Intelligent sensing socket based on equipment feature recognition and energy consumption management method thereof

Abstract

The invention discloses an intelligent sensing socket based on equipment feature recognition and an energy consumption management method thereof. The method comprises the steps that a sensing layer collects various feature parameters and environment data of connecting equipment in an omnibearing mode; the control layer is used for deeply processing and analyzing data acquired by the sensing layer and executing equipment identification, strategy management and communication coordination; the platform layer provides far-end service support and comprises an equipment management module, a data analysis module, a user interaction module, an energy consumption management module and a core algorithm module; the original data collected by the sensing layer is subjected to feature extraction and identification analysis of the control layer, and the result is uploaded to the platform layer. According to the invention, through a multi-modal feature fusion identification technology, extremely high identification accuracy can be maintained in various use scenes; the potential risk can be identified before the fault occurs through the trend analysis of the electrical parameters and the temperature change.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of socket, in particular to an intelligent sensing socket based on device feature recognition and an energy consumption management method thereof. BACKGROUND

[0002] With the popularization of Internet of Things technology and smart home concept, intelligent socket as the basic node of home energy management and device control has been widely used. However, the existing recognition type intelligent socket mainly adopts single identification technology, such as radio frequency identification or simple power detection, which has low identification accuracy, poor compatibility, needs to modify the appliance plug, and poor user experience.

[0003] The energy management function of the existing intelligent socket mostly stays at the level of simple timing switch and power consumption statistics, lacks intelligent decision-making ability based on device characteristics and use scenarios, and cannot realize fine energy management.

[0004] The safety protection function of the traditional intelligent socket mainly depends on basic electrical protection elements, which belongs to passive protection and cannot realize fault prediction and preventive protection. Most intelligent sockets adopt a closed system architecture, lack unified device interface and communication protocol standard, and are difficult to realize interconnection and function extension. SUMMARY

[0005] Therefore, the present application provides an intelligent sensing socket based on device feature recognition and an energy consumption management method thereof to solve the problems in the prior art.

[0006] In order to achieve the above purpose, the present application provides the following technical scheme:

[0007] An intelligent sensing socket based on device feature recognition, comprising a base, a shell, a socket, an internal metal conductor, a power cord, a mechanical switch and an overload protector, in addition, the intelligent sensing socket further comprises a sensing layer and a control layer;

[0008] The sensing layer collects various characteristic parameters and environmental data of the connected device in all directions; the sensing layer comprises a radio frequency identification module, an electric energy metering module, an environmental sensing module and a temperature monitoring module;

[0009] The control layer performs deep processing and analysis on the data collected by the sensing layer, and executes device recognition, strategy management and communication coordination; the control layer comprises an MCU main control module, a device recognition engine, a storage module and a communication module;

[0010] After the MCU main control module of the control layer receives multi-source data from the sensing layer, it performs data preprocessing and feature alignment; the device recognition engine starts multi-modal feature analysis in sequence, and performs fusion processing on the radio frequency identification, electrical characteristics and environmental parameters; the communication module uploads the recognition result and device state in real time, and receives the control strategy issued.

[0011] The intelligent sensing socket can interact with a platform layer arranged in the cloud; the platform layer provides remote service support, and the platform layer comprises a device management module, a data analysis module, a user interaction module, an energy consumption management module and a core algorithm module;

[0012] The device management module receives device state data from the control layer, and maintains the full life cycle information of the device; the core algorithm module continuously optimizes the device recognition model and the energy consumption optimization algorithm based on the data analysis result, and updates the algorithm model to the control layer through the communication module; the user interaction module provides intuitive system state display and convenient control interface for the user, and returns the user feedback and behavior data to the data analysis module;

[0013] The original data collected by the sensing layer is subjected to feature extraction and recognition analysis by the control layer, and the result is uploaded to the platform layer to update the device state; the data analysis module of the platform layer generates an optimization strategy based on historical data, processes and forms executable control instructions through the core algorithm module, and sends the control instructions to the control layer; the sensing layer monitors the execution effect after the control layer executes the instructions, and the data is fed back to the platform layer through the control layer.

[0014] Further, the radio frequency identification module uses high-frequency RFID technology to read the electronic tag of the electric appliance plug; the preset information of the device, including the basic data such as device model, rated power, energy efficiency level and manufacturer, can be obtained; multi-tag simultaneous identification and anti-collision mechanism are supported to ensure accurate identification when multiple devices are connected;

[0015] The electric energy metering module accurately measures the electrical parameters and energy consumption data of the device, integrates a high-precision electric energy metering chip, and monitors key electrical parameters in real time; high-speed sampling technology is used to capture the electrical characteristics of each state of the device; a harmonic analysis function is provided to detect the current waveform distortion of the device and identify the nonlinear load characteristics;

[0016] The environmental sensing module collects related parameters of the device operating environment, which integrates various environmental sensors, including temperature and humidity sensors, light sensors and human infrared sensors; the environmental sensing module monitors the temperature and humidity changes of the device operating environment; the light intensity detection provides environmental context for the intelligent control of light-sensitive devices; the human sensing function is used to determine the presence of personnel, and intelligent energy-saving control based on personnel activity is realized.

[0017] Furthermore, the temperature monitoring module is equipped with multiple high-precision temperature sensors at key locations in the socket, including the power input terminal, output terminal, and socket contact point; it monitors the temperature rise caused by the plug contact resistance in real time, enabling early detection of potential safety hazards; it establishes a device operating temperature model and analyzes the device's health status through temperature change trends; and it provides overheat warning and protection functions to prevent safety accidents caused by overheating.

[0018] Specifically, relative temperature rise The calculation formula is:

[0019] ;

[0020] in, This is the measured temperature; The ambient reference temperature;

[0021] The formula for calculating moving average temperature is:

[0022] ;

[0023] Where n is the size of the moving window; Historical temperature data;

[0024] The formula for estimating contact resistance is:

[0025] ;

[0026] in, L is the material resistivity; A is the contact length; and A is the contact area.

[0027] The formula for calculating the rate of temperature change is:

[0028] ;

[0029] in, k is the time interval; k is the heating rate; The temperature at the current moment; Indicates the temperature at the previous moment;

[0030] The overheat warning conditions are:

[0031] ; ;

[0032] in, This is the temperature rise threshold; For duration; t hold This is the minimum trigger duration.

[0033] Furthermore: the temperature monitoring module monitors the temperature change ΔT(t) and the moving average trend MA in real time. n(t) and the rate of change k, combined with the contact resistance model R contact , to realize the judgment of the device health state and the overheating warning; only when the temperature rise exceeds the threshold and lasts for a sufficient time , the protection is triggered to avoid false positives.

[0034] Further: the device recognition engine can intelligently recognize devices based on multi-modal features; it can analyze the fusion of radio frequency features, electrical features and behavior features collected by the perception layer; it can establish a device feature database, support incremental learning and model self-optimization; it can provide device recognition confidence assessment and start an interactive confirmation mechanism when the confidence is low; in addition, it also supports learning and feature modeling of new devices, continuously expanding the recognition capability;

[0035] Specifically, the Z-score standardization formula is:

[0036] ;

[0037] Where, is the original feature; is the mean; is the standard deviation;

[0038] The calculation formula of weighted feature fusion is: ;

[0039] Where, w1, w2, w3 all represent weights, which are determined according to experiments; w1+w2+w3=1; F fused is the fusion feature vector; F RFID is the radio frequency feature; F elec is the electrical feature; F env is the environmental feature;

[0040] The classification confidence is:

[0041] ;

[0042] Where, W is the weight matrix; b is the bias term; F fused is the fusion feature vector;

[0043] The incremental learning loss function is: ;

[0044] Where, is the loss function of the new task; is the forgetting factor; is the loss function of the old task; is the loss function of the new sample.

[0045] Further, the storage module provides data storage and management functions; adopts a multi-level storage architecture, including running memory, Flash storage, and extensible external storage; stores device feature templates, identification models, energy consumption strategies, and system configuration parameters; records device operation history data, energy consumption statistical information, and security event logs.

[0046] Further, the data analysis module can provide big data analysis and deep insight functions; aggregate multi-device operation data for cross-device, cross-user statistical analysis; implement power consumption mode identification and load characteristic analysis to provide power consumption behavior profiling; perform energy efficiency evaluation and energy saving potential analysis to generate personalized energy saving recommendations; establish a security risk assessment model to realize quantitative assessment of security posture; provide data visualization display to convert complex data into intuitive charts and reports.

[0047] Further, the energy consumption management module implements intelligent energy consumption management and optimization control; establishes intelligent control strategies based on device type, usage habits, and electricity price strategies; provides multiple control methods; implements demand response functions to support participation in grid peak shaving and demand side management; performs energy saving effect evaluation and optimization to continuously improve energy consumption management strategies; supports multi-objective optimization to balance user comfort, energy saving effect, and device life;

[0048] Specifically, the total electricity cost calculation formula is:

[0049] ;

[0050] Wherein, is the time-of-use electricity price; is the controllable load power; T is the statistical period;

[0051] The energy saving rate formula is:

[0052] ;

[0053] Wherein, E base is the baseline electricity consumption; E opt is the optimized electricity consumption;

[0054] The power factor correction formula is:

[0055] ;

[0056] Wherein, PF is the power factor; P real is the actual active power; S is the apparent power; V rms is the voltage effective value; I rms is the current effective value; is the phase angle cosine value;

[0057] Harmonic control target: THD represents total harmonic distortion; is a harmonic control target threshold.

[0058] Further: the core algorithm module is based on a federated learning framework, trains a lightweight model on a local MCU, and only uploads gradient updates; the federated learning update formula is: ;

[0059] wherein, is a global model parameter; is the gradient of the global loss function; f is the global loss function; is a learning rate.

[0060] Further: the radio frequency identification module is located inside the socket close to the socket position; the electric energy metering module is integrated on the socket circuit board; the environment perception module is arranged on the surface or inside the socket shell and is used for collecting environmental data such as temperature and humidity, illumination and the like to assist in equipment use scene judgment; the temperature monitoring module is installed at the plug-in point or the key part of the circuit; the MCU main control module is the core processing unit of the intelligent socket and is embedded in the circuit board, responsible for coordinating the work of each module, executing identification algorithms and control logic; the storage module is connected with the MCU; and the communication module is integrated on the circuit board.

[0061] In order to achieve the above purpose, the application further provides an energy consumption management method of an intelligent sensing socket based on equipment feature recognition, comprising the following steps:

[0062] S1: equipment access and feature collection; when an electrical equipment is inserted into the intelligent socket, a multi-dimensional feature collection process is started, including electrical parameter collection, radio frequency identification, working state recording and environmental parameter collection;

[0063] S2: multi-modal feature extraction and equipment identification; based on the collected raw data, multi-level feature extraction and equipment identification are performed, including time domain feature extraction, frequency domain feature analysis, transient behavior modeling and pattern recognition matching;

[0064] S3: equipment strategy matching and individualized setting; according to the equipment identification result, an exclusive energy consumption management strategy is automatically matched or generated, including strategy library retrieval, strategy adaptive adjustment, new equipment learning and strategy conflict resolution;

[0065] S4: real-time energy consumption monitoring and data analysis; entering a continuous monitoring state, the equipment energy consumption is tracked and analyzed in all directions, including multi-granularity data collection, energy efficiency state evaluation, load characteristic analysis and power consumption mode identification;

[0066] S5: Intelligent strategy generation and optimization; based on real-time monitoring data and historical behavior analysis, dynamically generate and optimize energy consumption management strategies, including habit learning and modeling, multi-objective optimization, scene strategy adaptation and strategy effect prediction;

[0067] Step S6: Fine-grained power control execution; according to the generated optimization strategy, execute specific power control operations, including working parameter adjustment, running time optimization, standby power consumption management and load collaborative control;

[0068] Step S7: Security monitoring and anomaly protection; execute security monitoring functions in parallel to ensure power safety, including multi-parameter safety monitoring, anomaly pattern recognition, risk level assessment and graded protection response;

[0069] Step S8: Energy efficiency evaluation and strategy feedback; evaluate the effect of energy consumption management regularly to form a closed-loop optimization, including energy saving effect quantification, user influence evaluation, strategy parameter tuning and abnormal strategy revision;

[0070] Step S9: Data reporting and cloud collaboration; keep data synchronization and strategy collaboration with the cloud platform, including running data compression upload, strategy library synchronization update, group intelligence learning and remote diagnosis and maintenance.

[0071] The present application has the following advantages: the present application can maintain high recognition accuracy in various use scenarios through multi-modal feature fusion recognition technology; a predictive safety protection system based on data driving is established, which can identify potential risks before failure by analyzing the trend of electrical parameters and temperature changes.

[0072] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0073] In order to more intuitively illustrate the prior art and the present application, the following exemplary drawings are given. It should be understood that the specific shapes, structures shown in the drawings should not be regarded as limiting conditions in the implementation of the present application; for example, based on the technical concept disclosed in the present application and the exemplary drawings, those skilled in the art can easily make routine adjustments or further optimization to some units (components) such as increase / decrease / attribute division, specific shape, positional relationship, connection mode, size ratio relationship, etc.

[0074] Figure 1 A kind of intelligent perception socket system architecture based on equipment feature identification provided for the embodiment of the present application.

[0075] Figure 2 A flow chart of the energy consumption management method of the intelligent perception socket based on equipment feature identification of the present application.

[0076] Figure 3 An implementation flowchart of an energy consumption management method of an intelligent sensing socket based on device feature recognition. DETAILED DESCRIPTION

[0077] The embodiments of the present application will be described in detail by specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the disclosed content. Obviously, the described embodiments are part of the embodiments of the present application, not all. It should be understood that these embodiments are only for further illustration of the present application, and cannot be understood as a limitation on the scope of protection of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0078] Please refer to Figure 1 An intelligent sensing socket based on device feature recognition includes a basic structure, a sensing layer, a control layer and a platform layer.

[0079] The basic structure usually includes: base, shell, socket, internal metal conductor (copper sheet), power line, mechanical switch and overload protector (such as fuse or circuit breaker).

[0080] In addition, the base of the socket is also provided with a circuit board and a plurality of electrical element modules; these electrical element modules belong to the sensing layer and the control layer according to function; the specific structural relationship is as follows:

[0081] The radio frequency identification module is located inside the socket near the socket position, which is used to read the electronic tag information of the access device, and realizes the device identity recognition;

[0082] The electric energy metering module is integrated on the socket circuit board, which monitors the electrical parameters such as current, voltage and power in real time, and realizes the energy consumption statistics and analysis of the electrical equipment;

[0083] The environmental sensing module is usually set on the surface or inside the socket shell, which is used to collect environmental data such as temperature, humidity and light, and assist in device use scene judgment;

[0084] The temperature monitoring module is installed at the plug-in point or the key part of the circuit, which is used to monitor the temperature change inside the socket and prevent the risk of overheating;

[0085] The MCU master module is the core processing unit of the intelligent socket, which is embedded in the circuit board, responsible for coordinating the work of each module, executing identification algorithm and control logic;

[0086] The storage module is connected with the MCU, which is used to store device feature template, running data, identification model, etc.

[0087] The communication module integrated on the circuit board supports communication protocols such as Wi-Fi, Bluetooth, Zigbee, etc., and realizes data interaction with the cloud platform or user terminal.

[0088] These modules are not independent of the socket structure, but are highly integrated into the internal circuit and shell design of the socket, forming a complete intelligent sensing and control system; these modules expand the electrical functions of traditional sockets, enabling them to transform from passive power interfaces to intelligent nodes with sensing, identification, communication, and control capabilities, achieving fine management and energy efficiency optimization of electrical equipment.

[0089] The sensing layer is used to collect various characteristic parameters and environmental data of connected devices; it includes a radio frequency identification module, an electric energy metering module, an environmental sensing module, and a temperature monitoring module.

[0090] The radio frequency identification module can read the electronic tag of the device plug (the electronic tag on the plug generally uses an RFID tag; for plugs with built-in electronic tags, information can be read directly; for plugs without built-in electronic tags, electronic tags can be attached or fixed to the outer surface of the plug; when the device is inserted into the socket, the tag on the plug enters the working area of the card reader and is identified), obtaining the basic identity information of the device; at the same time, the electric energy metering module starts high-precision sampling to capture the electrical characteristic changes when the device is connected, recording key parameters such as starting current and power waveform; the environmental sensing module synchronously collects environmental data such as temperature, humidity, and light intensity; the temperature monitoring module monitors the temperature state of the socket and connected devices in real time, monitoring the temperature rise at the plug-in point.

[0091] Specifically, the radio frequency identification module uses high-frequency RFID technology to read the electronic tag of the appliance plug; it can obtain preset information of the device, including basic data such as device model, rated power, energy efficiency level, and manufacturer; it supports multi-tag simultaneous identification and anti-collision mechanism to ensure accurate identification when multiple devices are connected.

[0092] The electric energy metering module accurately measures electrical parameters and energy consumption data of the device, integrates a high-precision electric energy metering chip, and monitors key electrical parameters such as voltage, current, power, and power factor in real time; it uses high-speed sampling technology (above 4kHz) to capture electrical characteristics in states such as device startup, operation, standby, and shutdown; it provides harmonic analysis function to detect current waveform distortion of the device and identify nonlinear load characteristics.

[0093] The environmental perception module collects relevant parameters of the device operating environment, and is integrated with multiple environmental sensors, including temperature and humidity sensors, light sensors, human infrared sensors, etc. The environmental perception module monitors the temperature and humidity changes of the device operating environment. Through light intensity detection, it provides environmental context for intelligent control of light-sensitive devices. Using human sensing function, it judges the presence of personnel and realizes intelligent energy-saving control based on personnel activities.

[0094] The temperature monitoring module is arranged with multiple high-precision temperature sensors at key positions of the socket, including power input end, output end and socket contact point. It monitors the temperature rise caused by plug contact resistance in real time, and early detects safety hazards such as poor contact. It establishes a device operating temperature model to analyze the health status of the device through temperature change trend. It provides overheat warning and protection function to prevent safety accidents caused by overheating.

[0095] Specifically, the relative temperature rise The calculation formula is:

[0096] ;

[0097] Wherein, is the measured temperature; is the environmental reference temperature;

[0098] The moving average temperature calculation formula is:

[0099] ;

[0100] Wherein, n is the moving window size; is the historical temperature data;

[0101] The contact resistance estimation formula is:

[0102] ;

[0103] Wherein, is the material resistivity; L is the contact length; A is the contact area;

[0104] The temperature change rate calculation formula is:

[0105] ;

[0106] Wherein, is the time interval; k is the heating rate; T(t) is the current temperature; represents the temperature at the last moment;

[0107] The overheat warning condition is:

[0108] ; ;

[0109] wherein, is a temperature rise threshold value; is a duration; t hold is a minimum trigger duration;

[0110] For example, typical parameters: ΔT threshold = 30℃, t hold = 60s, ρcopper = 1.72 x 10 -8 Ω·m.

[0111] The temperature monitoring module realizes the judgment of the device health state and the overheating early warning by monitoring the temperature change ΔT(t), the moving average trend MA n (t) and the change rate k in real time, in combination with the contact resistance model R contact .

[0112] The protection is triggered only when the temperature rise exceeds the threshold value and lasts for a sufficient time , to avoid false positives.

[0113] The multi-element data collected by each module of the perception layer are collected to the control layer through a unified interface protocol.

[0114] The control layer performs deep processing and analysis on the data collected by the perception layer, and performs functions such as device identification, strategy management and communication coordination; the control layer includes an MCU main control module, a device identification engine, a storage module and a communication module.

[0115] Among them, after the MCU main control module receives multi-source data from the perception layer, it first performs data preprocessing and feature alignment to ensure the time synchronization and integrity of the data;

[0116] The device identification engine then starts multi-modal feature analysis, fuses radio frequency identification, electrical characteristics and environmental parameters, and completes accurate identification of the device type through the built-in machine learning algorithm;

[0117] In this process, the storage module provides fast access support for the feature template library and the identification model, and records the running history data of the device at the same time;

[0118] The communication module is responsible for establishing a stable connection with the platform layer, uploading the identification results and device state in real time, and receiving the control strategy issued.

[0119] The entire processing process of the control layer forms a local decision-making closed loop, ensuring that the basic intelligent functions can still be maintained when the network is interrupted.

[0120] Further, the MCU master module adopts a high-performance 32-bit ARM Cortex-M series microprocessor, providing strong computing power and real-time response capability; coordinates the working sequence of each functional module to ensure stable and reliable operation of the system; executes device feature extraction algorithms, recognition classification algorithms and energy optimization algorithms; manages task scheduling and resource allocation, optimizes system power consumption and performance balance; handles abnormal situations and fault recovery to ensure the robustness and reliability of the system.

[0121] The device recognition engine is used to implement intelligent recognition of devices based on multi-modal features; integrates various machine learning algorithms, including support vector machines, random forests and lightweight neural networks; can perform fusion analysis on the radio frequency features, electrical features and behavior features collected by the perception layer; establishes a device feature database, supports incremental learning and model self-optimization; provides device recognition confidence assessment and starts an interactive confirmation mechanism when the confidence is low; in addition, it also supports learning and feature modeling of new devices, continuously expanding the recognition capability.

[0122] Specifically, the Z-score standardization formula is:

[0123] ;

[0124] wherein, is the original feature; is the mean; is the standard deviation;

[0125] The calculation formula of weighted feature fusion is: ;

[0126] wherein, w1, w2, w3 all represent weights, which are determined according to experiments; w1+w2+w3=1; F fused is the fusion feature vector; F RFID is the radio frequency feature; F elec is the electrical feature; F env is the environmental feature.

[0127] The classification confidence is:

[0128] ;

[0129] wherein, W is the weight matrix; b is the bias term; F fused is the fusion feature vector;

[0130] The incremental learning loss function is: ;

[0131] wherein, is the loss function of the new task; is the forgetting factor, usually taking 0.9; The loss function for the old task; Let be the loss function for the new sample.

[0132] The storage module provides data storage and management functions; it adopts a multi-level storage architecture, including RAM, Flash storage, and scalable external storage; it includes storage device feature templates, identification models, energy consumption strategies, and system configuration parameters; and it records historical device operation data, energy consumption statistics, and security event logs.

[0133] The communication module enables data exchange and communication between the system and external systems; it supports multiple communication protocols, including Wi-Fi, Bluetooth, Zigbee, and Ethernet; it provides a stable connection to the cloud platform to achieve data synchronization and remote management; it supports direct communication between devices to achieve local collaborative control and networking functions; and it integrates secure communication mechanisms, including data encryption, identity authentication, and integrity protection.

[0134] The platform layer provides remote service support, including device management, data analysis, user interaction, and advanced algorithms. The platform layer includes a device management module, a data analysis module, a user interaction module, an energy consumption management module, and a core algorithm module.

[0135] The equipment management module receives equipment status data from the control layer, maintains the full lifecycle information of the equipment, and provides a unified equipment query service for other modules.

[0136] The data analysis module performs in-depth mining of massive amounts of equipment operation data to identify power consumption patterns, assess energy efficiency levels, and predict equipment failures. These analysis results provide data support for the energy management module's strategy formulation.

[0137] Based on data analysis results, the core algorithm module continuously optimizes the device identification model and energy consumption optimization algorithm, and sends the updated algorithm model to the control layer through the communication module.

[0138] The user interaction module provides users with an intuitive display of system status and a convenient control interface, while also feeding user feedback and behavioral data back to the data analysis module, forming another important input for system optimization.

[0139] The data analysis module provides big data analysis and in-depth insights; it aggregates operational data from multiple devices to perform cross-device and cross-user statistical analysis; it identifies electricity consumption patterns and analyzes load characteristics to provide electricity consumption behavior profiles; it conducts energy efficiency assessments and energy-saving potential analysis to generate personalized energy-saving recommendations; it establishes a safety risk assessment model to achieve quantitative assessment of the safety situation; and it provides data visualization, transforming complex data into intuitive charts and reports.

[0140] When mining electricity consumption patterns, the data analysis module uses unsupervised learning to uncover the spatiotemporal patterns of user electricity consumption behavior and identify typical electricity consumption scenarios (such as peak hours, equipment start-up and shutdown cycles, and seasonal load changes), providing a basis for energy consumption optimization and fault diagnosis.

[0141] Specifically, the steps for operating an excavator in electric mode are as follows:

[0142] (1) Data preprocessing

[0143] Input data: Device-level / user-level power time series P(t), sampling interval Δt=1min); denoising: moving average filtering to eliminate random fluctuations; normalization: Z-score standardization to make features of different dimensions comparable.

[0144] (2) Time-series characteristic engineering

[0145] The key steps to transform a continuous time series into a discrete feature vector are as follows:

[0146] Sliding window partitioning: Divide the time series into fixed durations (e.g., 1 hour) to generate a window set W={w1,w2,...,wn}.

[0147] Feature extraction: Calculate multidimensional statistical features for each window:

[0148] Morphological characteristics: mean μ, standard deviation σ, peak-to-valley difference R, slope k.

[0149] Frequency domain characteristics: the dominant frequency component f after FFT transformation dom .

[0150] Entropy characteristics: Shannon entropy H measures waveform complexity.

[0151] Feature matrix construction:

[0152] Each window corresponds to a feature vector X. i =[μ,σ,R,k,f dom ,H].

[0153] (3) DBSCAN clustering implementation

[0154] Distance metric: Euclidean distance d(X) i ,X j ) Measures the similarity of feature vectors;

[0155] Neighborhood radius ε: Automatically selected based on the feature space density (e.g., through the k-distance graph method).

[0156] Minimum number of samples MinPts: defined as greater than or equal to a certain threshold (such as 1% of the total number of samples).

[0157] Key point: If the ε-neighborhood of a point contains ≥MinPts samples, then it is a core point.

[0158] Boundary point: A sample that falls into the neighborhood of a core point but is not itself a core point.

[0159] Noise points: isolated samples that are neither core points nor boundary points.

[0160] (4) Pattern Analysis and Application

[0161] Label assignment: Assign semantic labels to each cluster (such as "weekday office mode" and "nighttime standby mode").

[0162] Anomaly detection: Noise points are marked as potential abnormal power consumption events.

[0163] Visualization: Draw a heatmap of the time distribution of clustering results and overlay the original power curve to verify its rationality.

[0164] The energy management module enables intelligent energy management and optimized control; establishes intelligent control strategies based on equipment type, usage habits, and electricity pricing policies; provides multiple control methods such as timed control, conditional triggering, and scene linkage; realizes demand response functions, supports participation in grid peak shaving and demand-side management; conducts energy-saving effect evaluation and optimization, and continuously improves energy management strategies; supports multi-objective optimization to balance user comfort, energy-saving effect, and equipment lifespan.

[0165] The formula for calculating total electricity cost is:

[0166] ;

[0167] Where p(t) is the time-period electricity price; L(t) is the controllable load power; and T is the statistical period.

[0168] The energy saving rate formula is:

[0169] ;

[0170] Among them, E base E is the baseline electricity consumption. opt To optimize power consumption;

[0171] The power factor correction formula is:

[0172] ;

[0173] Where PF is the power factor; P real V represents actual active power; S represents apparent power; V represents actual active power. rms I represents the effective value of the voltage. rms This is the effective value of the current; This is the cosine value of the phase angle;

[0174] Harmonic mitigation objectives: THD represents Total Harmonic Distortion. The target threshold for harmonic control.

[0175] The core algorithm module provides advanced intelligent algorithms and model services; develops equipment identification optimization algorithms to continuously improve identification accuracy and efficiency; implements user habit learning algorithms to adaptively optimize control strategies; provides safety early warning algorithms to achieve fault prediction and health management; develops swarm intelligence algorithms to optimize individual strategies based on multi-user data; and supports online algorithm updates and model optimization to continuously improve the system's intelligence level.

[0176] The core algorithm module is based on a federated learning framework, training a lightweight model on a local MCU and uploading only gradient updates; the federated learning update formula is: ;

[0177] in, These are global model parameters; Let f be the gradient of the global loss function; f is the global loss function. This is the learning rate.

[0178] During use, in the device access identification stage, the raw data collected by the perception layer is processed by the feature extraction and identification analysis of the control layer, and the results are uploaded to the platform layer to update the device status, forming a data flow of "collection-identification-update".

[0179] In the intelligent decision-making process, the data analysis module at the platform layer generates optimization strategies based on historical data. Through processing by the core algorithm module, these strategies are transformed into executable control commands that are then sent to the control layer, completing the "analysis-decision-deployment" decision flow.

[0180] In the execution feedback phase, after the control layer executes the instruction, the perception layer monitors the execution effect, and the data is fed back to the platform layer through the control layer, realizing closed-loop control of "execution-monitoring-feedback".

[0181] In the security monitoring phase, the perception layer continuously collects security parameters, the control layer makes real-time risk assessments, and the platform layer performs trend analysis and strategy optimization, forming a "monitoring-judgment-protection" security system.

[0182] See Figures 2-3 A method for energy management of a smart sensing socket based on device feature recognition includes the following steps:

[0183] S1: Device access and feature acquisition; When an electrical device is plugged into a smart socket, the system initiates a multi-dimensional feature acquisition process, including electrical parameter acquisition, radio frequency identification, working status recording, and environmental parameter acquisition.

[0184] S2: Multimodal feature extraction and device identification; Based on the collected raw data, the system performs multi-level feature extraction and device identification, including time-domain feature extraction, frequency-domain feature analysis, transient behavior modeling, and pattern recognition and matching;

[0185] S3: Device strategy matching and personalized settings; Based on the device identification results, the system automatically matches or generates exclusive energy management strategies, including strategy library retrieval, adaptive strategy adjustment, new device learning, and strategy conflict resolution;

[0186] S4: Real-time energy consumption monitoring and data analysis; Enters continuous monitoring mode to conduct comprehensive tracking and analysis of equipment energy consumption, including multi-granularity data acquisition, energy efficiency status assessment, load characteristic analysis and power consumption pattern identification;

[0187] S5: Intelligent strategy generation and optimization; Based on real-time monitoring data and historical behavior analysis, dynamically generate and optimize energy management strategies, including habit learning and modeling, multi-objective optimization, scenario strategy adaptation and strategy effect prediction;

[0188] Step S6: Refined power consumption control execution; Based on the generated optimization strategy, the system executes specific power consumption control operations, including adjusting operating parameters, optimizing running time, managing standby power consumption, and coordinating load control.

[0189] Step S7: Safety monitoring and anomaly protection; execute safety monitoring functions in parallel to ensure electrical safety, including multi-parameter safety monitoring, anomaly pattern identification, risk level assessment, and graded protection response;

[0190] Step S8: Energy efficiency assessment and strategy feedback; Regularly assess the effectiveness of energy consumption management to form a closed-loop optimization, including quantifying energy-saving effects, assessing user impact, optimizing strategy parameters, and revising abnormal strategies;

[0191] Step S9: Data reporting and cloud collaboration; maintain data synchronization and policy collaboration with the cloud platform, including running data compression and uploading, policy library synchronization and updates, swarm intelligence learning, and remote diagnosis and maintenance.

[0192] See Figure 3 The implementation process of the intelligent sensing socket energy consumption management method of the present invention is as follows:

[0193] When an electrical device is plugged into a socket, the device access and feature acquisition process is initiated immediately. In this initial stage, the dual tasks of RFID identification and electrical feature acquisition are performed simultaneously: the radio frequency identification module reads the electronic tag in the device plug to obtain the device's preset information, while the power metering module captures the device's electrical parameters such as voltage, current, and power at a high sampling frequency of 4kHz, and the environmental sensing module and temperature monitoring module record the environmental context and basic temperature data, respectively.

[0194] After data collection is completed, the process enters the multimodal feature extraction and device identification stage. Through feature matching algorithms, the collected electrical features and radio frequency identification are compared with templates in the device feature library. At this point, a key decision is made: if the feature matching is successful and the confidence level exceeds the set threshold, the device is considered to have been successfully identified; if the matching fails or the confidence level is insufficient, the process switches to learning mode, creates a new feature profile for the device, and applies the default security policy to ensure basic electrical safety.

[0195] For successfully identified devices, the process proceeds to the device policy matching and personalization stage. At this point, an energy efficiency policy matching the device type is retrieved from the policy library, including operating parameter ranges, safety protection thresholds, and optimized control logic. Simultaneously, the basic policy parameters are adjusted to suit individual device models and operating environment characteristics.

[0196] Once the strategy is set, real-time energy consumption monitoring and data analysis are enabled. Through multi-granularity data acquisition, the power changes, energy consumption trends and operating status of the equipment are continuously tracked, while energy efficiency status assessment and load characteristic analysis are performed simultaneously. During this process, safety monitoring and anomaly protection functions are executed in parallel. Through multi-parameter safety monitoring and anomaly pattern recognition algorithms, abnormal states such as overload, short circuit, and leakage are diagnosed in real time.

[0197] When an anomaly is detected, an anomaly confirmation process will be initiated. Once the anomaly is confirmed, corresponding protective actions will be immediately executed, including tiered responses ranging from early warning alerts to emergency power outages, and event details will be recorded while an alarm is issued to the user. If the situation is confirmed to be normal, normal power supply will be maintained.

[0198] During normal power supply periods, strategies are continuously executed and optimized. Based on real-time monitoring data and historical behavior analysis, control strategies are dynamically adjusted using algorithms such as habit learning and modeling, and multi-objective optimization. Simultaneously, the effectiveness of strategy execution is periodically evaluated, forming a closed-loop optimization through indicators such as energy-saving effect quantification and user impact assessment.

[0199] Finally, the execution data is reported and coordinated with the cloud. The locally stored operational data is compressed and its features extracted before being uploaded to the platform layer. At the same time, updated device policies and optimization algorithms are obtained from the cloud. The platform layer further optimizes the policies based on swarm intelligence learning and distributes the updated device policies to the terminals, completing the closed loop of the entire energy consumption management process.

[0200] This complete implementation process embodies intelligent features: through an edge-cloud collaborative architecture, it ensures both the real-time nature of local control and the deep optimization capabilities of cloud computing; through multimodal feature recognition and adaptive learning, it continuously optimizes device identification accuracy and energy management effectiveness; and through the deep integration of security monitoring and energy management, it maximizes energy efficiency while ensuring electricity safety. The various stages of the entire process are closely linked, forming a complete closed loop from device identification to strategy optimization, providing users with a safe, intelligent, and efficient electricity management experience.

[0201] 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, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart sensing socket based on device feature recognition, comprising a base, a shell, sockets, an internal metal conductor, a power cord, a mechanical switch, and an overload protector, characterized in that, It also includes a perception layer and a control layer; The sensing layer collects various characteristic parameters and environmental data of the connected devices from all directions; the sensing layer includes a radio frequency identification module, an energy metering module, an environmental sensing module, and a temperature monitoring module; The control layer performs in-depth processing and analysis on the data collected by the perception layer, and performs device identification, policy management, and communication coordination; the control layer includes an MCU main control module, a device identification engine, a storage module, and a communication module; After receiving multi-source data from the sensing layer, the MCU main control module performs data preprocessing and feature alignment; the device identification engine then starts multimodal feature analysis, fusing RFID tags, electrical features, and environmental parameters. The communication module uploads the identification results and device status in real time, and receives the control policies sent out. The smart sensing socket can interact with a platform layer set in the cloud; the platform layer provides remote service support and includes a device management module, a data analysis module, a user interaction module, an energy consumption management module, and a core algorithm module. The raw data collected by the perception layer is processed by the control layer for feature extraction and identification analysis. The results are then uploaded to the platform layer to update the device status. The data analysis module of the platform layer generates optimization strategies based on historical data. Through the processing of the core algorithm module, executable control commands are generated and sent to the control layer. After the control layer executes the commands, the perception layer monitors the execution effect, and the data is fed back to the platform layer through the control layer.

2. The intelligent sensing socket based on device feature recognition according to claim 1, characterized in that, The radio frequency identification module uses high-frequency RFID technology to read the electronic tag on the electrical plug; it can obtain preset information of the device, including basic data such as device model, rated power, energy efficiency level, and manufacturer. It supports simultaneous identification of multiple tags and anti-collision mechanisms to ensure accurate identification even when multiple devices are connected; The power metering module accurately measures the electrical parameters and energy consumption data of the equipment, integrates a high-precision power metering chip, and monitors key electrical parameters in real time; it also uses high-speed sampling technology to capture the electrical characteristics of the equipment in various states. It provides harmonic analysis capabilities to detect current waveform distortion in equipment and identify nonlinear load characteristics; The environmental sensing module collects relevant parameters of the equipment's operating environment. It integrates multiple environmental sensors, including temperature and humidity sensors, light sensors, and human infrared sensors. The environmental sensing module monitors changes in temperature and humidity in the equipment's operating environment and provides environmental context for the intelligent control of photosensitive devices through light intensity detection. By utilizing human body sensing capabilities, the presence of people can be determined, enabling intelligent energy-saving control based on human activity.

3. The intelligent sensing socket based on device feature recognition according to claim 1, characterized in that, The temperature monitoring module consists of multiple high-precision temperature sensors arranged at key locations in the socket, including the power input terminal, output terminal, and socket contact point. Real-time monitoring of temperature rise caused by plug contact resistance enables early detection of potential safety hazards; establishment of equipment operating temperature models allows for analysis of equipment health status through temperature change trends; and overheat warning and protection functions prevent safety accidents caused by overheating. Specifically, relative temperature rise The calculation formula is: ； in, This is the measured temperature; The ambient reference temperature; The formula for calculating moving average temperature is: ； Where n is the size of the moving window; Historical temperature data; The formula for estimating contact resistance is: ； in, L is the material resistivity; A is the contact length; and A is the contact area. The formula for calculating the rate of temperature change is: ； Where Δt is the time interval; k is the heating rate; and T(t) is the current temperature. Indicates the temperature at the previous moment; The overheat warning conditions are: ； ； in, This is the temperature rise threshold; Duration; This is the minimum trigger duration.

4. The intelligent sensing socket based on device feature recognition according to claim 3, characterized in that, The temperature monitoring module monitors the temperature change ΔT(t) and the moving average trend MA in real time. n (t) and the rate of change k, combined with the contact resistance model R contact This enables the assessment of equipment health status and provides overheat warnings; only when the temperature rise exceeds the threshold ΔT... threshold And last for a sufficient period of time The protection is triggered in a timely manner to avoid false alarms.

5. A smart sensing socket based on device feature recognition according to claim 1, characterized in that, The device recognition engine can intelligently identify devices based on multimodal features; it can fuse and analyze the radio frequency features, electrical features, and behavioral features collected by the perception layer; it can establish a device feature database, support incremental learning and model self-optimization; it can provide device recognition confidence assessment and initiate an interactive confirmation mechanism when the confidence level is low; in addition, it can support the learning and feature modeling of new devices, continuously expanding the recognition capabilities. Specifically, the Z-score standardization formula is: ； in, Original features; The mean; Standard deviation; The formula for calculating weighted feature fusion is: ; Where w1, w2, and w3 represent weights, which are determined experimentally; w1 + w2 + w3 = 1; F fused For fusing feature vectors; F RFID Radio frequency characteristics; F elec Electrical characteristics; F env Environmental characteristics; Classification confidence for: ； Where W is the weight matrix; b is the bias term; F fused To fuse feature vectors; The incremental learning loss function is: ; in, The loss function for the new task; Forgetting factor; The loss function for the old task; Let be the loss function for the new sample.

6. A smart sensing socket based on device feature recognition according to claim 1, characterized in that, The storage module provides data storage and management functions; it adopts a multi-level storage architecture, including RAM, Flash storage, and scalable external storage; and includes storage device feature templates, identification models, energy consumption strategies, and system configuration parameters. Record historical data of equipment operation, energy consumption statistics and safety event logs.

7. A smart sensing socket based on device feature recognition according to claim 1, characterized in that, The device management module receives device status data from the control layer and maintains the device's full lifecycle information. Based on data analysis results, the core algorithm module continuously optimizes the device identification model and energy consumption optimization algorithm, and sends the updated algorithm model to the control layer through the communication module; the user interaction module provides users with an intuitive display of system status and a convenient control interface, while feeding user feedback and behavioral data back to the data analysis module. The data analysis module provides big data analysis and in-depth insight capabilities; Aggregate operational data from multiple devices to perform statistical analysis across devices and users; It enables electricity consumption pattern recognition and load characteristic analysis to provide a profile of electricity consumption behavior; it conducts energy efficiency assessment and energy-saving potential analysis to generate personalized energy-saving suggestions. Establish a security risk assessment model to achieve a quantitative assessment of the security situation; It provides data visualization, transforming complex data into intuitive charts and reports.

8. A smart sensing socket based on device feature recognition according to claim 1, characterized in that, The energy management module enables intelligent energy management and optimized control; establishes intelligent control strategies based on equipment type, usage habits, and electricity pricing policies; and provides multiple control methods. It enables demand response capabilities, supports participation in grid peak shaving and demand-side management; conducts energy-saving effect assessment and optimization, and continuously improves energy consumption management strategies; supports multi-objective optimization, balancing user comfort, energy-saving effect and equipment lifespan. Specifically, the formula for calculating the total electricity cost is as follows: ； in, Electricity price is based on time of day; The power is the controllable load; T is the statistical period. The energy saving rate formula is: ； in, Based on the baseline electricity consumption; To optimize power consumption; The power factor correction formula is: ； Where PF is the power factor; P real V represents actual active power; S represents apparent power; V represents actual active power. rms I represents the effective value of the voltage. rms This is the effective value of the current; This is the cosine value of the phase angle; Harmonic mitigation objectives: THD represents Total Harmonic Distortion. The target threshold for harmonic control.

9. A smart sensing socket based on device feature recognition according to claim 1, characterized in that, The core algorithm module is based on a federated learning framework, which trains a lightweight model on a local MCU and only uploads gradient updates. The federated learning update formula is: ; in, These are global model parameters; Let f be the gradient of the global loss function; f is the global loss function. This is the learning rate.

10. A method for energy management of an intelligent sensing socket based on device feature recognition, characterized in that, Includes the following steps: S1: Device access and feature acquisition; When an electrical device is plugged into a smart socket, a multi-dimensional feature acquisition process is initiated. S2: Multimodal feature extraction and device identification; Based on the collected raw data, perform multi-level feature extraction and device identification; S3: Device policy matching and personalized settings; automatically match or generate exclusive energy management policies based on device identification results; S4: Real-time energy consumption monitoring and data analysis; Entering continuous monitoring mode, the system performs comprehensive tracking and analysis of equipment energy consumption, including multi-granular data acquisition, energy efficiency status assessment, load characteristic analysis, and power consumption pattern identification. S5: Intelligent strategy generation and optimization; Based on real-time monitoring data and historical behavior analysis, energy consumption management strategies are dynamically generated and optimized. Step S6: Refined power consumption control execution; Based on the generated optimization strategy, execute specific power consumption control operations; Step S7: Safety monitoring and anomaly protection; execute safety monitoring functions in parallel to ensure electrical safety, including multi-parameter safety monitoring, anomaly pattern identification, risk level assessment, and graded protection response; Step S8: Energy efficiency assessment and strategy feedback; regularly evaluate the effectiveness of energy consumption management to form a closed-loop optimization; Step S9: Data reporting and cloud collaboration; maintain data synchronization and policy collaboration with the cloud platform.

Images