Data storage method and system for multi-functional energy meters

Abstract

This invention discloses a data storage method and system for multi-functional energy meters, relating to the field of energy meter data storage technology. The method includes: collecting multi-source energy data during the operation of the energy meter and constructing recovery value parameters; uniformly labeling the multi-source energy data to form a data stream with recovery priority identifiers; performing semantic layering processing to establish semantic layering results for electricity consumption behavior; constructing a differentiated structural organization expression; reading voltage fluctuation data from the power supply side of the energy meter, constructing a power outage risk prediction index, and performing adaptive storage control based on the differentiated structural organization expression. This invention solves the technical problems of existing energy meter data lacking hierarchical storage, leading to the easy loss of critical data and difficulty in reconstructing and recovering abnormal data. It achieves the technical effect of complete retention and rapid recovery of critical data in power outage scenarios, improving the security of energy meter data storage and the reliability of abnormal operating condition data reconstruction.

Description

Technical Field

[0001] This invention relates to the field of electricity meter data storage technology, and more specifically to a data storage method and system for multi-functional electricity meters. Background Technology

[0002] Existing multi-functional energy meters generally adopt a unified and fixed storage strategy, without classifying the importance and semantics of energy data. They lack the ability to dynamically perceive voltage fluctuations and power outage risks on the power supply side, and cannot be adapted to the storage needs of different risk conditions. Conventional storage modes are prone to the loss of key metering and electricity consumption behavior data. After a power outage, it is difficult to quickly restore the device's operating status. Under abnormal conditions, data integrity is poor and reconstruction is difficult, making it difficult to meet the actual application needs of reliable data retention and fault tracing for energy meters.

[0003] Existing technology for electricity meters lacks hierarchical storage of data, making it easy to lose critical data and causing technical problems such as difficulty in reconstructing and recovering abnormal data. Summary of the Invention

[0004] This application provides a data storage method and system for multi-functional energy meters, which addresses the technical problem of existing energy meter data lacking hierarchical storage, making it easy to lose key data and causing abnormal data to be difficult to reconstruct and recover.

[0005] In view of the above problems, this application provides a data storage method and system for multi-functional energy meters.

[0006] A first aspect of this application provides a data storage method for a multi-functional energy meter, the method comprising:

[0007] Multi-source energy data is collected during the operation of the electricity meter. Based on the magnitude, rate of change, and duration of the state of the multi-source energy data over time, a recovery value parameter is constructed to characterize the contribution of the data to the recovery of the state after a power outage. According to the recovery value parameter, the multi-source energy data is uniformly labeled to form a data stream with a recovery priority identifier. Semantic layering processing is performed on the multi-source energy data based on the characteristics of electricity consumption behavior changes during the operation of the electricity meter to establish a semantic layering result of electricity consumption behavior. A differentiated structure organization expression is constructed using the semantic layering result of electricity consumption behavior and the recovery priority identifier. Voltage fluctuation data on the power supply side of the electricity meter is read, and a power outage risk prediction index is constructed based on the voltage fluctuation data. Adaptive storage control of the differentiated structure organization expression is performed using the power outage risk prediction index.

[0008] A second aspect of this application provides a data storage system for a multi-functional energy meter, the system comprising: The system comprises the following modules: a recovery value parameter construction module, used to collect multi-source energy data during the operation of the electricity meter, and construct recovery value parameters based on the magnitude, rate, and duration of change of the multi-source energy data over time series to characterize the contribution of the data to the recovery of the state after a power outage; a data stream formation module, used to uniformly label the multi-source energy data according to the recovery value parameters to form a data stream with recovery priority identifiers; a semantic layering processing module, used to perform semantic layering processing on the multi-source energy data based on the electricity consumption behavior change characteristics during the operation of the electricity meter to establish semantic layering results of electricity consumption behavior; a differentiated structure organization expression construction module, used to construct differentiated structure organization expressions using the semantic layering results of electricity consumption behavior and recovery priority identifiers; and an adaptive storage control module, used to read voltage fluctuation data from the power supply side of the electricity meter, construct a power outage risk prediction index based on the voltage fluctuation data, and perform adaptive storage control of the differentiated structure organization expression using the power outage risk prediction index.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: This method involves collecting multi-source electrical energy data during the operation of an electricity meter and constructing recovery value parameters to characterize the contribution of the data to the recovery of the state after a power outage. The multi-source electrical energy data is then uniformly labeled to form a data stream with recovery priority identifiers. Semantic layering processing is performed on the multi-source electrical energy data based on the electricity consumption behavior changes during the meter's operation, establishing semantic layering results for electricity consumption behavior. Differentiated structural organization and expression are constructed using the semantic layering results for electricity consumption behavior and the recovery priority identifiers. Voltage fluctuation data from the power supply side of the electricity meter is read, and a power outage risk prediction index is constructed based on this data. Adaptive storage control based on the differentiated structural organization and expression is then performed using this power outage risk prediction index. This method achieves the technical effect of complete retention and rapid recovery of key data in power outage scenarios, improving the data storage security of electricity meters and the reliability of data reconstruction under abnormal operating conditions. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic diagram of a data storage method for a multi-functional energy meter provided in an embodiment of this application; Figure 2 This is a schematic diagram of a data storage system structure for a multi-functional energy meter provided in an embodiment of this application.

[0012] Figure labeling: 10 for restoring value parameters, 20 for data flow formation, 30 for semantic layering processing, 40 for differential structure organization and expression, and 50 for adaptive storage control. Detailed Implementation

[0013] This application provides a data storage method and system for multi-functional energy meters, which addresses the technical problem of existing energy meter data lacking hierarchical storage, making it easy to lose key data and difficult to reconstruct and recover abnormal data.

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0015] Example 1, as Figure 1 As shown, this application provides a data storage method for a multi-functional energy meter, the method comprising: Step S100: Collect multi-source power data during the operation of the power meter, and based on the change amplitude, change rate and state duration of the multi-source power data in the time series, construct a recovery value parameter to characterize the contribution of the data to the state recovery after power failure.

[0016] Specifically, the system collects multi-source electrical energy data such as voltage, current, and power during the operation of the electricity meter in real time. The multi-source electrical energy data is divided into multiple continuous time segments using a sliding time window. Within each time segment, the local change gradient of voltage, current, and power data is calculated to form a first characteristic quantity characterizing the intensity of instantaneous fluctuations. Based on the difference of the first characteristic quantity between adjacent time segments, the system calculates the continuity index of the evolution of the electrical energy operating state. It identifies state abrupt change points and state stable intervals to form a second characteristic quantity characterizing the predictability of the state. The system performs duration statistics on the data corresponding to the state stable intervals and constructs a state stability contribution factor based on the frequency of state abrupt change points to form a third characteristic quantity characterizing the ability to maintain data stability. The first, second, and third characteristic quantities are weighted and coupled for calculation, and finally, a recovery value parameter is established to characterize the contribution intensity of the electrical energy state reconstruction after a power outage.

[0017] Step S200: Based on the recovery value parameters, perform unified marking processing on the multi-source power data to form a data stream with recovery priority identifier.

[0018] Specifically, based on the constructed recovery value parameter used to characterize the contribution intensity of state recovery after power failure, the multi-source power data is classified, sorted, and uniformly labeled according to the parameter value. Each group of power data is assigned a recovery priority identifier that matches its recovery value. The labeled data is then arranged in a regular manner according to the time series and priority order, ultimately forming a data stream with a continuous time sequence and clear priority, labeled with recovery priority identifiers.

[0019] Step S300: Perform semantic layering processing on the multi-source power data based on the characteristics of power consumption behavior changes during the operation of the power meter, and establish semantic layering results of power consumption behavior.

[0020] Specifically, the collected multi-source power data is time-aligned according to a unified time base to construct a synchronous correlation sequence between voltage, current, and power data. Based on the synchronous correlation sequence, the load matching relationship between power and current and the coupling relationship between voltage disturbance and power response are calculated to form behavioral correlation features characterizing the operating status of electrical equipment. Based on the temporal change patterns of the behavioral correlation features, three typical behavioral units are identified: load access, load exit, and stable operation. Behavioral unit sequences are reconstructed using typical behavioral units as the basic granularity. Contextual correlation analysis is performed on the behavioral unit sequences. Based on the temporal connection relationship and recurrence pattern, periodic power consumption patterns and aperiodic disturbance patterns are identified and semantic aggregation is performed to form a set of hierarchical behavioral segments. Based on the composition structure and continuous characteristics of each behavioral segment, it is assigned power consumption behavior semantic labels of continuous metering semantics, behavioral change semantics, and steady-state maintenance semantics, and finally, a complete hierarchical result of power consumption behavior semantics is established.

[0021] Step S400: Construct a differentiated structure organization expression using the semantic hierarchical results of electricity consumption behavior and the priority identifier.

[0022] Specifically, based on the established semantic hierarchical relationship between electricity consumption behavior and the combination of recovery priority identifiers, multi-source power data is structurally mapped to generate continuous metering data units, behavior change data units, and steady-state expression data units. The storage structure granularity of each data unit is determined according to the recovery priority identifier. For behavior change data units, continuous data segments corresponding to the change process are extracted to construct an event segment structure containing the change start point, change process, and termination point. The data retention granularity of the event segment structure is adjusted according to the recovery priority identifier. For steady-state expression data units, statistical parameters representing steady-state characteristics are extracted and parameterized. Different levels of encoding processing are performed in combination with the recovery priority identifier to form a compressed storage structure. Finally, a complete and differentiated structural organization expression is established based on the three types of data units and their corresponding structural granularities.

[0023] Step S500: Read the voltage fluctuation data on the power supply side of the electricity meter, construct a power outage risk prediction index based on the voltage fluctuation data, and use the power outage risk prediction index to perform adaptive storage control of differentiated structure organization expression.

[0024] Specifically, the system reads voltage fluctuation data from the power supply side of the electricity meter in real time, segments the voltage fluctuation data through a sliding time window, and extracts features such as voltage drop amplitude, fluctuation frequency, and duration of continuous low voltage within each time window to construct fluctuation feature quantities characterizing power supply stability. Based on the changing trend of fluctuation feature quantities between adjacent time windows, the system identifies the power supply state degradation process and establishes a trend factor. The fluctuation feature quantities and trend factor are fused and calculated to generate a power outage risk prediction index characterizing the probability of power outages within a preset time range. According to the relationship between the power outage risk prediction index and preset first and second thresholds, low-risk, medium-risk, and high-risk storage strategies are implemented in stages. For data units in the differentiated structure organization, adaptive processing such as delayed writing, high-compression storage, segmented writing, increased writing frequency, immediate solidification, and forced truncation closure are applied to complete the adaptive storage control of the differentiated structure organization.

[0025] In one possible implementation, step S100 further includes: Step S110: Divide the multi-source power data into sliding time windows to establish multiple continuous time segments.

[0026] Step S120: Calculate the local change gradient of voltage, current and power data in each continuous time segment to form the first characteristic quantity characterizing the intensity of instantaneous fluctuations.

[0027] Step S130: Based on the difference of the first characteristic quantity between adjacent time segments, calculate the evolution continuity index of the power operation state, and use the evolution continuity index to identify the state change point and the state stability interval, forming a second characteristic quantity characterizing the predictability of the state.

[0028] Step S140: Perform duration statistical processing on the data corresponding to the state stability interval, and construct a state stability contribution factor by combining the occurrence frequency of state change points to form a third feature quantity characterizing the ability of data to maintain stability.

[0029] Step S150: Perform weighted coupling calculations on the first feature quantity, the second feature quantity, and the third feature quantity to establish a recovery value parameter that characterizes the contribution intensity of power state reconstruction during the power outage recovery process.

[0030] Specifically, the collected multi-source power data such as voltage, current, and power are divided into sliding time windows along the time axis using a preset window length and step size. The continuous time series power data is sequentially divided into multiple continuous time segments that are interconnected, non-overlapping, or partially overlapping, so that the power data in each time segment maintains temporal integrity and local independence.

[0031] Within each continuous time segment obtained by the sliding time window, the time-series rate of change of voltage data, current data, and power data are solved respectively. The magnitude of the change of each quantity in the current segment and the rate of change per unit time are calculated to obtain the corresponding local change gradient. This gradient value is used as a quantitative index to characterize the instantaneous fluctuation intensity of the power data in the current time segment, and is uniformly formed into the first characteristic quantity.

[0032] The first characteristic quantity corresponding to two consecutive consecutive time segments is subjected to difference calculation and normalization. The absolute difference of the first characteristic quantity between adjacent segments is divided by the maximum value of the first characteristic quantity in that time period to obtain the evolution continuity index of the power operation state. When the evolution continuity index is greater than the preset mutation judgment threshold, the current time is determined to be the power operation state mutation point. When the evolution continuity index is less than or equal to the preset stability judgment threshold, the current time period is determined to be the state stability interval. Based on the distribution characteristics of the state mutation point and the state stability interval, a second characteristic quantity is constructed to characterize the predictability of the power operation state.

[0033] The duration of the electrical data segments corresponding to the identified stable intervals is statistically processed to obtain the actual duration of each stable interval. At the same time, the number of occurrences of state abrupt changes per unit time is counted to calculate the frequency of state abrupt changes. The duration of the stable interval and the frequency of state abrupt changes are weighted and fused together. The stable interval duration is divided by (1 + state abrupt change frequency) to construct the state stability contribution factor. This state stability contribution factor is directly used as the third characteristic quantity to characterize the data stability retention capability, reflecting the degree of contribution of electrical data to the stability retention during power outage recovery.

[0034] Based on the preset weight configuration, fixed weighting coefficients are assigned to the first feature quantity representing the intensity of instantaneous fluctuations, the second feature quantity representing the predictability of the state, and the third feature quantity representing the ability to maintain data stability. Each feature quantity is multiplied by its weighting coefficient in turn and then summed to complete the weighted coupling calculation. Finally, a recovery value parameter is generated to quantify the contribution intensity of power data in the process of power outage state recovery and power state reconstruction.

[0035] In one possible implementation, step S300 further includes: Step S310: Align the multi-source power data according to a unified time reference, construct a synchronous correlation sequence between voltage, current and power data, and calculate the load matching relationship between power data and current and the coupling relationship between voltage disturbance and power response based on the synchronous correlation sequence, forming behavioral correlation features characterizing the operating status of electrical equipment.

[0036] Step S320: Based on the change pattern of the behavioral correlation features in the time series, identify typical behavioral units in the process of power use. The typical behavioral units include load access behavioral units, load exit behavioral units, and stable operation behavioral units. Reconstruct continuous data with typical behavioral units as the basic granularity to form a sequence of behavioral units.

[0037] Step S330: Perform context association analysis on the behavior unit sequence, identify periodic power consumption patterns and aperiodic disturbance patterns based on the temporal connection relationship and recurrence patterns between adjacent behavior units, and perform semantic aggregation processing on the behavior unit sequence to form a set of behavior segments with hierarchical relationships.

[0038] Step S340: Based on the composition structure and continuous characteristics of each behavior segment in the behavior segment set, assign electricity consumption behavior semantic labels to the behavior segment set respectively. The electricity consumption behavior semantic labels include metering continuity semantics, behavior change semantics and steady-state maintenance semantics, and establish a hierarchical result of electricity consumption behavior semantics.

[0039] Specifically, the collected voltage, current, and power multi-source electrical energy data are time-aligned and interpolated using a unified timestamp as a reference, ensuring a one-to-one correspondence between the electrical quantity data at the same moment and constructing a synchronous correlation sequence with a consistent time axis. Based on the synchronous correlation sequence, the instantaneous power relationship P=U×I is used, where P represents instantaneous active power, U represents real-time operating voltage, and I represents real-time operating current. The ratio of power to current at the same moment, R=P / I, is calculated, where R is the load equivalent impedance characteristic parameter, thus obtaining the load matching relationship characterizing the load equivalent impedance. Voltage disturbances are extracted from the synchronous correlation sequence using a sliding window. In the dynamic segment, the rate of change of voltage relative to the rated value within the calculated window, ΔU / U, is used as the disturbance input, where ΔU represents the voltage change and U represents the rated reference voltage. The rate of change of power corresponding to the same window, ΔP / P, is used as the response output, where ΔP represents the power change and P represents the rated reference power. The least squares method is used for linear fitting to obtain the fitting slope Kc of the voltage change rate and the power change rate. Kc represents the correlation coefficient between voltage disturbance and power fluctuation. This slope is used as the coupling relationship between voltage disturbance and power response. The load matching relationship and the coupling relationship are integrated to form a complete behavioral correlation feature that characterizes the operating status of electrical equipment.

[0040] Based on the continuous trend, amplitude jump, and rate of change of the obtained behavioral correlation features on the time axis, behavioral pattern recognition is performed on the entire process of electricity use, and typical behavioral units in the electricity use process are extracted and divided. These typical behavioral units include load access behavioral units, load exit behavioral units, and stable operation behavioral units. When the behavioral correlation features rise rapidly and exceed the access judgment threshold, it is identified as a load access behavioral unit. When the behavioral correlation features fall rapidly and fall below the exit judgment threshold, it is identified as a load exit behavioral unit. When the behavioral correlation features remain within a stable range and there are no obvious jumps, it is identified as a stable operation behavioral unit. Using the above typical behavioral units as the smallest analysis and storage granularity, continuous electricity data is segmented, labeled, and reassembled, and the time-series continuous electricity data is reconstructed into a sequence of behavioral units arranged in chronological order.

[0041] Temporal context association analysis is performed on the formed behavioral unit sequences to traverse the sequential connection order, transition interval, and duration between adjacent behavioral units, extracting temporal connection relationships. At the same time, the recurrence frequency and occurrence period of each typical behavioral unit within a set time range are counted to obtain recurrence patterns. Based on temporal connection relationships and recurrence patterns, behavioral combinations with fixed periods and regular alternations are identified as periodic electricity consumption patterns, while abnormal behavioral combinations without fixed patterns and sudden occurrences are identified as aperiodic disturbance patterns. The behavioral unit sequences are hierarchically classified and semantically aggregated according to behavioral type, temporal association, and pattern attributes. Behavioral units of the same type, pattern, and context are merged into structured fragments, ultimately forming a set of behavioral fragments with a multi-level structure including behavioral units, behavioral fragments, and electricity consumption patterns.

[0042] Based on the unit composition structure, temporal continuity characteristics, and stable and continuous characteristics of each behavior segment in the behavior segment set, a corresponding electricity consumption behavior semantic label is assigned to each behavior segment. When a behavior segment exhibits continuous and uninterrupted characteristics suitable for metering and traceability, a metering continuity semantic label is assigned. When a behavior segment contains obvious state jumps and process changes such as load access and load exit, a behavior change semantic label is assigned. When a behavior segment exhibits long-term characteristics of stable operating parameters without significant fluctuations, a steady-state maintenance semantic label is assigned. All labeled behavior segments are classified and organized according to semantic type and temporal hierarchy, ultimately forming a complete and standardized semantic hierarchical result for electricity consumption behavior.

[0043] In one possible implementation, step S400 further includes: Step S410: Based on the semantic hierarchical results of electricity consumption behavior and the combination relationship of recovery priority identifier, perform structural mapping on multi-source power data to generate continuous metering data units, behavior change data units and steady-state expression data units, and determine the structural granularity of each data unit according to the recovery priority identifier.

[0044] Specifically, continuous data segments corresponding to the change process are extracted from the behavior change data unit, an event segment structure containing the change start point, change process and termination point is constructed, and the data retention granularity of the event segment structure is adjusted according to the recovery priority identifier.

[0045] Step S420: For the steady-state expression data unit, extract the statistical parameters that characterize the steady-state features, express them in a parameterized manner, and combine them with the recovery priority identifier to perform encoding processing of different precision levels to form a compressed storage structure.

[0046] Step S430: Establish differentiated structural organization expressions based on the continuous measurement data units, behavioral change data units, and steady-state expression data units, and their corresponding structural granularities.

[0047] Specifically, by combining the matching and combination relationship between the constructed semantic hierarchical results of electricity consumption behavior and the labeled recovery priority identifiers, the multi-source power data composed of voltage, current and power is subjected to full-domain structured mapping and decomposition. According to the semantic attributes of electricity consumption and the importance of data recovery, continuous metering data units, behavioral change data units and steady-state expression data units are generated respectively. At the same time, based on the recovery priority identifier level bound to different data units, the exclusive data sampling density, recording range and storage format of each data unit are set and matched to accurately determine the structural granularity corresponding to each type of data unit.

[0048] Specifically, for the divided behavioral change data units, the time-series continuous raw power data segments corresponding to the dynamic switching process of power consumption status are completely extracted. The starting time of the state switch is taken as the starting point of the change, the intermediate time interval of the state transition is taken as the change process, and the ending time when the power consumption status tends to stabilize is taken as the ending point, so as to completely construct a standardized event segment structure. At the same time, combined with the level difference of the recovery priority identifier bound to the data unit, the data sampling frequency, effective data storage quantity and detail recording accuracy within the event segment are dynamically adjusted to complete the adaptive differential adjustment of data retention granularity, ensuring that high-priority data is completely retained and low-priority data is appropriately simplified.

[0049] For the divided steady-state expression data units, statistical parameters that can objectively characterize the steady-state operation characteristics of electrical energy, such as average voltage, average current, average power, data fluctuation variance, and operating duration, are uniformly extracted. The extracted statistical parameters are used to replace the original time-series sampling data to complete the simplified parameterized expression. At the same time, combined with the recovery priority identifier level matched to the steady-state expression data unit, differentiated encoding algorithms and data compression rules of high precision, medium precision, and low precision are applied in stages. The parameterized steady-state data is then formatted and compressed with corresponding precision, and finally a compressed storage structure with differentiated compression levels is generated to meet the steady-state data storage requirements.

[0050] For the three independent data units already generated—continuous metering data units, behavioral change data units, and steady-state expression data units—differentiated structural organization expressions are established to suit the characteristics of each data unit, based on the structural granularity corresponding to each type of data unit. Specifically, this is achieved by pre-determining the sampling density, recording range, storage format, and detail retention accuracy according to the recovery priority identifier. For continuous metering data units, a time-series continuous storage structure is adopted according to the continuous recording requirements corresponding to their structural granularity, ensuring the continuity and integrity required for metering traceability. For behavioral change data units, a structured organization of event start-change process-end point is adopted based on their structural granularity and the constructed event fragment structure, clearly presenting the complete process of power consumption state switching. For steady-state expression data units, a parametric encoding storage structure is adopted, combining their structural granularity and compressed storage structure, to achieve concise and efficient storage of steady-state data. Through the above differentiated structural organization design, the three types of data units maintain their independence and integrity while achieving adaptable storage and parsing according to their own characteristics and structural granularity, ultimately forming a unified and standardized differentiated structural organization expression that adapts to different needs.

[0051] In one possible implementation, step S500 further includes: Step S510: Divide the voltage fluctuation data into sliding time windows, extract the voltage drop amplitude, fluctuation frequency and duration of continuous low voltage in each time window, and construct fluctuation feature quantities that characterize power supply stability.

[0052] Step S520: Based on the trend of fluctuation characteristics between adjacent time windows, identify the evolution process of voltage from a stable state to an abnormal fluctuation state, and establish a trend factor characterizing the degree of power supply degradation.

[0053] Step S530: Perform fusion calculation on the fluctuation characteristic quantity and trend factor to generate a power outage risk prediction index that represents the probability of power outage occurring within a preset time range.

[0054] Specifically, a sliding time window with a fixed step size and fixed window length is applied to the real-time acquired voltage fluctuation data to achieve segmented analysis of the voltage time series data. Within each independent sliding time window, three core operating characteristics are quantitatively extracted: voltage drop amplitude, voltage fluctuation frequency, and duration of continuous low voltage. The voltage drop amplitude reflects the degree of voltage decline from the rated operating voltage, the voltage fluctuation frequency is used to count the number of abnormal voltage oscillations per unit time, and the duration of continuous low voltage is used to record the cumulative time when the voltage remains in the excessively low range. The above three quantitative characteristics are normalized and fused to form a multi-dimensional, quantifiable fluctuation characteristic quantity, thereby comprehensively and objectively characterizing the overall power supply stability level of the power grid in the current period.

[0055] The fluctuation characteristics corresponding to consecutive adjacent sliding time windows are extracted sequentially. The difference between the fluctuation characteristics of adjacent windows and the slope of change per unit time are used as input features. A trend factor quantification model is constructed using an LSTM (Long Short-Term Memory) network algorithm. This model consists of an input layer, a hidden layer, and an output layer. The input layer receives two types of temporal features: the normalized difference between the fluctuation characteristics of adjacent windows and the slope of change. The hidden layer uses three LSTM units, each containing 64 neurons. Through the synergistic effect of the forget gate, input gate, and output gate, the model captures the long-term dependence and short-term abrupt change patterns of the temporal changes in fluctuation characteristics, accurately identifying the voltage change from a stable state to an abnormal state. The model is designed to continuously evolve from a state of constant fluctuation. The output layer uses a fully connected layer to map the temporal evolution features extracted from the hidden layer into a single quantized value, namely the trend factor. During model training, the degradation degree label corresponding to the historical voltage anomaly evolution data is used as a supervision signal. The Adam optimizer minimizes the mean square error between the predicted value and the true label, and iteratively adjusts the model parameters until convergence. Finally, by inputting the real-time adjacent window fluctuation feature change data through the trained LSTM model, the model can output a trend factor that accurately represents the degree of power supply degradation. The larger the value of this factor, the more severe the power supply degradation and the faster the evolution towards an abnormal fluctuation state.

[0056] Normalized fluctuation features and trend factors are used as dual-dimensional input features. A lightweight multilayer perceptron (MLP) fusion algorithm is introduced to complete feature coupling calculation. The MLP consists of an input layer, a single-layer hidden layer, and an output layer. The input layer connects to two normalized features. The hidden layer completes feature crossover and deep fusion through a nonlinear activation function to explore the correlation between the voltage static fluctuation amplitude and the dynamic degradation trend. The output layer performs probability mapping through a sigmoid activation function to convert the fused comprehensive features into probability values ​​in the range of 0 to 1. Finally, a power outage risk prediction index that can accurately characterize the probability of a sudden power outage fault in the power load within a preset time range is generated.

[0057] In one possible implementation, step S500 further includes: Step S540: Perform adaptive storage control of differentiated structural organization expression using the power failure risk prediction index, including performing storage strategy segmentation of the power failure risk prediction index using a first threshold and a second threshold, and performing adaptive storage control of differentiated structural organization expression through corresponding low-risk storage strategy, medium-risk storage strategy and high-risk storage strategy.

[0058] Specifically, based on the generated power outage risk prediction index, which is a quantitative value representing the probability of power outage within a preset time range, adaptive storage control is performed on the established differentiated structural organization expression to achieve reasonable allocation of storage resources and data security. Two fixed quantitative thresholds are preset, namely a first threshold and a second threshold, where the first threshold is less than the second threshold, both being probability values ​​in the range of 0 to 1, used to divide different power outage risk levels. These two thresholds are used to segment the numerical range of the power outage risk prediction index for storage strategy, clearly defining the index intervals corresponding to low, medium, and high risk levels. When the power outage risk prediction index value is less than the first threshold, it is determined to be a low power outage risk, and a low-risk storage strategy is executed. Metered continuous data units, behavioral change data units, and steady-state expression data units in the differentiated structural organization expression are stored conventionally according to a preset basic granularity, with appropriate compression of non-critical data, balancing storage efficiency and basic requirements. Data retention: When the power outage risk prediction index value is between the first and second thresholds, it is determined to be of medium power outage risk. A medium-risk storage strategy is implemented to improve the granularity of event fragment structure retention for behavioral change data units, increase the encoding accuracy of steady-state expression data units, and ensure the integrity and continuity of metering continuous data units, balancing storage resources and data recovery needs. When the power outage risk prediction index value is greater than the second threshold, it is determined to be of high power outage risk. A high-risk storage strategy is implemented to maximize the structural granularity and data retention accuracy of the three types of data units, cancel the compression processing of non-critical data, and fully retain the original data and event fragment details to ensure that data can be quickly and completely recovered after a power outage. Through the above three-level storage strategy based on threshold segmentation, adaptive storage control for differentiated structural organization and expression is achieved, which avoids the waste of storage resources in low-risk scenarios and ensures data security and recoverability in high-risk scenarios.

[0059] In one possible implementation, step S540 further includes: Step S541: The behavior change data unit is cached using a delayed write method, and the data write frequency is reduced.

[0060] Step S542: Use a parameterized storage method with a high compression ratio for the steady-state expression data unit to reduce storage resource consumption and reduce write loss.

[0061] Specifically, for the divided behavioral change data units, a delayed write caching storage mechanism is adopted throughout the process. The real-time generated original time-series data of behavioral changes is temporarily stored in the high-speed cache area for buffering, and no immediate persistent write operation is performed. This delays the data write time, while actively reducing the number of batch writes per unit time and reasonably controlling the data write frequency. On the basis of completely preserving the key data information of the behavioral change process, the frequency of continuous read and write consumption of the storage medium is reduced, thereby realizing lightweight caching storage management of behavioral change data units.

[0062] For the established steady-state expression data unit, the direct storage mode of the full original time-series data is abandoned. Instead, a parameterized storage method with a high compression ratio is adopted. The core statistical feature parameters that can characterize the steady-state operating condition are extracted to replace the massive redundant original data. Combined with an efficient compression coding algorithm, the data is simplified. While fully preserving the key feature information of steady-state operation, the overall storage capacity of steady-state data is significantly reduced, effectively reducing the occupation of device storage resources. At the same time, the data writing volume is reduced, the frequent data writing operations are reduced, and the read and write load of the storage medium is reduced. This significantly reduces the write loss during long-term operation, realizing energy-saving, lightweight and efficient storage of the steady-state expression data unit.

[0063] In one possible implementation, step S540 further includes: Step S543: Perform segmented writing processing on the event fragment structure in the behavior change data unit, and trigger fragment solidification when the change process ends.

[0064] Step S544: Increase the write cycle frequency of the continuous metering data unit to ensure the phased recoverability of the metering status.

[0065] Specifically, for the event fragment structure built within the behavior change data unit, a differentiated storage processing mechanism of segmented writing is adopted throughout the process. The change starting point data and dynamic data of the change process contained in the event fragment are cached and temporarily stored in batches, without immediate overall solidification storage. After the complete change process corresponding to a single voltage state switch is completely finished and the operating conditions return to stability, a unified storage trigger signal is issued to centrally complete the data solidification writing operation of the entire event fragment structure. This not only ensures the temporal integrity of the data throughout the behavior change process, but also reduces fragmented writing operations and optimizes the overall read and write load.

[0066] For independently divided continuous metering data units, the data writing time interval is proactively shortened, the overall writing cycle frequency is increased, and the real-time recording frequency of raw time-series metering data is encrypted to continuously and uninterruptedly retain continuous and complete metering operation data. By strengthening the high-density time-series storage and recording of metering data, the operating condition information of different operating stages is completely retained, effectively ensuring that the equipment metering status has clear and complete stage traceability capabilities and reliable data recoverability in scenarios such as power supply anomalies, voltage fluctuations, or sudden power outages, meeting the usage requirements for accurate metering data restoration and fault review.

[0067] In one possible implementation, step S540 further includes: Step S545: Perform immediate writing and state solidification processing on the recovery priority identifier in the differential structure organization expression according to the priority order.

[0068] Step S546: Forcefully truncate the incomplete event fragments in the behavior change data unit and generate a temporary closed fragment in the current state to ensure data reconstructability at the time of power failure.

[0069] Specifically, for the recovery priority identifiers bound to each data unit within the differentiated structure organization expression system, hierarchical scheduling is performed in sequence according to the pre-set high and low priority sorting rules. In strict accordance with the priority order corresponding to the identifier, high priority data is given priority to perform immediate write operations and real-time status solidification processing, skipping the cache delay storage mechanism, and synchronously persisting key data to the storage medium, ensuring that high recovery level data is effectively retained in real time, and strengthening the overall data system's ability to resist loss under abnormal working conditions.

[0070] When the power outage risk exceeds the standard or the emergency protection logic is triggered by abnormal power supply conditions, a forced truncation operation is performed on the incomplete event segments in the behavior change data unit that are in the cache temporary storage stage and have not yet completed the complete time sequence record. The current system acquisition time is used as the segment termination boundary, and the segment structure identifier and status description information are supplemented to quickly generate a temporary closed segment that conforms to the standardized format. This avoids the process data from being broken or lost due to abnormal power outages, and completely preserves the real-time operating condition evolution information before the power outage occurs. This effectively ensures that the original operating data at the moment of sudden power outage has complete analytical basis and reliable data reconstructability.

[0071] Example 2, based on the same inventive concept as the data storage method for multi-functional energy meters in the foregoing examples, such as... Figure 2 As shown, this application provides a data storage system for a multi-functional energy meter. The system and method embodiments in this application are based on the same inventive concept. The system includes: The recovery value parameter construction module 10 is used to collect multi-source power data during the operation of the power meter, and construct recovery value parameters to characterize the contribution of the data to the recovery of the state after power failure based on the change amplitude, change rate and state duration of the multi-source power data in the time series.

[0072] The data stream forming module 20 is used to uniformly mark the multi-source power data according to the recovery value parameters to form a data stream with a recovery priority identifier.

[0073] The semantic layering processing module 30 is used to perform semantic layering processing on multi-source power data based on the characteristics of power consumption behavior changes during the operation of the power meter, and to establish semantic layering results of power consumption behavior.

[0074] The differential structure organization expression construction module 40 is used to construct differential structure organization expressions using the semantic hierarchical results of electricity consumption behavior and the recovery priority identifier.

[0075] The adaptive storage control module 50 is used to read voltage fluctuation data from the power supply side of the electricity meter, construct a power outage risk prediction index based on the voltage fluctuation data, and use the power outage risk prediction index to perform adaptive storage control with differentiated structure organization expression.

[0076] Furthermore, the system is also used to implement the following functions: The multi-source power data is divided into sliding time windows to establish multiple continuous time segments. Within each continuous time segment, the local variation gradients of voltage, current, and power data are calculated to form a first characteristic quantity representing the intensity of instantaneous fluctuations. Based on the difference in the first characteristic quantity between adjacent time segments, an evolution continuity index of the power operating state is calculated. This evolution continuity index is used to identify state abrupt change points and stable state intervals, forming a second characteristic quantity representing state predictability. The data corresponding to the stable state intervals undergo duration statistical processing, and a state stability contribution factor is constructed based on the frequency of state abrupt change points, forming a third characteristic quantity representing the ability to maintain data stability. The first, second, and third characteristic quantities are weighted and coupled to establish a recovery value parameter representing the contribution intensity of power state reconstruction during power outage recovery.

[0077] Furthermore, the system is also used to implement the following functions: The multi-source power data is aligned according to a unified time reference to construct a synchronous correlation sequence between voltage, current, and power data. Based on this synchronous correlation sequence, the load matching relationship between power data and current, and the coupling relationship between voltage disturbance and power response are calculated to form behavioral correlation features characterizing the operating status of electrical equipment. Based on the change patterns of these behavioral correlation features in the time series, typical behavioral units in the power consumption process are identified. These typical behavioral units include load access behavioral units, load exit behavioral units, and stable operation behavioral units. Continuous data is reconstructed using these typical behavioral units as the basic granularity to form a behavioral unit sequence. Contextual correlation analysis is performed on the behavioral unit sequence. Based on the temporal connection relationship and recurrence pattern between adjacent behavioral units, periodic power consumption patterns and aperiodic disturbance patterns are identified. Semantic aggregation processing is then performed on the behavioral unit sequence to form a hierarchical set of behavioral segments. Based on the composition structure and persistence characteristics of each behavioral segment in the behavioral segment set, each behavioral segment set is assigned a power consumption behavior semantic label. These labels include metering continuity semantics, behavioral change semantics, and steady-state maintenance semantics, establishing a hierarchical result for power consumption behavior semantics.

[0078] Furthermore, the system is also used to implement the following functions: Based on the semantic hierarchical results of electricity consumption behavior and the combination relationship of recovery priority identifiers, multi-source power data is structurally mapped to generate continuous metering data units, behavior change data units, and steady-state expression data units. The structural granularity of each data unit is determined according to the recovery priority identifier. Specifically, for the behavior change data units, continuous data segments corresponding to the change process are extracted to construct an event segment structure containing the change start point, change process, and termination point. The data retention granularity of the event segment structure is adjusted according to the recovery priority identifier. For the steady-state expression data units, statistical parameters representing steady-state characteristics are extracted and parameterized. Different levels of encoding processing are performed in combination with the recovery priority identifier to form a compressed storage structure. Differentiated structural organization expressions are established according to the continuous metering data units, behavior change data units, and steady-state expression data units and their corresponding structural granularities.

[0079] Furthermore, the system is also used to implement the following functions: The voltage fluctuation data is divided into sliding time windows. Within each time window, the voltage drop amplitude, fluctuation frequency, and duration of continuous low voltage are extracted to construct fluctuation characteristic quantities that characterize power supply stability. Based on the changing trend of fluctuation characteristic quantities between adjacent time windows, the evolution process of voltage from a stable state to an abnormal fluctuation state is identified, and a trend factor characterizing the degree of power supply degradation is established. The fluctuation characteristic quantities and trend factor are fused and calculated to generate a power outage risk prediction index that characterizes the probability of power outage occurring within a preset time range.

[0080] Furthermore, the system is also used to implement the following functions: Adaptive storage control for differentiated structural organization expression is performed using the power loss risk prediction index, including performing storage strategy segmentation based on the power loss risk prediction index using a first threshold and a second threshold, and performing adaptive storage control for differentiated structural organization expression through corresponding low-risk storage strategy, medium-risk storage strategy, and high-risk storage strategy.

[0081] Furthermore, the system is also used to implement the following functions: For behavioral change data units, a delayed write method is used for cache storage to reduce the data write frequency; for steady-state expression data units, a parameterized storage method with a high compression ratio is used to reduce storage resource consumption and write loss.

[0082] Furthermore, the system is also used to implement the following functions: Segmented writing is performed on the event fragment structure in the behavior change data unit, and the fragment is solidified when the change process ends; the writing cycle frequency is increased for the metering continuous data unit to ensure the phased recoverability of the metering state.

[0083] Furthermore, the system is also used to implement the following functions: In the expression of differentiated structure organization, the priority identifier is restored and written immediately and the state is solidified in order of priority. Unfinished event fragments in the behavioral change data unit are forcibly truncated and a temporary closed fragment in the current state is generated to ensure the data reconstructability at the time of power failure.

[0084] It should be noted that the order of the embodiments described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. Specific embodiments of this specification have been described above. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0085] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0086] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A data storage method for a multi-functional energy meter, characterized in that, The method includes: Collect multi-source power data during the operation of the power meter, and construct a recovery value parameter to characterize the contribution of the data to the recovery of the state after a power outage based on the change amplitude, change rate and state duration of the multi-source power data in the time series. Based on the recovery value parameters, the multi-source power data is uniformly marked to form a data stream with recovery priority identifiers; Semantic hierarchical processing is performed on multi-source power data based on the characteristics of power consumption behavior changes during the operation of power meters, and semantic hierarchical results of power consumption behavior are established. We utilize the semantic hierarchical results of electricity consumption behavior and restore priority identifiers to construct a differentiated structural organization expression; Read voltage fluctuation data from the power supply side of the electricity meter, construct a power outage risk prediction index based on the voltage fluctuation data, and use the power outage risk prediction index to perform adaptive storage control with differentiated structure organization expression.

2. The data storage method for a multi-functional energy meter as described in claim 1, characterized in that, Construct recovery value parameters to characterize the contribution of data to post-power-out state recovery, including: The multi-source power data is divided into sliding time windows to establish multiple continuous time segments; The local variation gradients of voltage, current, and power data are calculated in each continuous time segment to form the first characteristic quantity characterizing the intensity of instantaneous fluctuations; Based on the difference in the first characteristic quantity between adjacent time segments, the evolution continuity index of the power operation state is calculated, and the evolution continuity index is used to identify the state change points and the state stability intervals, forming a second characteristic quantity that characterizes the predictability of the state. Perform duration statistical processing on the data corresponding to the state stability interval, and construct a state stability contribution factor by combining the occurrence frequency of state change points, forming a third characteristic quantity characterizing the data stability maintenance capability; The first, second, and third characteristic quantities are weighted and coupled to calculate the recovery value parameter that characterizes the contribution intensity of power state reconstruction during the power outage recovery process.

3. The data storage method for a multi-functional energy meter as described in claim 1, characterized in that, Semantic layering processing is performed on multi-source electricity data based on the changing characteristics of electricity consumption behavior during the operation of electricity meters, and the semantic layering results of electricity consumption behavior are established, including: The multi-source power data is aligned according to a unified time base to construct a synchronous correlation sequence between voltage, current and power data. Based on the synchronous correlation sequence, the load matching relationship between power data and current and the coupling relationship between voltage disturbance and power response are calculated to form behavioral correlation features characterizing the operating status of electrical equipment. Based on the changing patterns of the aforementioned behavioral correlation features in the time series, typical behavioral units in the process of electricity use are identified. These typical behavioral units include load access behavioral units, load exit behavioral units, and stable operation behavioral units. Continuous data is then reconstructed using these typical behavioral units as the basic granularity to form a sequence of behavioral units. Context association analysis is performed on the behavioral unit sequence. Based on the temporal connection relationship and recurrence pattern between adjacent behavioral units, periodic power consumption patterns and aperiodic disturbance patterns are identified. Semantic aggregation processing is then performed on the behavioral unit sequence to form a set of behavioral segments with hierarchical relationships. Based on the composition structure and continuous characteristics of each behavior segment in the set of behavior segments, the set of behavior segments is assigned a semantic label for electricity consumption behavior. The semantic label for electricity consumption behavior includes metering continuity semantic, behavior change semantic, and steady-state maintenance semantic, and a hierarchical result for electricity consumption behavior semantic is established.

4. The data storage method for a multi-functional energy meter as described in claim 1, characterized in that, Using semantic hierarchical results of electricity consumption behavior and priority identifiers, a differentiated structural organization expression is constructed, including: Based on the semantic hierarchical results of electricity consumption behavior and the combination relationship of recovery priority identifiers, the multi-source power data is structurally mapped to generate continuous metering data units, behavior change data units and steady-state expression data units, and the structural granularity of each data unit is determined according to the recovery priority identifier. Specifically, continuous data segments corresponding to the change process are extracted from the behavior change data unit, an event segment structure containing the change start point, change process and termination point is constructed, and the data retention granularity of the event segment structure is adjusted according to the recovery priority identifier. For the steady-state expression data unit, statistical parameters characterizing steady-state features are extracted and parameterized, and encoding processing of different precision levels is performed in combination with the recovery priority identifier to form a compressed storage structure; Differentiated structural organization expressions are established based on continuous measurement data units, behavioral change data units, and steady-state expression data units, and their corresponding structural granularities.

5. The data storage method for a multi-functional energy meter as described in claim 1, characterized in that, Based on the voltage fluctuation data, a power outage risk prediction index is constructed, including: The voltage fluctuation data is divided into sliding time windows, and the voltage drop amplitude, fluctuation frequency and duration of continuous low voltage are extracted within each time window to construct fluctuation feature quantities that characterize power supply stability. Based on the trend of fluctuation characteristics between adjacent time windows, the evolution process of voltage from a stable state to an abnormal fluctuation state is identified, and a trend factor characterizing the degree of power supply degradation is established. The fluctuation characteristics and trend factors are fused and calculated to generate a power outage risk prediction index that represents the probability of power outage occurring within a preset time range.

6. The data storage method for a multi-functional energy meter as described in claim 1, characterized in that, Adaptive storage control for differentiated structural organization expression is performed using the power loss risk prediction index, including performing storage strategy segmentation based on the power loss risk prediction index using a first threshold and a second threshold, and performing adaptive storage control for differentiated structural organization expression through corresponding low-risk storage strategy, medium-risk storage strategy, and high-risk storage strategy.

7. The data storage method for a multi-functional energy meter as described in claim 6, characterized in that, When the power failure risk prediction index is lower than a first threshold, a low-risk storage strategy is implemented, including: The behavioral change data units are cached using a delayed write method, and the data write frequency is reduced; A parameterized storage method with a high compression ratio is adopted for steady-state expression data units to reduce storage resource consumption and reduce write losses.

8. The data storage method for a multi-functional energy meter as described in claim 6, characterized in that, When the power outage risk prediction index is between a first threshold and a second threshold, a medium-risk storage strategy is executed, including: Perform segmented writing processing on the event fragment structure in the behavior change data unit, and trigger fragment solidification when the change process ends; Increase the write cycle frequency of continuous metering data units to ensure the phased recoverability of the metering status.

9. The data storage method for a multi-functional energy meter as described in claim 6, characterized in that, When the power failure risk prediction index exceeds the second threshold, a high-risk storage strategy is implemented, including: In the expression of differential structural organization, the priority markers are restored and written immediately and the state is solidified according to the priority order. Unfinished event segments in the behavior change data unit are forcibly truncated, and temporary closed segments in the current state are generated to ensure data reconstructability at the time of power failure.

10. A data storage system for a multi-functional energy meter, characterized in that, The system is used to implement the data storage method for a multi-functional energy meter according to any one of claims 1-9, the system comprising: The recovery value parameter construction module is used to collect multi-source power data during the operation of the power meter, and construct recovery value parameters to characterize the contribution of the data to the recovery of the state after power failure based on the change amplitude, change rate and state duration of the multi-source power data in the time series. The data stream forming module is used to uniformly mark the multi-source power data according to the recovery value parameters, and form a data stream with a recovery priority identifier; The semantic layering processing module is used to perform semantic layering processing on multi-source power data based on the characteristics of power consumption behavior changes during the operation of the power meter, and to establish semantic layering results of power consumption behavior. The differential structure organization expression construction module is used to construct differential structure organization expressions by utilizing the semantic hierarchical results of electricity consumption behavior and the recovery priority identifier; An adaptive storage control module is used to read voltage fluctuation data from the power supply side of the electricity meter, construct a power outage risk prediction index based on the voltage fluctuation data, and use the power outage risk prediction index to perform adaptive storage control with differentiated structure organization expression.

Images