Intelligent socket integrated with electric energy recording and counting functions and operation method thereof

Abstract

The invention discloses an intelligent socket integrated with electric energy recording and counting functions and an operation method thereof, and relates to the technical field of intelligent electric appliances. The intelligent socket comprises a sensor unit, a data processing and executing unit and a storage unit. The method comprises the steps that a sensor unit collects voltage and current waveform data and generates original sampling data with a timestamp; the data processing and executing unit preprocesses and extracts features, and standardized data is generated through fusion; then, identifying load types by adopting an edge lightweight classification method, dynamically updating an energy consumption mode in combination with a federal incremental learning strategy, generating an energy consumption statistical result and an anomaly detection result, outputting an energy-saving prompt and a corresponding safety protection signal, summarizing the energy-saving prompt and the corresponding safety protection signal into a comprehensive judgment result, and executing corresponding operation; and the storage unit generates a periodic energy consumption statistical report and an operation log. According to the invention, electric energy can be accurately recorded and counted, energy-saving and safe operation can be executed in a targeted manner, and diversified power utilization management requirements of users are met.

Description

Technical Field

[0001] This invention relates to the field of intelligent electrical appliance technology, and more specifically, to an intelligent socket with integrated power recording and statistical functions and its operating method. Background Technology

[0002] With the rapid development of smart home appliances, the number and types of electrical devices are constantly increasing, and users' needs for recording, statistics, and safe management of electricity usage are rising. Currently, traditional sockets only provide basic power supply and cannot effectively record and statistically analyze electricity usage. Users find it difficult to know the energy consumption distribution of different electrical devices, which is not conducive to achieving energy-saving electricity use.

[0003] While some sockets with basic electricity metering functions can roughly estimate total electricity consumption, they lack the ability to accurately identify load types and cannot perform differentiated energy consumption analysis for different types of loads (such as resistive, inductive, and mixed loads), resulting in low practicality of energy consumption statistics. Furthermore, these sockets are deficient in updating energy consumption modes, often using fixed modes that cannot dynamically adjust based on users' actual electricity usage habits and equipment usage, making it difficult to accurately reflect real-time energy consumption.

[0004] In terms of safety protection, the anomaly detection functions of existing sockets are relatively simple, typically only able to detect serious faults such as overload or short circuit. They struggle to effectively identify potential safety hazards such as poor socket contact or abnormal power fluctuations. Furthermore, after detecting anomalies, they often only perform a simple power-off operation, lacking targeted safety protection strategies and follow-up measures, thus failing to comprehensively guarantee electrical safety. In addition, existing sockets lack a comprehensive energy-saving prompt and optimization mechanism, unable to provide users with effective energy-saving suggestions and automatic optimization operations based on energy consumption, making it difficult to meet users' dual needs for electrical safety and energy efficiency. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a smart socket with integrated power recording and statistical functions and its operating method. The following solutions address the problems of inaccurate power recording and statistics, low practicality of energy consumption analysis, insufficient identification of safety hazards, and poor energy-saving effects mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for operating a smart socket integrating power recording and statistical functions, characterized in that it includes: S1: The smart socket collects voltage and current signals through sensors, and combines them with the built-in high-speed data acquisition module and real-time clock module to generate raw sampling data with timestamps. S2: Perform digital filtering and time window segmentation on the original sampled data; then extract spectral features, harmonic components and transient disturbance information based on multi-scale time-frequency decomposition, and generate standardized feature data through an adaptive weighted fusion strategy; S3: Based on the standardized feature data, the load type is identified using the edge lightweight classification method, including the discrimination of resistive, inductive and mixed loads; an initial energy consumption pattern model is constructed and the energy consumption pattern model is improved and updated by combining the federated incremental learning strategy, and energy consumption statistics and anomaly detection results are generated; the anomaly detection results include overload, short circuit, poor socket contact and abnormal power fluctuation; S4: Based on the energy consumption statistics and anomaly detection results, compare the energy consumption level with the set threshold. When the energy consumption level exceeds the threshold, generate an energy-saving prompt. When the anomaly detection results indicate overload, short circuit, poor socket contact, or abnormal power fluctuation, generate a corresponding safety protection signal. Combine the energy-saving prompt and the safety protection signal into a comprehensive judgment result. S5: Execute operations based on the comprehensive judgment results, including energy-saving optimization operations and safety hazard elimination operations, and finally generate a periodic performance consumption statistical report and operation log; A smart socket includes a sensor unit, a data processing unit, and a storage unit. The sensor unit acquires voltage and current signals and generates raw sampling data with timestamps. The data processing unit processes the raw sampling data, generates a comprehensive judgment result, and executes corresponding operations. The storage unit generates and stores operation logs and periodic performance consumption statistics reports based on the comprehensive judgment result.

[0007] The technical effects and advantages of this invention are as follows: 1. This invention collects raw voltage and current waveform data with timestamps, and then processes it through digital filtering, time window segmentation, multi-scale time-frequency decomposition, and adaptive weighted fusion to obtain standardized characteristic data that can comprehensively reflect the power consumption status, thus achieving accurate and comprehensive power recording and statistics. 2. This invention uses a lightweight edge classification method based on standardized feature data to obtain the identification results of resistive, inductive and mixed load types, and calculates energy consumption indicators according to load type. It has the effect of accurately generating periodic energy consumption statistical reports and realizing differentiated energy consumption analysis. 3. This invention obtains efficient and accurate load identification results by using a lightweight convolutional neural network and edge-side fine-tuning. At the same time, it combines a federated incremental learning strategy to achieve dynamic model updates, which has the effect of real-time matching of energy consumption patterns with actual power consumption conditions, thereby ensuring the reliability of energy consumption statistics and anomaly detection. 4. This invention sets differentiated thresholds and corresponding safety protection signals for overload, short circuit, poor socket contact, and abnormal power fluctuations, thereby obtaining a safety protection mechanism covering multiple scenarios. It has the effects of quickly cutting off power, reducing power and prompting maintenance, thus comprehensively ensuring electrical safety. 5. This invention generates energy-saving prompts and performs local power adjustments, delayed power outages, and power consumption period optimization operations when energy consumption exceeds a threshold, resulting in significantly reduced energy consumption. It achieves the effect of improving energy efficiency while providing safety protection, thereby meeting the diverse power management needs of users. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2 This is a schematic diagram of the overall method flow of the present invention; Figure 3 This is a schematic diagram of the feature extraction process of the present invention; Figure 4 This is a schematic diagram of the load identification and anomaly detection process of the present invention. Detailed Implementation

[0009] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] As attached Figures 2 to 4 The operation method of the smart socket with integrated power recording and statistics functions shown includes: S1: The smart socket collects voltage and current signals through sensors, and generates raw sampling data with timestamps by combining the built-in high-speed data acquisition module and real-time clock module; S2: Perform digital filtering and time window segmentation on the original sampled data; then extract spectral features, harmonic components and transient disturbance information based on multi-scale time-frequency decomposition, and generate standardized feature data through an adaptive weighted fusion strategy; S3: Based on the standardized feature data, the load type is identified using the edge lightweight classification method, including the discrimination of resistive, inductive and mixed loads; an initial energy consumption pattern model is constructed and the energy consumption pattern model is improved and updated by combining the federated incremental learning strategy, and energy consumption statistics and anomaly detection results are generated; the anomaly detection results include overload, short circuit, poor socket contact and abnormal power fluctuation; S4: Based on the energy consumption statistics and anomaly detection results, compare the energy consumption level with the set threshold. When the energy consumption level exceeds the threshold, generate an energy-saving prompt. When the anomaly detection results indicate overload, short circuit, poor socket contact, or abnormal power fluctuation, generate a corresponding safety protection signal. Combine the energy-saving prompt and the safety protection signal into a comprehensive judgment result. S5: Execute operations based on the comprehensive judgment results, including energy-saving optimization operations and safety hazard elimination operations, and finally generate periodic performance consumption statistics reports and operation logs.

[0011] It should be noted that, firstly, the sensor unit collects the voltage and current waveforms at the input of the socket, forming raw sampling data with timestamps.

[0012] It should be further explained that high-precision voltage and current sensors are installed inside the smart socket to collect voltage and current signals from the socket's input terminals, respectively. The voltage sensor is a high-precision Hall effect sensor based on the Hall effect principle. This type of sensor can isolate and convert high-voltage signals into electrical signals proportional to the input voltage, featuring high accuracy, fast response speed, and good electrical isolation performance. It can effectively avoid interference to the circuit during measurement and ensure the accuracy of the collected data. The current sensor is a closed-loop Hall current sensor, which utilizes the magnetic balance principle to accurately measure DC, AC, and pulse currents, and has good linearity and stability.

[0013] It should be further explained that the sensor transmits the collected analog voltage and current signals to the high-speed data acquisition module built into the smart socket. This data acquisition module uses an ADC (analog-to-digital converter) chip with high-speed sampling capability, such as an ADC chip with a sampling rate of up to 1MSPS (millions of samples per second). It can sample analog signals at an extremely fast speed and convert them into digital signals. During the sampling process, the voltage and current signals are periodically sampled according to the set sampling frequency (such as 10kHz, i.e., 10,000 samples per second), thereby obtaining a series of discrete voltage and current data points. These data points are connected to form voltage and current waveform data.

[0014] It should be further explained that after each sampling, the current accurate time information is obtained through the real-time clock module (RTC) inside the smart socket. The real-time clock module is usually composed of a dedicated clock chip and a backup battery, which can ensure accurate time operation even when the smart socket is powered off. The obtained time information is added to the corresponding voltage and current data in the form of timestamps. Finally, the voltage and current data with timestamps together constitute the original sampling data.

[0015] Specifically, digital filtering and time-window segmentation are performed on the original sampling data; subsequently, spectral features, harmonic components, and transient disturbance information are extracted based on multi-scale time-frequency decomposition, and standardized feature data is generated through an adaptive weighted fusion strategy.

[0016] It should be further noted that the specific method of the digital filtering is to use a finite impulse response (FIR) filter. Let the original sampling data sequence be , N is the number of sampling points, and the filter coefficient sequence is , M is the filter order, and the filtered signal is calculated through . The value range of n is 0 ≤ n ≤ N - 1. When n < k, it is default that x(n - k) = 0 to ensure that there is no boundary distortion in the filtering result; the specific method of the time-window segmentation is to divide the continuous filtered signal into multiple independent data segments according to a fixed time length (such as 1 second) (each time window contains 10,000 sampling points, based on a 10 kHz sampling frequency), so that the originally continuous power data is transformed into discrete units convenient for time-segment analysis.

[0017] It should be further noted that subsequently, multi-scale time-frequency decomposition is performed on the data within each time window: using discrete wavelet transform (DWT), the signal within the time window is processed according to the formula . Among them: is the wavelet transform result, representing the wavelet coefficient of the signal at the scale a and displacement b, reflecting the characteristic intensity of the signal at a specific scale and position; a is the scale parameter, used to control the stretching of the wavelet basis function , and a > 0. When a > 1, the wavelet basis function stretches in the time domain and the frequency domain bandwidth becomes narrower, which is suitable for extracting spectral features such as the fundamental frequency when electrical equipment operates stably in the signal; when a < 1, the wavelet basis function compresses in the time domain and the frequency domain bandwidth becomes wider, which is suitable for extracting high-frequency components in the signal, such as high-order harmonic components and transient disturbance information; b is the displacement parameter, used to control the translation of the wavelet basis function on the time axis. By changing the value of b, the wavelet basis function can slide on the time axis to accurately locate the position of signal features (such as the moment when harmonics appear, the starting point of transient disturbances) in the time dimension; is the selected wavelet basis function (here the db4 wavelet basis is used), It is its conjugate function, used to perform convolution operation with the original signal to achieve time-frequency decomposition; n is the sampling point number, and N is the number of sampling points in each time window; by adjusting a and b in the above formula, the signal can be decomposed into different frequency sub-bands, and the time of occurrence of features can be located in the time domain, thereby accurately extracting spectral features (energy distribution of the signal at different frequencies, reflecting the power frequency characteristics of the equipment), harmonic components (integer multiple frequency components in the current or voltage other than the fundamental frequency, reflecting load or power quality problems) and transient disturbance information (the time and characteristics of sudden anomalies such as voltage surges and drops).

[0018] It should be further explained that standardized feature data is finally generated through adaptive weighted fusion: let the extracted feature data of each type be... , where i is the index of the feature data, and its value range is L represents the number of feature data types; the weight corresponding to each feature data type is . The weights will be automatically adjusted based on the importance of different feature data in reflecting the working status of electrical equipment and power quality, and will meet the following requirements. Through formula Multi-dimensional feature data is fused into a set of standardized feature data F.

[0019] It should be specifically noted that, based on the standardized feature data, a lightweight edge classification method is used to identify the load type, including the distinction between resistive, inductive, and mixed loads; at the same time, a federated incremental learning strategy is combined to dynamically update the energy consumption pattern model, generating energy consumption statistics and anomaly detection results; the anomaly detection results include overload, short circuit, poor socket contact, and abnormal power fluctuations.

[0020] It should be further explained that, based on the standardized feature data, a lightweight edge classification method is used to identify the load type. Specifically, a lightweight convolutional neural network (CNN) is used as the classification model. The model input is the standardized feature data F (dimension 1×L, where L is the feature dimension), and classification is achieved through 3 convolutional layers, 2 pooling layers, and 1 fully connected layer. The convolutional layer operation formula is: in, On the output feature map of the convolutional layer The pixel value of the location, To input the feature values ​​at the corresponding positions of the standardized feature data, The weight parameters are for the 3×3 convolution kernel. As a bias term, unique patterns in the spectral characteristics of different load types are extracted through convolution operations (such as the fundamental frequency proportion characteristics of resistive loads and the harmonic characteristics of inductive loads); the pooling layer uses max pooling, with the formula: in The pooling layer outputs values ​​by selecting the maximum value within a 2×2 region to reduce data dimensionality and computational cost; the fully connected layer outputs the predicted probability of the load type. ,in Let be the predicted probability for load type t (t=1 represents resistive, t=2 represents inductive, and t=3 represents hybrid), and W be the weight matrix of the fully connected layer. This is the flattened vector output by the pooling layer. As the bias term for the corresponding category, the category with the highest probability is taken as the final discrimination result, and the load type label T is output.

[0021] It should be further explained that the specific method of model pre-training is to construct a training set using a public load dataset (containing standardized feature data of resistive / inductive / mixed loads) and 1000 sets of actual collected load data of the three classes. The pre-training is completed through 50 iterations of the Adam optimizer, with an initial classification accuracy of ≥95%. After deployment, edge-side fine-tuning is supported.

[0022] It should be further explained that the energy consumption pattern model is then dynamically updated using a federated incremental learning strategy to generate energy consumption statistics and anomaly detection results. Specifically, the smart socket uses historical standardized feature data locally to apply an autoregressive integral moving average (ARIMA) model to different load types T. Initial energy consumption model is constructed, where p is the autoregressive order, d is the differencing order, and q is the moving average order. p is determined by the PACF (partial autocorrelation function) truncation property (e.g., p=2 if PACF is trunculated at lag 2); d is determined by the ADF stationarity test (the minimum differencing order that makes the data stationary, usually d=0 or 1); q is determined by the ACF (autocorrelation function) truncation property (e.g., q=1 if ACF is trunculated at lag 1). The model formula is as follows: ,in Let be the predicted energy consumption at time u, and c be a constant term. These are the autoregressive coefficients. for Historical energy consumption at any given time The moving average coefficient, for Error term at time; Federated Incremental Learning Update: Multiple smart sockets act as edge nodes, extracting features from newly collected standardized feature data based on the identified load type T, forming feature vectors that include time information, load type label T, power change trends, and other information. The energy consumption model is updated using a weighted aggregation method, and the calculation formula is as follows: in, These are the global model parameters corresponding to load type T; It is the weight of the k-th edge node. The weight is determined by the amount and quality of historical data of the node, specifically... ,in Let k be the amount of historical data for the k-th node. Assign a data quality score (0-1, calculated based on data integrity and noise rate). Where data integrity = effective data volume / total data volume, and data noise rate = noisy data points / total data points), α = 0.6 and β = 0.4 are weighting coefficients; These are the local model parameters for the k-th node specific to load type T; Energy consumption statistics are generated based on the updated energy consumption pattern model, calculating energy consumption indicators by time dimensions such as hour, day, and month, combined with load type T. For example, the daily average power consumption is calculated as follows: in, The number of time windows in a day; Let T be the power value of load type T in the s-th time window. Through this calculation, the daily average power and total power consumption of resistive, inductive, and mixed loads can be obtained, forming energy consumption statistics results for each load type. Anomaly detection result generation: Anomaly detection is achieved by comparing real-time standardized feature data. Energy consumption model prediction values (Based on the initial energy consumption pattern model, after optimization by federated incremental learning, the output is a set of multi-dimensional energy consumption and electrical parameter prediction values ​​that match the dimensions of real-time standardized feature data and are used for anomaly detection comparison), and a differentiated threshold is set in conjunction with the load type T for judgment: Overload detection: For resistive loads (T=1), an overload current threshold is set. ( (Rated power, U is rated voltage), inductive load (T=2), considering starting impact, threshold is set to... ( The power factor is calculated in real time by the phase difference between voltage and current. For inductive loads, the default value is... If the real-time detected value deviates from this range, the threshold calculation result will be automatically corrected. Exceeding the corresponding threshold and the duration exceeds (e.g., 500 milliseconds) is considered an overload; Short circuit detection: when current is detected... The voltage instantly increases to more than three times the rated current. A sudden drop to below 50% of the rated voltage, and the duration exceeding [a certain threshold]. (e.g., 100 milliseconds) is considered a short circuit; Socket contact failure detection: Based on historical data, a model of the relationship between contact resistance and power fluctuation is trained. When power fluctuates frequently with small amplitudes, and the calculated contact resistance... ,in (Contact point voltage drop, mV), I (load current, A), if Greater than the normal threshold ,like The issue was determined to be poor socket contact; power fluctuation detection: for resistive loads, the allowable power fluctuation range is set to ±10% of the average power, and for mixed loads, it is set to ±30%. When the real-time power... Deviation from average power The amplitude exceeds the corresponding range and the duration exceeds (e.g., 200 milliseconds) is judged as abnormal power fluctuation; finally, abnormal detection results covering overload, short circuit, poor socket contact and abnormal power fluctuation are generated, along with energy consumption statistics results for different load types.

[0023] It should be specifically explained that, based on the energy consumption statistics and anomaly detection results, the energy consumption level is compared with a set threshold. When the energy consumption level exceeds the threshold, an energy-saving prompt is generated. When the anomaly detection results indicate overload, short circuit, poor socket contact, or abnormal power fluctuation, a corresponding safety protection signal is generated. The energy-saving prompt and the safety protection signal are combined into a comprehensive judgment result.

[0024] It should be further explained that the smart socket has preset energy consumption thresholds for different scenarios. These thresholds are based on historical electricity consumption data statistical analysis and user-defined settings. For example, the daily energy consumption threshold is calculated by analyzing the average daily electricity consumption over the past 30 days and taking 1.2 times the average as the daily energy consumption threshold. Monthly energy consumption threshold: Set based on historical average monthly electricity consumption, and adjusted according to factors such as season and electricity consumption habits. Then, the actual energy consumption data in the energy consumption statistics are compared with the corresponding threshold: Daily energy consumption judgment: If the total daily energy consumption... Greater than the daily energy consumption threshold ,Right now If the daily energy consumption level is too high, then the monthly energy consumption level is determined by the cumulative monthly energy consumption. Exceeding the monthly energy consumption threshold If the energy consumption level is too high, then the energy consumption level for that month is determined to be too high.

[0025] It should be further explained that the smart socket performs real-time analysis of the anomaly detection results and generates corresponding safety protection signals according to different anomaly types. Specifically, this includes: Overload anomaly handling: When an overload anomaly is detected, an "overload protection signal" is generated. This signal triggers the smart socket to perform the following operations: Firstly, it cuts off the power to the sockets containing non-critical devices (only when the overload condition lasts for more than 500 milliseconds will the power to the non-critical device sockets be cut off to avoid accidental operation due to instantaneous impact), and pushes an emergency notification including the time of the overload, the affected devices, and the current handling measures; Short circuit anomaly handling: Once a short circuit anomaly is detected, a "short circuit emergency power-off signal" is generated to cut off the main power supply to the smart socket as quickly as possible to prevent the circuit fault from escalating, and simultaneously... The system sends an alarm to remind users not to attempt repairs themselves but to contact a professional electrician. For socket contact issues, if a poor connection is detected, a "poor contact warning signal" is generated. This reduces the output power of the corresponding socket to prevent overheating due to excessive contact resistance. It also displays the location of the faulty socket and provides repair suggestions (such as checking for loose plugs and cleaning the socket's metal contacts). After repair, the contact resistance must be verified to be ≤0.5Ω using the resistance detection function. For abnormal power fluctuations, a "fluctuation monitoring signal" is generated to continuously monitor equipment power changes. If the fluctuation persists and intensifies, the equipment power is gradually reduced or the power supply is cut off. Detailed data on the abnormal power fluctuation and possible causes are then sent to the user.

[0026] It should be further explained that the generated energy-saving prompts and safety protection signals will then be summarized to form a comprehensive judgment result, which includes: if only excessive energy consumption exists, the comprehensive judgment result is "energy-saving optimization is needed", with detailed energy-saving prompt information; if only an abnormal situation is detected, the comprehensive judgment result is "there is a safety hazard", including the type of abnormality, handling measures, and the operation that the user needs to cooperate with; if both excessive energy consumption and abnormal situation occur, the comprehensive judgment result will prioritize "safety hazard", with additional energy-saving prompts to ensure that users pay attention to safety issues as soon as possible and understand the direction of energy saving; when multiple abnormalities occur at the same time, the operation will be performed in the order of 'short circuit > overload > poor contact > power fluctuation'.

[0027] It should be noted that, based on the comprehensive judgment results, the final operations are performed, including energy-saving optimization operations and safety hazard elimination operations, and finally, a periodic performance consumption statistical report and operation log are generated.

[0028] It should be further explained that the smart socket triggers corresponding execution mechanisms based on the type and priority of the comprehensive judgment results. Specifically, these include: operations for the "energy-saving optimization required" result: local power adjustment, automatically adjusting the power of sockets containing high-energy-consuming devices (such as electric water heaters and air conditioners) according to preset energy-saving modes; delayed power-off management, triggering delayed power-off for idle devices (such as fully charged mobile phones and unused desk lamps) when the power is detected to be below 5W for 30 consecutive minutes, automatically cutting off the power to the corresponding socket after a 10-minute countdown with audio and visual prompts; and power consumption period optimization, identifying peak power consumption periods (such as 18:00-22:00) based on historical data, and pushing time period adjustment suggestions to devices that can operate off-peak (such as washing machines), which are automatically set to start during off-peak periods (such as 0:00-6:00) after user confirmation. For the result of "potential safety hazard", the following actions are taken: Local power cut-off control is implemented. For a single socket identified as having poor contact, power is immediately cut off and locked, displaying a "fault locked" indicator. The user must manually unlock and repair the socket before it can be reused. Overall power scheduling is implemented. When a short circuit or severe overload is detected, the main power cut-off mechanism is triggered, disconnecting the main circuit via a relay and storing data such as the time of the anomaly, current, and voltage waveforms locally. Simultaneously, an "emergency power outage" alarm and fault location information are pushed out. Hierarchical protection linkage is implemented. If abnormal power fluctuations are detected, the socket containing the fluctuation source is first cut off. If the line abnormality is still detected within 3 seconds, the action is expanded to cut off two adjacent sockets (adjacent sockets refer to physically adjacent sockets located on the same circuit branch as the fluctuation source socket), until the abnormality is eliminated.

[0029] It should be further explained that the smart socket automatically generates energy consumption statistics reports according to preset cycles (daily, monthly), specifically including: Daily report: summarizing the total electricity consumption of the day, the energy consumption ratio of each load type (resistive, inductive, and hybrid), peak electricity consumption periods, and the effectiveness of energy-saving optimization measures; Monthly report: statistics on the total energy consumption of the month, the rate of change compared with the previous month's energy consumption, the energy consumption ranking of each device, and analysis of energy-saving results in conjunction with the implementation of energy-saving strategies; The smart socket records all operation behaviors and generates operation logs, specifically including: real-time recording of all operation behaviors of the smart socket, including the execution time of energy-saving optimization measures, the handling process of safety hazards, the report generation time, and other information; Finally, the energy consumption statistics reports and operation logs are stored in the smart socket's storage unit.

[0030] The present invention also provides a smart socket for implementing the above-described method, such as... Figure 1 As shown, the smart socket includes: a sensor unit, a data processing and execution unit, and a storage unit.

[0031] The sensor unit is the core module for the smart socket to acquire basic power consumption data. It mainly uses a voltage sensor based on the Hall effect principle and a closed-loop Hall current sensor to collect voltage signals (converted into proportional electrical signals) and current signals (covering AC, DC, and pulse current) at the socket input terminal, respectively. At the same time, it works with a high-speed data acquisition module to periodically sample the above signals, obtain discrete data points, and form voltage and current waveform data. Then, a real-time clock module (including a dedicated clock chip and a backup battery to ensure accurate time after power failure) adds timestamps to the waveform data, finally generating original sampled data with timestamps, providing a complete basic data source for subsequent data processing.

[0032] The data processing and execution unit is responsible for the full-process analysis and judgment of the raw sampling data output by the sensor unit. First, it generates standardized feature data through digital filtering, time window segmentation (dividing data segments at fixed intervals), multi-scale time-frequency decomposition (extracting spectral features, harmonic components, and transient disturbance information based on discrete wavelet transform), and adaptive weighted fusion (setting weights for various features and combining them). Then, based on the standardized feature data, it uses a lightweight edge classification method (such as a convolutional neural network model) to identify resistive, inductive, and mixed loads, and updates the energy consumption pattern model in combination with a federated incremental learning strategy. Then, based on the updated model, it calculates the energy consumption statistics results at different time dimensions, identifies anomalies such as overload and short circuit by comparing real-time data with model predictions, and finally summarizes the comparison results of energy consumption level and threshold (energy saving prompts) and anomaly detection results (safety protection signals) to form a comprehensive judgment result.

[0033] The data processing and execution unit undertakes the core computation and operation execution tasks of the smart socket: First, it preprocesses the raw sampled data output by the sensor unit, including digital filtering through a finite impulse response (FIR) filter (to avoid boundary distortion), time window segmentation at fixed time intervals (forming data segments containing set sampling points), multi-scale time-frequency decomposition based on discrete wavelet transform (extracting spectral features, harmonic components, and transient disturbance information), and generating standardized feature data through an adaptive weighted fusion strategy; then, it uses a lightweight edge classifier built with a convolutional neural network to identify resistive, inductive, and mixed loads, updates the energy consumption pattern model with a federated incremental learning strategy, calculates the power and energy consumption data of each load type in different time dimensions based on the model (generating energy consumption statistics), and identifies anomalies such as overload and short circuit by comparing real-time data with model predictions (generating anomaly detection results); finally, it integrates energy consumption statistics and anomaly detection results to generate energy-saving prompts (such as over-threshold reminders) and safety protection signals, and implements energy-saving optimization operations such as local power adjustment and delayed power-off through execution units such as relay components, as well as operations to eliminate safety hazards such as single socket disconnection and overall power scheduling.

[0034] The storage unit is responsible for recording and retaining key data during the operation of the smart socket. It mainly stores two types of core content: first, operation logs, including detailed execution information of energy-saving optimization operations (such as local power adjustment time and delayed power-off settings) and safety protection operations (such as reasons for disconnection of individual sockets and total power scheduling records); second, energy consumption statistics, covering periodic energy consumption statistics reports such as total energy consumption data generated by time dimensions such as hours, days, and months, energy consumption ratio of various loads, and energy consumption ranking of equipment. At the same time, in order to support the calculation of the data processing unit, the storage unit will also temporarily store intermediate data such as raw sampling data and standardized feature data to ensure the continuity and traceability of the data processing process.

[0035] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. The method for operating the smart socket integrated with the electric energy recording and statistical functions, characterized in that, The method comprises the following steps: S1: The smart socket collects voltage and current signals through sensors, combines a built-in high-speed data acquisition module and a real-time clock module, and generates original sampling data with time stamps; S2: The original sampling data is subjected to digital filtering and time window segmentation; then, frequency spectrum features, harmonic components and transient disturbance information are extracted based on multi-scale time-frequency decomposition, and standardized feature data is generated through an adaptive weighted fusion strategy; S3: Based on the standardized feature data, an edge lightweight classification method is used to identify the load type, including the discrimination of resistive, inductive and mixed loads; an initial energy consumption mode model is constructed, and the energy consumption mode model is improved and updated in combination with a federal incremental learning strategy, and energy consumption statistical results and abnormal detection results are generated; the abnormal detection results include overload, short circuit, poor socket contact and abnormal power fluctuation; S4: Based on the energy consumption statistical results and abnormal detection results, the energy consumption level is compared with the set threshold value, and an energy-saving prompt is generated when the energy consumption level exceeds the threshold value; When the abnormal detection results indicate that there is overload, short circuit, poor socket contact or abnormal power fluctuation, a corresponding safety protection signal is generated; The energy-saving prompt and the safety protection signal are summarized as a comprehensive judgment result; S5: According to the comprehensive judgment result, the operation is executed, including energy-saving optimization operation and safety hazard elimination operation, and finally a periodic performance consumption statistical report and operation log are generated.

2. The method of claim 1, wherein the method further comprises: The sensor comprises a voltage sensor and a current sensor; the voltage sensor is a Hall principle voltage sensor based on the Hall effect principle, which is used to convert the input voltage signal into a proportional electrical signal; the current sensor is a closed-loop Hall current sensor, which is used to collect alternating current, direct current and pulse current; the high-speed data acquisition module uses a digital-to-analog conversion chip to periodically sample the voltage and current signals to obtain discrete data points, and the discrete data points are connected to form voltage and current waveform data; The real-time clock module adds the obtained time information to the corresponding voltage and current waveform data in the form of a time stamp to obtain voltage and current waveform data with time stamps, and constitutes the final original sampling data.

3. The method of claim 1, wherein the method further comprises: The time window segmentation method divides the continuous original sampling data into multiple data segments according to a fixed time interval, and each data segment contains a certain number of sampling points; the multi-scale time-frequency decomposition method is based on discrete wavelet transform to decompose the data segments and extract fundamental frequency spectrum features, harmonic components and transient disturbance information.

4. The method of claim 1, wherein the smart socket operation method integrated with the electric energy recording and statistical functions is characterized in that: The adaptive weighted fusion strategy sets weight parameters for the frequency spectrum features, harmonic components and transient disturbance information, and combines the features by weighting according to the weight parameters in the fusion process, and finally generates unified standardized feature data.

5. The method of claim 1, wherein the method further comprises: The edge lightweight classification method uses a convolutional neural network model to construct a classifier, inputs the standardized feature data and outputs the category label of resistive load, inductive load or mixed load.

6. The method of claim 1, wherein the smart socket operation method integrated with the electric energy recording and statistical functions is characterized by: The improved updating method of the energy consumption mode model is to introduce a federal incremental learning strategy, incrementally train locally at each smart socket based on the standardized feature data, generate improved local model parameters, and aggregate the local model parameters to form an improved energy consumption mode model.

7. The method of claim 1, wherein the method further comprises: The generation method of the energy consumption statistical result is to calculate the power and energy consumption data of each load type in different time dimensions based on the energy consumption mode model to form the energy consumption statistical result; the abnormality detection method is to compare real-time standardized feature data with predicted values of the energy consumption mode, and generate corresponding abnormal types according to different detection conditions.

8. The method of claim 1, wherein the smart socket operation method integrated with the electric energy recording and statistical functions is characterized by: The comprehensive judgment result is formed by the energy consumption statistical result and the abnormality detection result, the energy-saving optimization operation includes local power adjustment, delayed power-off management and power consumption time period adjustment, the safety hazard operation includes single socket cut-off, total power supply scheduling and hierarchical linkage cut-off, the periodic performance energy consumption statistical report includes total energy consumption, energy consumption proportion of each load type and equipment energy consumption ranking, and the operation log includes execution records of energy-saving operation and safety operation.

9. A smart socket, characterized in that: The sensor unit, the data processing unit and the storage unit are included; the function of the sensor unit is to acquire voltage signals and current signals and generate original sampling data with time stamps; the function of the data processing unit is to process the original sampling data, generate a comprehensive judgment result, and execute corresponding operations; The function of the storage unit is to generate and store operation logs and periodic performance energy consumption statistical reports according to the comprehensive judgment result.

Images