stability levels and setting differentiated data reading strategies, the present application realizes low-frequency perception of high-stability-level devices, medium-frequency perception of medium-stability-level devices, and high-frequency perception of low-stability-level devices, optimizes perception resource allocation, and reduces communication and calculation overhead. BRIEF DESCRIPTION OF DRAWINGS

[0050] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate the present application and explain the principles of the present application, and do not constitute a limitation of the present application. In the drawings:

[0051] Figure 1 a whole flowchart of a big data full life cycle state monitoring method and a running self-feedback electric energy meter proposed by the present application;

[0052] Figure 2 a feedback inducing action set generation and execution flow structure diagram of a big data full life cycle state monitoring method proposed by the present application. DETAILED DESCRIPTION

[0053] The application will now be described in further detail with reference to the drawings. These drawings are simplified schematic illustrations of the basic structure of the application and therefore only show what is relevant to the present application.

[0054] Reference Figure 1 and Figure 2 A big data full life cycle state monitoring method, comprising the following steps:

[0055] S1. Constructing an electricity meter twin set based on the data of each electricity meter;

[0056] S2. Extracting the state parameter vector, spatial coordinate and layer attribute representation vector of each electricity meter twin, inputting them into a modeling structure for fusing state and spatial features, introducing a position-aware attention mechanism, position encoding the spatial coordinates, participating in feature fusion calculation, and generating a nested state embedding tensor;

[0057] S3. Outputting a state reasoning tensor field based on the nested state embedding tensor, the difference between the state parameter vectors of adjacent electricity meter twins and the spatial distance, and the specific operation being: in the construction process of the state reasoning tensor field, first extract the current state expression based on the nested state embedding tensor of each electricity meter twin, then select the adjacent twin set and calculate the element-by-element difference between the state parameter vectors to represent the local state change; then calculate the spatial distance combining the spatial coordinate information between the electricity meter twins, introduce a Gaussian decay weight function to perform position-sensitive weighting processing on the state difference, enhance the influence of the neighbor change, and finally superimpose and map the weighted state difference in the spatial adjacency structure to form a state reasoning tensor field of the continuous state evolution trend;

[0058] S4. Extracting the continuous state change rate of each electricity meter twin from the state reasoning tensor field, calculating the abnormal trend cumulative score value, and when the abnormal trend cumulative score value exceeds the preset threshold, automatically marking the electricity meter twin as a state that needs feedback;

[0059] S5. Generating a feedback inducing action set according to the trend direction, change amplitude and adjacent state parameter vector density of the electricity meter twin that needs feedback in the state reasoning tensor field;

[0060] S6. The electricity meter automatically performs data reading period adjustment, communication frequency adjustment and running state self-marking operation according to the feedback inducing action set, records the feedback execution data, and generates a feedback execution record;

[0061] S7, update the feedback execution record and the state parameter vector after execution to the electric energy meter twin, update the nested state embedding tensor, reconstruct the state reasoning tensor field, realize the state monitoring and autonomous feedback closed loop of the electric energy meter, the specific content is: first, the feedback execution record generated by each electric energy meter after executing the feedback induced action and the updated state parameter vector are written into the corresponding electric energy meter twin in synchronization, forming a dynamically evolving state entity; then, based on the latest twin set, the state parameter vector, the spatial coordinate and the layer attribute representation vector are extracted again and input into the fusion modeling structure, and the nested state embedding tensor is updated through the position perception attention mechanism; on this basis, the state difference and the spatial distance between adjacent twins are combined to regenerate the state reasoning tensor field, realize the continuous monitoring and real-time adjustment of the overall state evolution trend, and thus form a closed loop control process of self updating, self perception and self feedback.

[0062] The application realizes the whole-process closed loop control of the electric energy meter from perception, judgment to feedback by constructing the electric energy meter twin set with the state parameter vector as the core, fusing the position perception attention modeling, constructing the state reasoning tensor field and the feedback induction mechanism. Compared with the traditional mode of only monitoring or reporting regularly, the application has the advantages of high state evolution monitoring accuracy, flexible feedback triggering mechanism and strong structure response timeliness, and can adapt to the remote intelligent management of multiple types of electric energy meters in different operating environments.

[0063] In the embodiment, S1 specifically includes: according to the running data volatility and communication stability of each electric energy meter in the latest continuous period, a stability score value is calculated, the stability score value is calculated by weighting the voltage variance, current fluctuation rate, active power offset rate and communication interruption times, and the electric energy meter is divided into high stability level, medium stability level and low stability level according to the stability score value: the high stability level electric energy meter is the electric energy meter with the stability score value lower than the first threshold value, the data reading period is set to 6 hours; the medium stability level electric energy meter is the electric energy meter with the stability score value between the first threshold value and the second threshold value, the data reading period is set to 1 hour; the low stability level electric energy meter is the electric energy meter with the stability score value higher than the second threshold value, the data reading period is set to 10 minutes, and the constructed electric energy meter twin includes the state parameter vector, the spatial coordinate and the layer attribute representation vector; the data reading level and the data reading period jointly constitute the state configuration parameter, which is embedded into the electric energy meter state parameter vector as an additional field, used in S2 to participate in the state feature fusion calculation, in S3 to dynamically control the calculation window of the state difference, in S4 to form a level sensitive early warning combined with the abnormal trend cumulative score, and in S6 to guide the execution frequency and feedback priority of the feedback induced action set.

[0064] The present application aims at the poor stability difference of electric energy meter state, and proposes a grading data reading strategy based on stability score, and constructs a multi-sampling frequency twin set of high stability, medium stability and low stability levels, which can improve data utilization efficiency, reduce redundant communication burden, ensure timely capture of key state changes, and has on-demand scheduling capability, enhances the refinement and flexibility of electric energy meter state monitoring, and solves the problems of extensive and lagging feedback response of the prior art.

[0065] In the embodiment, S2 specifically includes:

[0066] S21, extract the state parameter vector, spatial coordinates, layer attribute representation vector and stability level identifier of each electric energy meter twin, the state parameter vector includes voltage, current, active power, reactive power, power factor and sampling time identifier, the spatial coordinates are the longitude and latitude of the electric energy meter, the layer attribute representation vector includes the supply path level, the physical installation type and the topology structure identifier, and the stability level identifier indicates that the electric energy meter belongs to the high stability level, the medium stability level or the low stability level;

[0067] S22, perform position coding based on the spatial coordinates, encode the longitude and latitude respectively by using the improved multi-scale sine and cosine function, generate a position coding vector, and splice the position coding vector and the layer attribute representation vector to form a spatial structure representation vector;

[0068] S23, splice the state parameter vector, the spatial structure representation vector and the stability level identifier into a fusion input vector set, and input it into a position-aware attention mechanism, the position-aware attention mechanism generates attention weights by calculating the spatial correlation between the fusion input vectors, which is used to enhance the state linkage representation capability between the spatially adjacent electric energy meters, and dynamically adjust the feature fusion strength based on the stability level difference;

[0069] S24, weight the fusion input vector set based on the attention weights, and generate a nested state embedding tensor for each electric energy meter twin.

[0070] The present application introduces a bidirectional embedding mechanism of layer attribute representation and spatial position coding in the fusion modeling structure, combines a position-aware attention mechanism to enhance the coupling expression capability of state and spatial features, constructs a nested state embedding tensor as the core intermediate representation, compared with the existing isolated modeling method, this method can improve the spatial perception capability and state expression granularity, so that the subsequent state reasoning has better context adaptability.

[0071] In the embodiment, the process of encoding the longitude and latitude respectively by using the improved multi-scale sine and cosine function in step S22 includes:

[0072] S221, set the longitude of each electric energy meter twin as x, the latitude as y, the coding dimension as d, the coding serial number of each dimension as i, the disturbance parameter sequence corresponding to the i-th dimension as δ i , the frequency disturbance parameter as φ i , the phase disturbance parameter as ψ i , the tensor weighting coefficient corresponding to the stability level as ω s , the longitude coding value e i and the latitude coding value l i are calculated

[0073]

[0074] wherein ω s is the weighting coefficient corresponding to the stability level of the electric energy meter, the high stability level is 1.0, the medium stability level is 1.3, and the low stability level is 1.7

[0075] S222, the longitude coding values e1, e2, …, e d and the latitude coding values l1, l2, …, l d in all dimensions are spliced into a position coding vector with a length of 2d.

[0076] In the position coding formula constructed by the present application, the dimension disturbance parameter, the frequency disturbance parameter, the phase disturbance parameter and the tensor weighting coefficient corresponding to the stability level are introduced, the traditional sine-cosine coding mode is improved and optimized, the longitude and latitude under different scales are dynamically weighted, the modulation signal driven by the layer category is superimposed, the generated position coding vector has the characteristics of strong direction resolution, high layer correlation and fast local change response, compared with the existing fixed period coding mode, this coding method enhances the structure distinction degree of space representation and the fusion efficiency of state embedding, provides a more distinguishable and stable space structure representation for subsequent state-space coupling modeling, and can improve the spatial adaptability and state modeling robustness of the electric energy meter twin in multiple scenarios.

[0077] In this embodiment, the position-aware attention mechanism in step S23 specifically includes: the processing process of the position-aware attention mechanism includes: based on the state parameter vector and the position encoding vector, performing normalization operations to form a state input sequence and a position input sequence respectively, splitting the position input sequence according to the longitude and latitude dimensions, generating a sensitive guidance sequence for controlling the direction of attention distribution, combining the Euclidean distance of the spatial coordinates between the electricity meter twins, the grouping difference of the stability level and the change amplitude of the state parameter vector, constructing a multi-factor guidance matrix for measuring the degree of mutual correlation as the position guidance weight in the attention mechanism, in the attention calculation process, the weight distribution relationship between the state input sequences is jointly adjusted according to the sensitive guidance sequence and the position guidance weight, and the state input sequence is dynamically weighted before fusion according to the weight adjustment coefficients corresponding to the high stability, medium stability and low stability levels.

[0078] The position-aware attention mechanism designed in the present invention integrates a weighted strategy based on spatial position guidance, a stability level sensitivity adjustment factor and an attention gating layer control structure to generate a fused feature vector to enhance the accuracy of state reasoning. This mechanism can solve the technical difficulties of the existing attention mechanism, such as insufficient sensitivity to spatial distribution and poor response to data reading stability, and can realize the dynamic coupling expression of position and stability in state modeling, thereby improving the model's generalization ability for heterogeneous states.

[0079] In this embodiment, the S4 specifically includes:

[0080] S41. Extract the state parameter vector of each electric energy meter twin at continuous sampling time steps in the state inference tensor field and construct a time series state trajectory;

[0081] S42, performing a time difference operation on each state trajectory to obtain the state change rate between each time step, and marking the time step with a change rate greater than the fluctuation threshold corresponding to the stability level as an abnormal fluctuation point;

[0082] S43. For each electric energy meter twin, count the number of consecutive abnormal fluctuation points within a fixed sampling time window, combine the change rate value and the change direction consistency coefficient, and accumulate to generate a trend disturbance score;

[0083] S44. Perform time window integration processing based on the trend disturbance score and the duration of the abnormal fluctuation point to form an abnormal trend cumulative score value;

[0084] S45. Compare the accumulated score of the abnormal trend with the preset threshold of the stability level classification. If the score exceeds the upper threshold corresponding to the stability level of the current electricity meter twin, the electricity meter twin is automatically marked as requiring feedback.

[0085] The application constructs a state change rate time sequence, and realizes accurate marking of abnormal state trends of the electric energy meter based on dynamic trend extraction, abnormality accumulation score calculation and stability weighted evaluation.

[0086] In the embodiment, the S5 specifically includes:

[0087] In the electric energy meter twin of the state feedback marking type, the state parameter vector of the electric energy meter twin in the state reasoning tensor field within three continuous time steps is extracted, the state evolution trend vector is formed according to the change direction of the state parameter vector, the state parameter vector density distribution of the adjacent electric energy meter twin within the spatial neighborhood of the electric energy meter twin is calculated, the density gradient change in each direction is counted, the density direction vector is constructed, the trend consistency degree is measured by the cosine value between the state evolution trend vector and the density direction vector, the consistency degree, the average change amplitude and the density intensity are input as joint features, the rule mapping is performed, and the optimal feedback direction and the feedback action type of the electric energy meter twin in the current state are judged.

[0088] The strategy of the stable level feedback regulation includes the following contents: when the stable level is high, the data reading period adjustment step is ten percent of the present period, the communication frequency adjustment factor is the basic frequency minus twenty seconds, and the state self-marking type is state retention; when the stable level is medium, the data reading period adjustment step is thirty percent of the present period, the communication frequency adjustment factor is the basic frequency minus ten seconds, and the state self-marking type is dynamic observation; when the stable level is low, the data reading period adjustment step is fifty percent of the present period, the communication frequency adjustment factor is not adjusted, and the state self-marking type is high-frequency monitoring.

[0089] The application realizes the trend sensitivity, steady-state response and false alarm suppression in the abnormality identification process, can improve the accuracy of abnormality detection and the reliability of the triggering mechanism, and also enhances the state feedback practicability of the digital twin model.

[0090] An electric energy meter according to an embodiment of the application includes the following modules:

[0091] The data reading module is configured to read the original measurement data and construct the state parameter vector of the electric energy meter, and generate the electric energy meter twin set.

[0092] a spatial fusion module for extracting a state parameter vector, a spatial coordinate, and a layer attribute representation vector, and performing fusion feature modeling;

[0093] a position encoding module for position-aware encoding of the spatial coordinate to generate a nested state embedding tensor;

[0094] a state inference module for calculating a state difference value and a spatial distance, and outputting a state inference tensor field;

[0095] an anomaly detection module for extracting a continuous state change rate, calculating an anomaly trend cumulative score value, and marking a state requiring feedback;

[0096] a feedback generation module for generating a feedback inducing action set according to a trend direction, a change amplitude, and a density;

[0097] a self-feedback execution module for performing a sampling period adjustment, a communication frequency adjustment, and a state self-labeling operation, and generating a feedback execution record;

[0098] a state update module for updating a state parameter vector and a nested state embedding tensor of an electric energy meter twin, and reconstructing a state inference tensor field.

[0099] The electric energy meter structure proposed by the application includes a data reading unit, a position encoding unit, an embedding modeling unit, a trend identification unit, a feedback execution unit, and a state update unit, each unit function corresponds to the logic of the claim one by one, and constitutes a closed-loop feedback platform of software and hardware fusion. This structure effectively supports the execution of the whole process of digital twin, spatial modeling, and feedback control, can realize the modular implementation path from state monitoring to self-feedback, and can improve the deployability and engineering landing ability of the overall system.

[0100] Embodiment 1:

[0101] In order to verify the feasibility of the application in implementation, the application is applied to a large-scale intelligent electric energy meter state monitoring and operation feedback task deployed in a certain regional power distribution network. The region contains different types of residential buildings, office buildings and small industrial users, the total number of electric energy meters exceeds 6000, has a variety of operating states, communication frequencies and spatial distribution characteristics, and is a typical scene for testing the performance of the application.

[0102] In practical applications, first, the real-time running data of all electric energy meters is centrally read, and the data content includes active power, reactive power, voltage, current and device temperature parameters sampled every 15 minutes, and a corresponding state parameter vector is generated. The longitude and latitude coordinates and the layer attributes of the electric energy meter are obtained through GIS information, fused with the state parameter vector, and the electric energy meter twin set is constructed. Combined with the position perception attention mechanism and the spatial structure encoding strategy proposed in the application, the spatial coordinates are improved by multi-scale sine and cosine encoding, and the position encoding and layer attribute representation vector are spliced to form a spatial structure representation vector.

[0103] The spatial structure representation vector and the state parameter vector are jointly input into the nested state modeling structure to generate a multi-dimensional nested state embedding tensor. By comparing the state parameter difference and spatial distance relationship between adjacent electric energy meter twins, a state reasoning tensor field is constructed, from which the continuous state change rate is extracted, the abnormal trend cumulative score is calculated, and the electric energy meter twin with a score value exceeding the preset threshold is automatically identified as a state that needs to be fed back. Combined with the trend direction, change amplitude and adjacent state parameter density, a feedback induced action set is generated.

[0104] In the feedback induction process, the system performs differential operations according to the stability level. Electric energy meters with high stability level mainly perform periodic state self-marking; electric energy meters with medium stability level trigger sampling period compression and communication interval adjustment; and electric energy meters with low stability level perform continuous high-frequency communication within a certain time window and record state fluctuations in real time. All feedback execution results are recorded as feedback execution records, which are updated to the corresponding twin with new read state data, and the nested state tensor and the state reasoning tensor field are reconstructed to form a monitoring-feedback closed loop.

[0105] Through two months of continuous operation monitoring comparison, it is found that the application can effectively improve the accuracy and reaction speed of abnormal trend identification. The average delay of the traditional method to detect serious fluctuation state is 2.1 hours, while the average delay of the application is 17 minutes. The false positive rate in the traditional strategy is about 6.3%, and the false positive rate of the application scheme is reduced to 2.1% under the same data volume. In the feedback closed loop test, the average stability score of the state after feedback is improved from 0.78 to 0.94, indicating that the feedback strategy has good intervention and control ability in actual operation.

[0106] In the actual deployment area, the state recovery time of the electric energy meter after feedback is shortened by an average of 36%, and the stability score of some edge low stability level nodes is improved by more than 0.2 due to the introduction of the high-frequency self-feedback mechanism. At the same time, the response ability of the system to the spatially aggregated abnormal state is significantly improved, and the nested state embedding model can accurately identify the trend diffusion path in the local area, giving an early warning in the reasoning tensor field to ensure that the risk range is controllable. The following table is a summary of the state comparison data before and after the feedback of the electric energy meter:

[0107] Table 1: State stability score comparison table before and after power meter feedback

[0108]

[0109]

[0110] From the "state stability score comparison table before and after power meter feedback", it can be seen that the improvement effect of the state stability of the power meter before and after the feedback is remarkable. First, from the score improvement amplitude, all 7 power meters have increased in stability score after feedback, with an average score improvement of 0.17, among which the low stability level power meter has a particularly outstanding improvement amplitude; the score of C-3112 increases from 0.53 to 0.81, with an increase of 0.28; E-4578 increases from 0.49 to 0.78, with an increase of 0.29, showing that the present application has strong intervention ability in dealing with power meters with poor stability; the medium stability level devices B-2093, D-1548 and G-3982 increase by 0.16, 0.18 and 0.15 respectively, showing the effectiveness of the medium amplitude feedback strategy on state adjustment; the high stability level devices such as A-1035 and F-6671 increase by 0.04 and 0.05 respectively, with a relatively small improvement amplitude, but showing the ability of the present application in fine-tuning the state accuracy.

[0111] From the state recovery time shortening ratio, the average shortening ratio reaches 32.6%, among which the recovery effect of the low stability device is the most obvious, E-4578 shortens the recovery time by 52%, C-3112 shortens the recovery time by 48%, which shows that the high-frequency communication and state re-sampling strategy have fast response speed and compact feedback rhythm in dealing with unstable state. The medium stability device G-3982 shortens by 36%, B-2093 shortens by 34%, and D-1548 shortens by 31%, showing the balance ability of the medium frequency control mechanism in considering efficiency and resource consumption. High stability devices A-1035 and F-6671 shorten by 15% and 12% respectively, indicating that even in the case of good original state, small fluctuations can be adjusted and optimized through the state self-marking mechanism.

[0112] From the abnormal score decrease amplitude, the abnormal score of the low stability power meter decreases most significantly, E-4578 decreases by 0.29, C-3112 decreases by 0.28, which shows that the state fluctuation of high-risk devices is effectively inhibited; the decrease amplitudes of medium stability devices D-1548, B-2093 and G-3982 are 0.18, 0.16 and 0.15 respectively, and the score improvement is concentrated and stable; high stability devices A-1035 and F-6671 also achieve a decrease amplitude of 0.04 and 0.05 respectively, proving that the state sensing mechanism still has optimization value on high stability power meters.

[0113] Overall, the three types of stable level electric energy meters benefit from the combination strategy of nested state modeling, state reasoning tensor field and feedback induction mechanism proposed by the present application; the low stability level electric energy meter improves the operation stability and response sensitivity through high frequency communication and state resampling strategy; the medium stability level electric energy meter realizes stable state transition through sampling period compression and communication interval optimization; the high stability level electric energy meter maintains state accuracy through lightweight state self-labeling mechanism to avoid potential disturbance accumulation; the present application embodies good consistency and control ability in the aspects of hierarchical feedback strategy matching, feedback action execution effect and state adjustment persistence.

[0114] In summary, the feedback method based on the construction of state reasoning tensor field by embedding the nested state into the tensor and combined with the trend scoring mechanism has the characteristics of accurate identification, timely response and effective control, which can realize continuous state monitoring and closed-loop self-feedback control of large-scale electric energy meter twin bodies, and at the same time improve the operation stability and data reliability of the power perception terminal.

[0115] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art within the technical range disclosed by the present application, according to the technical scheme and the inventive concept of the present application, equivalent replacement or change, should be covered within the protection scope of the present application.

Claims

1. A method for monitoring the status of a large data life cycle, characterized in that: The steps include: S1. Based on the data of each electricity meter, a set of electricity meter twins is constructed; S2. Extract the state parameter vector, spatial coordinates, and layer attribute representation vector of each electricity meter twin, input them into the modeling structure for fusing state and spatial features, introduce the position-aware attention mechanism, perform position encoding on the spatial coordinates, participate in the feature fusion calculation, and generate a nested state embedding tensor; S3, outputting a state inference tensor field based on the nested state embedding tensor, the difference in state parameter vectors between adjacent electricity meter twins, and the spatial distance; S4. Extract the continuous state change rate of each electric energy meter twin from the state reasoning tensor field, calculate the abnormal trend cumulative score value, and automatically mark the electric energy meter twin as requiring feedback when the abnormal trend cumulative score value exceeds a preset threshold; S5. Generate a feedback-induced action set based on the trend direction, change amplitude, and adjacent state parameter vector density of the electric energy meter twin in the state requiring feedback in the state reasoning tensor field; S6. The electric energy meter automatically performs data reading cycle adjustment, communication frequency adjustment, and operation status self-marking operations according to the feedback induced action set, records feedback execution data, and generates feedback execution records; S7. Update the feedback execution record and the state parameter vector after execution to the electricity meter twin, update the nested state embedding tensor, reconstruct the state reasoning tensor field, and realize the electricity meter state monitoring and autonomous feedback closed loop.

2. A big data full life cycle status monitoring method according to claim 1, characterized in that: The S1 specifically includes: calculating a stability score value based on the operating data volatility and communication stability of each electric energy meter in the most recent continuous period, wherein the stability score value is obtained by weighted calculation of voltage variance, current fluctuation rate, active power offset rate and number of communication interruptions, and dividing the electric energy meter into high stability level, medium stability level and low stability level according to the stability score value: a high stability level electric energy meter is an electric energy meter with a stability score value lower than the first threshold, and the data reading cycle is set to 6 hours; a medium stability level electric energy meter is an electric energy meter with a stability score value between the first threshold and the second threshold, and the data reading cycle is set to 1 hour; a low stability level electric energy meter is an electric energy meter with a stability score value higher than the second threshold, and the data reading cycle is set to 10 minutes. The constructed electric energy meter twin includes a state parameter vector, spatial coordinates and layer attribute representation vector.

3. A big data full life cycle status monitoring method according to claim 2, characterized in that: The S2 specifically includes: S21. Extract the state parameter vector, spatial coordinates, layer attribute representation vector, and stability level identifier of each electric energy meter twin. The state parameter vector includes voltage, current, active power, reactive power, power factor, and sampling time identifier. The spatial coordinates are the longitude and latitude of the electric energy meter. The layer attribute representation vector includes the power supply path level, physical installation type, and topology structure identifier. The stability level identifier indicates whether the electric energy meter is of high stability level, medium stability level, or low stability level. S22. Perform position encoding based on spatial coordinates, use an improved multi-scale sine and cosine function to encode longitude and latitude respectively, generate a position encoding vector, and concatenate the position encoding vector with the layer attribute representation vector to form a spatial structure representation vector; S23, concatenating the state parameter vector, the spatial structure representation vector, and the stability level identifier into a fused input vector set, and inputting the fused input vector set into a position-aware attention mechanism, which generates an attention weight by calculating the spatial correlation between the fused input vectors; S24. Perform weighted fusion on the fusion input vector set based on the attention weight to generate a nested state embedding tensor for each electricity meter twin.

4. A big data full life cycle status monitoring method according to claim 3, characterized in that: The process of encoding the longitude and latitude using the improved multi-scale sine and cosine functions in step S22 includes: S221, let the longitude of each electric energy meter twin be x, the latitude be y, the coding dimension be d, the coding sequence number of each dimension be i, and the perturbation parameter sequence corresponding to the i-th dimension be δ i , the frequency perturbation parameter is φ i , the phase perturbation parameter is ψ i , the tensor weight coefficient corresponding to the stability level is ω s , calculate the longitude code value e i With the latitude code value l i The combined nested expressions are: Among them, ω s is the weighted coefficient corresponding to the stability level of the electricity meter, with a high stability level of 1.0, a medium stability level of 1.3, and a low stability level of 1.7; S222, longitude encoding values ​​e1, e2, ..., e on all dimensions d and latitude encoding values ​​l1,l2,…,l d Concatenate into a positional encoding vector of length 2d.

5. A big data full life cycle status monitoring method according to claim 4, characterized in that: The position-aware attention mechanism in step S23 specifically includes: the processing process of the position-aware attention mechanism includes: based on the state parameter vector and the position encoding vector, performing normalization operations respectively to form a state input sequence and a position input sequence, splitting the position input sequence according to the longitude and latitude dimensions, generating a sensitive guidance sequence for controlling the direction of attention distribution, combining the Euclidean distance of the spatial coordinates between the electricity meter twins, the grouping difference of the stability level and the change amplitude of the state parameter vector, constructing a multi-factor guidance matrix for measuring the degree of mutual correlation as the position guidance weight in the attention mechanism, in the attention calculation process, adjusting the weight distribution relationship between the state input sequences jointly according to the sensitive guidance sequence and the position guidance weight, and dynamically weighting the state input sequence before fusion according to the weight adjustment coefficients corresponding to the high stability, medium stability and low stability levels.

6. A big data full life cycle status monitoring method according to claim 5, characterized in that: The S4 specifically includes: S41. Extract the state parameter vector of each electric energy meter twin at continuous sampling time steps in the state inference tensor field and construct a time series state trajectory; S42, performing a time difference operation on each state trajectory to obtain the state change rate between each time step, and marking the time step with a change rate greater than the fluctuation threshold corresponding to the stability level as an abnormal fluctuation point; S43. For each electric energy meter twin, count the number of consecutive abnormal fluctuation points within a fixed sampling time window, combine the change rate value and the change direction consistency coefficient, and accumulate to generate a trend disturbance score; S44. Perform time window integration processing based on the trend disturbance score and the duration of the abnormal fluctuation point to form an abnormal trend cumulative score value; S45. Compare the accumulated score of the abnormal trend with the preset threshold of the stability level classification. If the score exceeds the upper threshold corresponding to the stability level of the current electricity meter twin, the electricity meter twin is automatically marked as requiring feedback.

7. A big data full life cycle status monitoring method according to claim 6, characterized in that: The S5 specifically includes: In the electric energy meter twin marked as requiring feedback, the state parameter vector of the electric energy meter twin in three consecutive time steps in the state reasoning tensor field is extracted, and the state evolution trend vector is formed according to the change direction of the state parameter vector. The density distribution of the state parameter vectors of the adjacent electric energy meter twins within the third order in the spatial neighborhood of the electric energy meter twin is calculated, and the density gradient changes in each direction are counted to construct the density direction vector. The cosine value of the angle between the state evolution trend vector and the density direction vector is used to measure the degree of trend consistency. The consistency degree, the mean value of the change amplitude and the density intensity are used as joint feature inputs for rule mapping to determine the optimal feedback direction and feedback action type of the electric energy meter twin in the current state. The strategies for feedback control based on the stability level include the following: when the stability level is high, the data reading cycle adjustment step is 10% of the current cycle, the communication frequency adjustment factor is the basic frequency minus 20 seconds, and the status self-marking type is status maintenance; when the stability level is medium, the data reading cycle adjustment step is 30% of the current cycle, the communication frequency adjustment factor is the basic frequency minus 10 seconds, and the status self-marking type is dynamic observation; when the stability level is low, the data reading cycle adjustment step is 50% of the current cycle, the communication frequency adjustment factor is not adjusted based on the basic frequency, and the status self-marking type is high-frequency monitoring.

8. An electric energy meter, used to implement the big data full life cycle status monitoring method according to any one of claims 1 to 7, characterized in that: The energy meter includes: The data reading module is used to read the original measurement data and construct the state parameter vector of the electric energy meter to generate the electric energy meter twin set; The spatial fusion module is used to extract state parameter vectors, spatial coordinates and layer attribute representation vectors, and perform fusion feature modeling; Positional encoding module, which performs position-aware encoding of spatial coordinates to generate nested state embedding tensors; The state reasoning module is used to calculate the state difference and spatial distance and output the state reasoning tensor field; Anomaly detection module, used to extract the continuous state change rate, calculate the cumulative score of abnormal trends and mark the state requiring feedback; Feedback generation module, used to generate feedback-induced action sets based on trend direction, change amplitude and density; Self-feedback execution module, used to perform sampling period adjustment, communication frequency adjustment and state self-marking operations, and generate feedback execution records; The state update module is used to update the state parameter vector and nested state embedding tensor of the electricity meter twin and reconstruct the state reasoning tensor field.

Images