An embedded load identification method and device for an electric energy meter and the electric energy meter

Abstract

The application provides an embedded load identification method and device of an electric energy meter and the electric energy meter. The method comprises: high-frequency sampling of a bus of a power grid to obtain current parameters and voltage parameters of the bus; adopting a preset edge algorithm to cut the current parameters and the voltage parameters according to continuous time to obtain windowed power consumption parameters after cutting; adopting a preset hybrid neural network model to extract appliance characteristics of the power consumption parameters, and outputting power consumption analysis results corresponding to each power consumption device in different time periods according to the appliance characteristics of the power consumption parameters. The appliance characteristics include power consumption information and corresponding power consumption time sequence information of different power consumption devices, and the power consumption analysis results include power consumption device identification results and power consumption analysis data of each power consumption device. Through the combination of edge algorithm preprocessing and a hybrid neural network, high-precision load identification is realized, power consumption analysis results corresponding to each power consumption device in different time periods are outputted, and this is helpful for the control of each power consumption device.

Description

Technical Field

[0001] This application relates to the field of electricity meter technology, and in particular to an embedded load identification method, device and electricity meter. Background Technology

[0002] With the development of technology, electricity meters used in homes or public areas are no longer traditional electricity metering devices, but are gradually becoming intelligent, capable of recording or analyzing the energy consumption of different devices.

[0003] Existing electricity meters can generally only perform simple distinctions and records of the energy consumption status of electrical equipment, or upload the data to the power company's server for detailed analysis, such as analyzing the electricity consumption of a specific appliance over a period of time. However, using existing technology is inefficient due to its reliance on the server, and the analysis results are simplistic with little reference value for electricity consumption and energy conservation. Summary of the Invention

[0004] The purpose of this application is to provide an embedded load identification method, device, and energy meter, which obtains fine electrical signals by high-frequency sampling of the bus, and combines edge algorithm preprocessing and hybrid neural network to obtain the power consumption analysis results of each electrical device at different time periods, thereby achieving high-precision load identification.

[0005] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows:

[0006] In a first aspect, embodiments of this application provide an embedded load identification method for electricity meters, applied to electricity meters, the method comprising:

[0007] The power grid bus is sampled at high frequency to obtain the current and voltage parameters of the bus. The current parameters include instantaneous current value, effective current value, current phase, and current harmonic components, which are used to describe the energy consumption intensity and waveform characteristics of electrical equipment. The voltage parameters include instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, which are used to calculate power and analyze the impact of power grid quality on the load.

[0008] The current parameters and voltage parameters are segmented according to continuous time using a preset edge algorithm to obtain the segmented windowed power consumption parameters;

[0009] A preset hybrid neural network model is used to extract the electrical appliance features of the power consumption parameters, and output the power consumption analysis results of each electrical device for different time periods based on the electrical appliance features of the power consumption parameters. The electrical appliance features include: power consumption information and corresponding power consumption time sequence information for different electrical devices. The power consumption analysis results include: electrical device identification results and power consumption analysis data for each electrical device.

[0010] In an optional implementation, the step of using a preset edge algorithm to segment the current parameters and voltage parameters according to continuous time to obtain the segmented windowed power consumption parameters includes:

[0011] The preset edge algorithm is used to divide the current parameter and the voltage parameter into continuous time segments to obtain window current and window voltage corresponding to multiple time windows;

[0012] A voltage-current trajectory diagram is generated based on the window current and the window voltage.

[0013] In an optional implementation, after using the preset edge algorithm to divide the current parameters and voltage parameters according to continuous time to obtain window currents and window voltages corresponding to multiple time windows, the method further includes:

[0014] Calculate the power consumption parameters within each time window, including: active power, reactive power, and power factor;

[0015] Generate a time-series vector corresponding to the electricity consumption parameters based on the electricity consumption parameters within each time window.

[0016] In an optional implementation, the preset hybrid neural network model includes: a convolutional neural network (CNN) model and a long short-term memory (LSTM) network model;

[0017] The method of extracting electrical appliance features from the electricity consumption parameters using a preset hybrid neural network model includes:

[0018] The voltage-current trajectory diagram is input into the CNN model, and the voltage-current spatial feature vector is output through image dimension recognition.

[0019] The time-series vector corresponding to the power consumption parameters is input into the LSTM model, and the power information and time-series dependency features are output.

[0020] Based on the voltage-current spatial feature vector, the power information, and the time-series dependency characteristics, the probability of different electrical devices being turned on or off in different time windows is identified.

[0021] In an optional implementation, identifying the probability of different electrical devices being turned on or off in different time windows based on the voltage-current spatial feature vector, the power information, and the time-series dependency features includes:

[0022] The voltage-current spatial feature vector, the power information, and the time-series dependency features are fused to obtain the fused feature vector;

[0023] Based on the fused feature vector, different electrical devices are identified, and the probability of different electrical devices being turned on or off in different time windows is output.

[0024] In an optional implementation, the preset hybrid neural network model is trained using the CNN model, the LSTM model, and a preset dataset, wherein the preset dataset includes current data and voltage data corresponding to different electrical equipment usage states or off states.

[0025] In an optional implementation, the preset hybrid neural network model is a lightweight model, wherein the lightweight processing includes: pruning the trained hybrid neural network model and then performing quantization-based perceptual training on the pruned model.

[0026] In an optional implementation, after extracting the electrical appliance characteristics of the power consumption parameters using a preset hybrid neural network model and outputting the power consumption analysis results for each electrical device at different time periods based on the electrical appliance characteristics of the power consumption parameters, the method further includes:

[0027] Based on the power consumption analysis results, a power consumption control strategy is generated;

[0028] Send the power control strategy to the appliance intelligent control platform.

[0029] Secondly, embodiments of this application provide an embedded load identification device for an electricity meter, applied to an electricity meter, the device comprising:

[0030] The sampling module is used to perform high-frequency sampling of the power grid bus to obtain the current and voltage parameters of the main incoming line. The current parameters include: instantaneous current value, effective current value, current phase, and current harmonic components, used to describe the energy consumption intensity and waveform characteristics of electrical equipment. The voltage parameters include: instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, used to calculate power and analyze the impact of power grid quality on the load.

[0031] The processing module is used to segment the current parameters and the voltage parameters according to continuous time using a preset edge algorithm to obtain the segmented windowed power consumption parameters;

[0032] The extraction module is used to extract the electrical appliance features of the power consumption parameters using a preset hybrid neural network model, and output the power consumption analysis results corresponding to different time periods for each electrical appliance based on the electrical appliance features of the power consumption parameters. The electrical appliance features include: power consumption information and corresponding power consumption time sequence information for different electrical appliances. The power consumption analysis results include: electrical appliance identification results and power consumption analysis data for each electrical appliance.

[0033] Thirdly, embodiments of this application provide an energy meter, including: a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the computer device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to perform the steps of the energy meter embedded load identification method as described in any of the first aspects.

[0034] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the embedded load identification method for electricity meters described in any of the first aspects above.

[0035] Compared to existing technologies, this application provides an embedded load identification method, device, and energy meter. The method includes: high-frequency sampling of the power grid bus to obtain current and voltage parameters of the bus; wherein, the current parameters include: instantaneous current value, effective current value, current phase, and current harmonic components, used to describe the energy consumption intensity and waveform characteristics of electrical equipment; the voltage parameters include: instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, used to calculate power and analyze the impact of power grid quality on the load; using a preset edge algorithm to segment the current and voltage parameters according to continuous time to obtain segmented windowed power consumption parameters; using a preset hybrid neural network model to extract the electrical features of the power consumption parameters, and outputting the power consumption analysis results corresponding to different time periods for each electrical equipment based on the electrical features of the power consumption parameters, wherein the electrical features include: power consumption information and corresponding power consumption time sequence information corresponding to different electrical equipment, and the power consumption analysis results include: electrical equipment identification results and power consumption analysis data for each electrical equipment. The method of this application obtains fine electrical signals by high-frequency sampling of the bus, and combines edge algorithm preprocessing and hybrid neural network to achieve high-precision load identification, reduce data transmission power consumption and privacy leakage risks; it decomposes the total power consumption to each power-consuming device and outputs the power consumption analysis results of each power-consuming device at different time periods, which helps to control each power-consuming device.

[0036] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 One of the flowcharts for an embedded load identification method in an energy meter provided in this application embodiment;

[0039] Figure 2 A second schematic flowchart illustrating an embedded load identification method for an energy meter provided in this application embodiment;

[0040] Figure 3 The third schematic flowchart of an embedded load identification method for an energy meter provided in this application embodiment;

[0041] Figure 4 The fourth flowchart illustrates a method for identifying an embedded load in an energy meter, as provided in this application embodiment.

[0042] Figure 5 Fifth schematic flowchart of an embedded load identification method for an energy meter provided in this application embodiment;

[0043] Figure 6 A schematic flowchart of an embedded load identification method for an energy meter provided in this application embodiment is shown in Figure 6.

[0044] Figure 7 A schematic diagram of the functional modules of an embedded load identification device in an energy meter provided in this application embodiment;

[0045] Figure 8 This is a schematic diagram of an electricity meter provided in an embodiment of this application. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, 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 some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0047] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0048] In the description of this application, it should be noted that if the terms "upper", "lower", etc. appear to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of this application is usually placed in, it is only for the convenience of describing this application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0049] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Additionally, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0050] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0051] This application provides an embedded load identification method for electricity meters, which is applied to electricity meters. Please refer to [link to relevant documentation]. Figure 1 , Figure 1 This application provides a flowchart illustrating one method for embedded load identification in an electricity meter, comprising:

[0052] S101. Perform high-frequency sampling on the power grid bus to obtain the bus current and voltage parameters.

[0053] In this embodiment, the electricity meter serves as a terminal measurement device in the power system, used to monitor the total electricity consumption in residential or industrial settings in real time. High-frequency sampling aims to acquire high-resolution electrical signal data to capture subtle electricity consumption characteristics of different devices, such as sudden current changes during switching and power fluctuations during steady-state operation, providing a raw data foundation for subsequent load identification. Specifically, the electricity meter can perform high-frequency synchronous sampling of the voltage and current signals transmitted via the power grid's bus.

[0054] Among them, electricity meters integrate high-precision analog-to-digital converter (ADC) chips at the hardware level, such as the ATT7022E, which internally uses a 24-bit ADC to achieve high-precision analog signal conversion. These ADCs can convert analog voltage and current signals from the bus into digital signals, offering high resolution and accurately capturing minute changes in voltage and current. Furthermore, to improve measurement accuracy and anti-interference capabilities, the signal input section of electricity meters often employs a differential input method. By comparing the voltage difference between the two input terminals to obtain the signal, common-mode noise can be effectively suppressed, ensuring the accuracy of the acquired current and voltage signals.

[0055] High-frequency sampling typically refers to sampling frequencies higher than those of conventional power metering (e.g., more than 10 times the 50Hz grid frequency, commonly 1kHz to 10kHz). This ensures that high-frequency components such as appliance start-up and shutdown, and harmonics can be captured. Precise clock signals are required to control the sampling frequency and timing, ensuring the accuracy and consistency of sampling points, satisfying the Nyquist sampling theorem, and avoiding aliasing.

[0056] Specifically, the microprocessor of the electricity meter triggers sampling at a preset frequency through timer interrupts and other methods. For example, it triggers 1500 timer interrupts per second, recording the current voltage and current data each time, thus achieving high-frequency, precise sampling of transient changes in voltage and current thousands of times per second. The sampled data may contain noise and interference. The electricity meter processes the sampled data using a digital filter to remove high-frequency noise and other interference signals, ensuring the accuracy of the data ultimately used for calculation and analysis. Finally, the collected high-frequency data is temporarily stored in the internal storage area of ​​the electricity meter's microprocessor. When certain conditions are met, such as if the number of times the data is recorded in the internal storage area exceeds a preset number, the data is transferred to external memory for subsequent analysis and processing. Simultaneously, the data can be transmitted to the power system's main station or other terminal equipment via the communication module.

[0057] The current parameters include: instantaneous current value, effective current value, current phase, and current harmonic components, which are used to describe the energy consumption intensity and waveform characteristics of electrical equipment; the voltage parameters include: instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, which are used to calculate power and analyze the impact of power grid quality on the load.

[0058] S102. The current and voltage parameters are segmented according to continuous time using a preset edge algorithm to obtain the segmented windowed power consumption parameters.

[0059] Specifically, the preset edge algorithm refers to a lightweight algorithm that runs locally on the electricity meter. It is used to perform initial processing on current and voltage parameters to obtain electricity consumption parameters. The purpose of the initial processing is to clean, compress, and preliminarily extract features from the original sampled data.

[0060] By employing a pre-defined edge algorithm, the continuously acquired high-frequency current and voltage signals are divided into several time segments according to the continuous time, and the dynamically changing electrical signals are transformed into local data units that can be analyzed independently, so as to capture the electricity consumption characteristics in different time intervals.

[0061] S103. Using a preset hybrid neural network model, extract the electrical appliance characteristics of the power consumption parameters, and output the power consumption analysis results corresponding to different time periods for each electrical device based on the electrical appliance characteristics of the power consumption parameters.

[0062] Among them, the electrical appliance characteristics include: the power consumption information and the corresponding power consumption time sequence information for different electrical appliances, and the power consumption analysis results include: the identification results of electrical appliances, and the power consumption analysis data of each electrical appliance.

[0063] Specifically, the pre-defined hybrid neural network combines the advantages of different networks to more comprehensively capture the characteristics of electrical appliances. These characteristics include power consumption information such as power output, waveform shape, and power consumption timing information such as start-up time, runtime, and state change patterns. The final output is the power consumption analysis results for each electrical device at different time periods. For devices including smart home devices, the power consumption analysis results may include: device identification results, i.e., identifying which devices are using the appliances at different time periods; and power consumption analysis data, including the on / off status of each device, and possibly real-time power, power consumption, and runtime of each device.

[0064] In summary, this application provides an embedded load identification method for electricity meters, applied to electricity meters. The method includes: high-frequency sampling of the power grid bus to obtain current and voltage parameters; wherein the current parameters include: instantaneous current value, effective current value, current phase, and current harmonic components, used to describe the energy consumption intensity and waveform characteristics of electrical equipment; the voltage parameters include: instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, used to calculate power and analyze the impact of power grid quality on the load; using a preset edge algorithm to segment the current and voltage parameters according to continuous time, obtaining windowed electricity consumption parameters after segmentation; using a preset hybrid neural network model to extract the electrical appliance features of the electricity consumption parameters, and outputting electricity consumption analysis results corresponding to different time periods for each electrical device based on the electrical appliance features of the electricity consumption parameters. The electrical appliance features include: electricity consumption information and corresponding electricity consumption time sequence information corresponding to different electrical devices; the electricity consumption analysis results include: electrical device identification results and electricity consumption analysis data for each electrical device. The method of this application obtains fine electrical signals by high-frequency sampling of the bus, and combines edge algorithm preprocessing and hybrid neural network to achieve high-precision load identification, reduce data transmission power consumption and privacy leakage risks; it decomposes the total power consumption to each power-consuming device and outputs the power consumption analysis results of each power-consuming device at different time periods, which helps to control each power-consuming device.

[0065] Figure 2 This is a second schematic flowchart illustrating an embedded load identification method for an electricity meter, as provided in this application embodiment. Figure 2 As shown, a preset edge algorithm is used to segment the current and voltage parameters according to continuous time, obtaining the segmented windowed power consumption parameters, including:

[0066] S201. The current and voltage parameters are divided into continuous time segments using a preset edge algorithm to obtain the window current and window voltage corresponding to multiple time windows.

[0067] In this embodiment, the time window is defined as a fixed time length (e.g., 10ms, 50ms) set by a preset edge algorithm, dividing the continuous current / voltage sampling sequence into multiple non-overlapping or partially overlapping subsequences. That is, the current and voltage parameters are divided according to continuous time to obtain the window current and window voltage corresponding to multiple time windows. For example, if the sampling frequency is 1kHz (1000 points per second), setting a 50ms window means each window contains 50 sampling points. The window length needs to be set according to the power consumption characteristics: a short window, such as 10ms, is suitable for capturing transient changes during device start-up and shutdown, such as the current spike during motor startup; a long window, such as 100ms, is suitable for analyzing steady-state operating characteristics, such as the stable current of an incandescent lamp. Additionally, an overlap rate, such as 50%, can be set to avoid feature loss at window boundaries and ensure the integrity of the continuous signal.

[0068] S202. Generate a voltage-current trajectory diagram based on the window current and window voltage.

[0069] Specifically, the current and voltage data within each time window are converted into two-dimensional image features to intuitively present the correspondence between voltage and current within that time period.

[0070] For example, for each power frequency cycle within each window, a closed VI trajectory curve is plotted with the instantaneous voltage v(t) of that cycle as the X-axis and the instantaneous current i(t) as the Y-axis. This trajectory is rasterized and normalized to generate a fixed-size two-dimensional grayscale image (e.g., 32x32 pixels). Different types of electrical equipment will produce VI trajectory maps with significantly different shapes, resulting in a highly recognizable electrical fingerprint. The electrical parameters include: voltage-current trajectory map.

[0071] In the method provided in this application embodiment, a preset edge algorithm is used to segment the current and voltage parameters according to continuous time, obtaining window current and window voltage corresponding to multiple time windows; based on the window current and window voltage, a voltage-current trajectory diagram is generated. This transforms the original high-frequency current and voltage parameters into a structured and visualized voltage-current trajectory diagram, preserving the detailed features of local time intervals while adapting to the feature extraction requirements of neural networks through image format, laying the foundation for accurate recognition by hybrid models, and simultaneously achieving efficient local processing through edge computing.

[0072] Figure 3 This is the third flowchart illustrating an embedded load identification method for an electricity meter provided in this application. Figure 3 As shown, after using a preset edge algorithm to divide the current and voltage parameters into continuous time segments and obtaining the window current and window voltage corresponding to multiple time windows, the process also includes:

[0073] S301. Calculate the power consumption parameters within each time window.

[0074] The electrical parameters include: active power, reactive power, and power factor.

[0075] In this embodiment, key numerical parameters reflecting the energy consumption characteristics and electrical operating status of electrical equipment are extracted from the current and voltage data within the time window, providing quantitative indicators for subsequent time series analysis.

[0076] Specifically, within the same window, macroscopic parameters such as total active power P, reactive power Q, and power factor PF are calculated, forming a time series vector S = [P(t), Q(t), PF(t)]. Active power refers to the power actually consumed in the circuit for doing work, calculated by integrating the instantaneous power (instantaneous voltage × instantaneous current) within the time window and taking the average value. Reactive power is calculated based on the phase difference between voltage and current, using the orthogonal components of the instantaneous power. The power factor is the ratio of active power to apparent power (RMS voltage × RMS current), used to reflect electricity efficiency.

[0077] S302. Generate a time-series vector corresponding to the power consumption parameters based on the power consumption parameters within each time window.

[0078] Specifically, the power consumption parameters (active power, reactive power, and power factor) of discrete time windows are connected in series in time order to form a continuous time-series characteristic sequence, i.e., time-series vector, in order to capture the dynamic change pattern of power consumption parameters over time.

[0079] The structure of the time-series vector can be assumed to be divided into N consecutive windows in chronological order (e.g., window 1, window 2, ..., window N), with the power consumption parameters of each window being [Pi, Qi, PFi] (where i is the window number). Then the time-series vector would be: [[P1, Q1, PF1], [P2, Q2, PF2], ..., [Pn, Qn, PFn]], resulting in an n×3 two-dimensional vector (n is the number of windows, and 3 is the parameter dimension). The processed power consumption parameters then include the time-series vector.

[0080] The method provided in this application calculates the power consumption parameters within each time window, including active power, reactive power, and power factor. A time-series vector corresponding to the power consumption parameters is generated based on these parameters within each time window. The electrical signals within the time window are converted into data forms that combine quantification features and time correlation: parameters such as active / reactive power, quantifying the energy consumption and electrical characteristics of the electrical equipment. The time-series vector preserves the dynamic changes of these parameters over time.

[0081] Figure 4 This is the fourth flowchart illustrating an embedded load identification method for an electricity meter provided in this application. The preset hybrid neural network model includes: a Convolutional Neural Network (CNN) model and a Long Short-Term Memory (LSTM) model; as shown... Figure 4 As shown, a pre-defined hybrid neural network model is used to extract the electrical appliance features of power consumption parameters, including:

[0082] S401. Input the voltage-current trajectory diagram into the CNN model, and output the voltage-current spatial feature vector through image dimension recognition.

[0083] In this embodiment, the strong image feature extraction capability of convolutional neural networks (CNN) is utilized to capture the spatial fingerprints of different electrical devices from the voltage-current (VI) trajectory map, and the visual features of the image are transformed into quantifiable vector representations.

[0084] Specifically, given a 32×32 pixel voltage-current trajectory map as input, the CNN model structure includes: Convolutional layer C1: 16 3x3 convolutional kernels with ReLU activation function, used to extract low-level features such as edges and corners of the trajectory map. Max pooling layer P1: 2x2 pooling window, used to reduce the dimensionality of the feature map and retain the most salient features. Convolutional layer C2: 32 3x3 convolutional kernels with ReLU activation function, used to combine low-level features to form more complex shape features. Max pooling layer P2: 2x2 pooling window. Flattening layer: Converts the two-dimensional feature map output by C2 into a one-dimensional feature vector, thus obtaining the voltage-current spatial feature vector corresponding to the voltage-current trajectory map.

[0085] S402. Input the time-series vectors corresponding to the power consumption parameters into the LSTM model, and output the power information and the time-series dependency features.

[0086] Among them, the ability of Long Short-Term Memory (LSTM) network to model time-series data is utilized to capture the dynamic changes of power consumption parameters (active power, reactive power, and power factor) over time, and to extract the time-series fingerprint of the operation of power equipment.

[0087] Specifically, the input is a temporal vector S_seq = [S(t-N+1), ..., S(t)] of length N (e.g., N=10, representing 10 consecutive time steps). Since the LSTM model structure is: LSTM layer L1 contains 64 hidden units, this layer can learn the temporal dependencies of the temporal vector. The output is the output vector of the LSTM layer at the last time step t. This yields power information and temporal dependency features.

[0088] S403. Based on the voltage-current spatial feature vector, power information, and time-series dependency characteristics, identify the probability of different electrical devices being turned on or off in different time windows.

[0089] Specifically, the voltage-current spatial feature vector extracted by the CNN model is combined with the power information and temporal dependency features extracted by the LSTM model to comprehensively judge the operating status of each electrical device within each time window, and output probability values ​​to reflect the credibility of the judgment.

[0090] It should be noted that after identifying the probability of different electrical devices being turned on or off in different time windows, the power consumption information of the electrical devices that are turned on can be obtained, and the power consumption analysis results of each electrical device in different time periods can be output.

[0091] The method provided in this application involves inputting a voltage-current trajectory diagram into a CNN model, identifying the image dimension, and outputting a voltage-current spatial feature vector. The time-series vectors corresponding to the power consumption parameters are input into an LSTM model, outputting power information and time-series dependency features. Based on the voltage-current spatial feature vectors, power information, and time-series dependency features, the probability of different electrical devices being turned on or off in different time windows is identified. This achieves a comprehensive characterization of electrical appliance features: spatial features distinguish the inherent electrical attributes of different devices, and time-series features capture the operating mode of the devices over time. The fusion of these two features significantly improves the accuracy and robustness of load identification. The final output probability value not only determines the status of each electrical device but also reflects the reliability of the identification through the probability magnitude, providing detailed basic data for subsequent power consumption analysis.

[0092] Figure 5This is the fifth flowchart illustrating an embedded load identification method for an electricity meter, as provided in the embodiments of this application. Figure 5 As shown, based on the spatial feature vectors of voltage and current, power information, and the time-series dependency characteristics, the probability of different electrical devices being turned on or off in different time windows is identified, including:

[0093] S501. The voltage-current spatial feature vector, power information and time-series dependency features are fused to obtain the fused feature vector.

[0094] In this embodiment, the voltage-current spatial feature vector, power information, and timing dependency features are integrated to form a comprehensive feature containing multi-dimensional information, avoiding the limitations of a single feature dimension and improving the accuracy of subsequent identification.

[0095] The following methods can be used for fusion, specifically:

[0096] (1) Directly concatenate the two types of feature vectors according to their dimensions to form a longer composite vector. For example: the voltage-current spatial feature vector is 256 (encoding trajectory shape, harmonic characteristics, etc.); the temporal dependency feature vector is 128-dimensional (encoding power change trend, start-stop timing, etc.); after fusion, it becomes a 256+128=384-dimensional feature vector, retaining the complete information of both types of features. The advantage is that it is simple and direct, does not lose the original feature details, and is suitable for scenarios where both types of features are equally important.

[0097] (2) The two types of feature vectors are summed after being assigned different weights. The weights are learned automatically through model training or set manually. For example: fusion vector = α × spatial feature vector + (1-α) × relational feature vector (α is the weight, 0<α<1). The advantage is that the weights can be dynamically adjusted according to the importance of the features.

[0098] (3) Introduce an attention layer to allow the model to automatically learn the importance of different features in different time windows. For example, calculate attention weights for spatial features and relational features separately (e.g., the weight of spatial features is 0.7 and the weight of relational features is 0.3 in a certain window); sum the weights to obtain a fusion vector, thereby dynamically focusing on key features. The advantage is that it can adaptively adapt to different scenarios (e.g., when complex loads are superimposed, it automatically amplifies the weights of features with high discriminative power).

[0099] S502. Based on the fused feature vector, identify different electrical devices and output the probability of different electrical devices being turned on or off in different time windows.

[0100] Based on the integrated features after fusion, a classifier is used to determine which electrical appliances are in operation within each time window, and the reliability of the judgment is quantified in the form of probability, providing a detailed status basis for subsequent power consumption analysis.

[0101] Specifically, the input fused feature vector is used in a classification model with a fully connected layer and a softmax activation function. The fully connected layer maps the high-dimensional fused features to dimensions corresponding to the types of appliances to be identified, completing the conversion from features to category probabilities. The softmax function normalizes the output of the fully connected layer into probability values ​​(the probability of each dimension ∈ [0,1], and the sum of the probabilities of all dimensions is 1), ensuring the output conforms to the probability distribution characteristics. For example, the output is an M-dimensional probability vector Y = [y_1, y_2, ..., y_M], where y_i represents the probability that the i-th type of appliance is in the on state.

[0102] The method provided in this application fuses voltage-current spatial feature vectors, power information, and temporal dependency features to obtain a fused feature vector. Based on the fused feature vector, it identifies different electrical appliances and outputs the probability of each appliance being turned on or off in different time windows. This achieves a precise mapping from multi-dimensional features to appliance states: the fusion process integrates the advantages of spatial and temporal features, compensating for the information loss of single features; the probability output provides the operating status and reliability of each appliance in a quantitative form, supporting both single-device identification and handling complex scenarios where multiple devices operate simultaneously, providing basic data for subsequent power consumption analysis.

[0103] This application provides a possible implementation of an embedded load identification method for electricity meters. The preset hybrid neural network model is trained using a CNN model, an LSTM model, and a preset dataset. The preset dataset includes current data and voltage data corresponding to the usage or off states of different electrical devices.

[0104] In this embodiment, a pre-defined hybrid neural network model architecture is constructed. The CNN model includes: initial convolutional layers, pooling layers, and fully connected layers (e.g., 3 convolutional layers + 2 fully connected layers) for extracting spatial features of the VI trajectory map; the LSTM model includes: initial LSTM units and subsequent fully connected layers for extracting dynamic features of the temporal vector; the fusion layer includes initial feature fusion methods (e.g., concatenation, weighted fusion) and the final classification layer (fully connected layer + softmax) for outputting the probability of appliance switching.

[0105] The pre-processed dataset yields training samples including voltage-current trajectory maps and time-series vectors corresponding to power consumption parameters. These training samples are then input into a CNN model and an LSTM model, respectively. The VI trajectory map is input into the CNN model, which outputs a voltage-current spatial feature vector after convolution and pooling operations. The time-series vectors are input into the LSTM model, which outputs a power time-dependent feature vector after gating. Finally, a fusion layer integrates the two types of feature vectors, and a classification layer outputs the predicted on / off probability of each appliance (e.g., a sample predicts an air conditioner on with a probability of 0.8).

[0106] Cross-entropy loss is used to calculate the difference between the model's predicted probability and the true label (e.g., 1 for air conditioner). A smaller loss value indicates a prediction closer to the true state. The loss value is then propagated back from the output layer to each layer of the CNN and LSTM using gradient descent to adjust network parameters (weights, biases) and reduce the loss. For example, if the model's predicted probability of the microwave oven being turned on is too low, the weights of the convolutional kernels in the CNN related to the microwave oven's VI trajectory features are increased through backpropagation.

[0107] Finally, the pre-defined dataset is divided into a training set (70%, used for parameter learning), a validation set (20%, used to monitor overfitting), and a test set (10%, used for final performance evaluation). The forward propagation, loss calculation, and backpropagation processes are repeated, with performance evaluated on the validation set after each iteration. If the validation set performance decreases, it indicates that the model is overfitting the training data and needs optimization through early stopping, regularization, etc. When the training set loss steadily decreases and the validation set performance no longer improves, training is stopped, and the model parameters at this point are saved; this is the pre-defined hybrid neural network model that has been successfully trained.

[0108] Optionally, the hybrid neural network model is a lightweight model, wherein the lightweight processing includes: pruning the trained hybrid neural network model and then performing quantization-aware training on the pruned model.

[0109] Specifically, the essence of pruning training is to remove redundant or unimportant parameters / structures from the model, retain the core parts that are crucial to recognition performance, and avoid a significant drop in accuracy while reducing the number of parameters and computational load.

[0110] After model training, the importance of parameters in each layer is analyzed (usually measured by the absolute value of weights), and redundant parameters with minimal impact on the output are filtered out: For CNNs, low-contribution convolutional kernels and unimportant neurons in fully connected layers are identified; for LSTMs, parameters with weights close to 0 in gating mechanisms (input gate, forget gate, output gate) or hidden layer units that contribute little to temporal dependency modeling are identified. A threshold is set (e.g., retaining the top 80% of parameters by absolute weight), and redundant parameters below the threshold are directly removed: For CNNs, redundant convolutional kernels are deleted, or corresponding neurons are deleted from fully connected layers; for LSTMs, redundant weights in gating mechanisms are deleted, or the number of hidden layer units is reduced. After pruning, the number of parameters in the model can be reduced, and the network structure becomes more streamlined.

[0111] Pruning may disrupt the original parameter balance, leading to a temporary decrease in accuracy. By performing quantized perception training on a pre-set dataset, the problem of accuracy redundancy is solved, further reducing storage requirements and operating energy consumption. The resulting lightweight model can still maintain load identification accuracy under the limited hardware resources of the electricity meter, with low computing power, small memory, and low power consumption, achieving a balance between high accuracy and embedded deployment.

[0112] Figure 6 This is the sixth flowchart illustrating an embedded load identification method for an electricity meter, as provided in this application embodiment. Figure 6 As shown, after using a pre-set hybrid neural network model to extract the electrical appliance characteristics of power consumption parameters, and outputting the power consumption analysis results of each electrical device for different time periods based on the electrical appliance characteristics of power consumption parameters, it also includes:

[0113] S601. Generate an electricity control strategy based on the electricity consumption analysis results.

[0114] S602, Send the power control strategy to the appliance intelligent control platform.

[0115] In this embodiment, based on the power consumption analysis results output by the hybrid neural network, including the identification results of electrical equipment and the power consumption analysis data of each electrical equipment, an executable power consumption control strategy is formulated, transforming passive monitoring into active control and realizing intelligent power consumption management.

[0116] For example, if the electricity consumption analysis results show that electrical device A is on from 12:00 to 2:00 and from 18:00 to 21:00, then an electricity control strategy will be generated for electrical device A. That is, in the future, the intelligent electrical control platform will automatically control electrical device A to be on from 12:00 to 2:00 and from 18:00 to 21:00, and will not control electrical device A at other times. Similarly, the same method will be used to generate electricity control strategies for other electrical devices in the electricity consumption analysis results.

[0117] Based on the results of electricity consumption analysis, the switching status of each electrical device at different time periods is clearly identified, and an automatic control strategy is generated. This strategy can accurately match the future start-up and shutdown times of the equipment, reducing standby and ineffective operation energy consumption and lowering users' electricity costs. It can also proactively adapt to users' historical electricity consumption habits, realizing automated start-up and shutdown of electrical devices, eliminating the burden of manual operation and improving ease of use. At the same time, it can avoid disordered start-up and shutdown or overload operation of electrical devices, extending the service life of equipment, and can also help the power grid balance peak and valley loads and optimize overall energy utilization efficiency.

[0118] The following will continue to explain the embedded load identification device and the energy meter provided in any of the above embodiments of this application. The specific implementation process and the resulting technical effects are the same as those in the corresponding method embodiments. For the sake of brevity, the parts not mentioned in this embodiment can be referred to the corresponding content in the method embodiment.

[0119] Figure 7 This is a schematic diagram of the functional modules of an embedded load identification device in an electricity meter, provided as an embodiment of this application. Figure 7 As shown, the embedded load identification device 100 in the electricity meter includes:

[0120] The sampling module 110 is used to perform high-frequency sampling of the power grid bus to obtain the current and voltage parameters of the main incoming line. The current parameters include: instantaneous current value, effective current value, current phase, and current harmonic components, which are used to describe the energy consumption intensity and waveform characteristics of electrical equipment. The voltage parameters include: instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, which are used to calculate power and analyze the impact of power grid quality on the load.

[0121] Processing module 120 is used to divide the current parameters according to continuous time using a preset edge algorithm to obtain windowed power consumption parameters after division;

[0122] The extraction module 130 is used to extract the electrical appliance features of the power consumption parameters using a preset hybrid neural network model, and output the power consumption analysis results corresponding to different time periods for each electrical appliance based on the electrical appliance features of the power consumption parameters. The electrical appliance features include: power consumption information and corresponding power consumption time sequence information for different electrical appliances. The power consumption analysis results include: electrical appliance identification results and power consumption analysis data for each electrical appliance.

[0123] Optionally, the processing module 120 is further configured to use a preset edge algorithm to divide the current parameters and voltage parameters according to continuous time, obtain window current and window voltage corresponding to multiple time windows, and generate a voltage-current trajectory diagram based on the window current and the window voltage.

[0124] Optionally, the processing module 120 is also used to calculate the power consumption parameters within each time window, including: active power, reactive power, and power factor; and to generate a time-series vector corresponding to the power consumption parameters based on the power consumption parameters within each time window.

[0125] Optionally, the preset hybrid neural network model includes a convolutional neural network (CNN) model and a long short-term memory (LSTM) network model; the extraction module 130 is also used to input the voltage-current trajectory map into the CNN model, identify the voltage-current spatial feature vector through image dimension recognition, and output the voltage-current spatial feature vector; input the time-series vector corresponding to the power consumption parameters into the LSTM model, and output the power information and time-series dependency features; and identify the probability of different electrical devices being turned on or off in different time windows based on the voltage-current spatial feature vector, power information and time-series dependency features.

[0126] Optionally, the extraction module 130 is also used to fuse the voltage-current spatial feature vector, power information and time-series dependency features to obtain the fused feature vector; based on the fused feature vector, to identify different electrical devices and output the probability of different electrical devices being turned on or off in different time windows.

[0127] Optionally, the preset hybrid neural network model is trained using a CNN model, an LSTM model, and a preset dataset, wherein the preset dataset includes current data and voltage data corresponding to the operating or off states of different electrical devices.

[0128] Optionally, the hybrid neural network model is a lightweight model, wherein the lightweight processing includes: pruning the trained hybrid neural network model and then performing quantization-aware training on the pruned model.

[0129] Optionally, the device further includes:

[0130] The generation module is used to generate power control strategies based on the power consumption analysis results;

[0131] The sending module is used to send power control strategies to the intelligent appliance control platform.

[0132] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0133] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0134] Figure 8 This is a schematic diagram of an electricity meter provided in an embodiment of this application. This electricity meter can be used for embedded load identification. Figure 8 As shown, the computer device includes: a processor 210, a storage medium 220, and a bus 230.

[0135] Storage medium 220 stores machine-readable instructions executable by processor 210. When the computer device is running, processor 210 communicates with storage medium 220 via bus 230, and processor 210 executes the machine-readable instructions to perform the steps of the above method embodiment. The specific implementation and technical effects are similar, and will not be described again here.

[0136] Optionally, this application also provides a storage medium 220, on which a computer program is stored. When the computer program is run by a processor, it executes the steps of the above-described method embodiments. The specific implementation methods and technical effects are similar, and will not be described again here. The above-described apparatus is used to execute the methods provided in the foregoing embodiments, and its implementation principles and technical effects are similar, so will not be described again here.

[0137] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0138] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0139] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0140] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0141] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for embedded load identification in an electricity meter, characterized in that, Applied to electricity meters, the method includes: The power grid bus is sampled at high frequency to obtain the current and voltage parameters of the bus. The current parameters include instantaneous current value, effective current value, current phase, and current harmonic components, which are used to describe the energy consumption intensity and waveform characteristics of electrical equipment. The voltage parameters include instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, which are used to calculate power and analyze the impact of power grid quality on the load. The current parameters and voltage parameters are segmented according to continuous time using a preset edge algorithm to obtain the segmented windowed power consumption parameters; A preset hybrid neural network model is used to extract the electrical appliance features of the power consumption parameters, and output the power consumption analysis results of each electrical appliance for different time periods based on the electrical appliance features of the power consumption parameters. The electrical appliance features include: power consumption information and corresponding power consumption time sequence information for different electrical appliances. The power consumption analysis results include: electrical appliance identification results and power consumption analysis data for each electrical appliance. The step of using a preset edge algorithm to segment the current and voltage parameters according to continuous time to obtain windowed power consumption parameters after segmentation includes: The preset edge algorithm is used to divide the current parameter and the voltage parameter into continuous time segments to obtain window current and window voltage corresponding to multiple time windows; Generate a voltage-current trajectory diagram based on the window current and the window voltage; After using the preset edge algorithm to segment the current and voltage parameters according to continuous time to obtain window current and window voltage corresponding to multiple time windows, the method further includes: Calculate the power consumption parameters within each time window, including: active power, reactive power, and power factor; Generate a time-series vector corresponding to the electricity consumption parameters based on the electricity consumption parameters within each time window; The preset hybrid neural network model includes: a convolutional neural network (CNN) model and a long short-term memory (LSTM) network model; The method of extracting electrical appliance features from the electricity consumption parameters using a preset hybrid neural network model includes: The voltage-current trajectory diagram is input into the CNN model, and the voltage-current spatial feature vector is output through image dimension recognition. The time-series vector corresponding to the power consumption parameters is input into the LSTM model, and the power information and time-series dependency features are output. Based on the voltage-current spatial feature vector, the power information, and the time-series dependency characteristics, the probability of different electrical devices being turned on or off in different time windows is identified.

2. The method according to claim 1, characterized in that, The step of identifying the probability of different electrical devices being turned on or off in different time windows based on the voltage-current spatial feature vector, the power information, and the time-series dependency features includes: The voltage-current spatial feature vector, the power information, and the time-series dependency features are fused to obtain the fused feature vector; Based on the fused feature vector, different electrical devices are identified, and the probability of different electrical devices being turned on or off in different time windows is output.

3. The method according to claim 2, characterized in that, The preset hybrid neural network model is trained using the CNN model, the LSTM model, and a preset dataset, wherein the preset dataset includes current data and voltage data corresponding to different power equipment usage states or off states.

4. The method according to claim 3, characterized in that, The preset hybrid neural network model is a lightweight model, wherein the lightweight processing includes: pruning the trained hybrid neural network model and then performing quantization-aware training on the pruned model.

5. The method according to claim 1, characterized in that, After extracting the electrical appliance characteristics of the power consumption parameters using a preset hybrid neural network model, and outputting the power consumption analysis results for each electrical device at different time periods based on the electrical appliance characteristics of the power consumption parameters, the method further includes: Based on the power consumption analysis results, a power consumption control strategy is generated; Send the power control strategy to the appliance intelligent control platform.

6. A load identification device embedded in an electricity meter, characterized in that, Applied to electricity meters, the device includes: The sampling module is used to perform high-frequency sampling of the power grid bus to obtain the current and voltage parameters of the main incoming line. The current parameters include: instantaneous current value, effective current value, current phase, and current harmonic components, used to describe the energy consumption intensity and waveform characteristics of electrical equipment. The voltage parameters include: instantaneous voltage value, effective voltage value, voltage phase, and voltage fluctuation, used to calculate power and analyze the impact of power grid quality on the load. The processing module is used to segment the current parameters and the voltage parameters according to continuous time using a preset edge algorithm to obtain the segmented windowed power consumption parameters; The extraction module is used to extract the electrical appliance features of the power consumption parameters using a preset hybrid neural network model, and output the power consumption analysis results corresponding to different time periods for each electrical appliance based on the electrical appliance features of the power consumption parameters. The electrical appliance features include: power consumption information and corresponding power consumption time sequence information for different electrical appliances. The power consumption analysis results include: electrical appliance identification results and power consumption analysis data for each electrical appliance. The processing module is further configured to use the preset edge algorithm to divide the current parameter and the voltage parameter according to continuous time, and obtain the window current and window voltage corresponding to multiple time windows; and generate a voltage-current trajectory diagram based on the window current and the window voltage. The processing module is also used to calculate the power consumption parameters within each time window, the power consumption parameters including: active power, reactive power, and power factor; and to generate a time-series vector corresponding to the power consumption parameters based on the power consumption parameters within each time window; The preset hybrid neural network model includes a Convolutional Neural Network (CNN) model and a Long Short-Term Memory (LSTM) model. The extraction module is further configured to input the voltage-current trajectory map into the CNN model, identify the voltage-current spatial feature vector through image dimension recognition, and output the voltage-current spatial feature vector; input the time-series vector corresponding to the power consumption parameters into the LSTM model, and output the power information and time-series dependency features; and identify the probability of different electrical devices being turned on or off in different time windows based on the voltage-current spatial feature vector, the power information and time-series dependency features.

7. An electricity meter, characterized in that, include: The device includes a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the computer device is running, the processor communicates with the storage medium via the bus. The processor executes the program instructions to perform the steps of the embedded load identification method for an electricity meter as described in any one of claims 1 to 5.

Images