Socket electricity data intelligent analysis and interaction system based on internet of things

Abstract

The application relates to the technical field of intelligent sockets, and discloses a socket power data intelligent analysis and interaction system based on the Internet of Things, which comprises an electrical characteristic acquisition module, a mixed analysis module, a safety boundary adjustment module and an intervention module.The electrical characteristic acquisition module is used for acquiring the instantaneous current and voltage waveform of an electrical appliance on a socket and extracting an electrical characteristic fingerprint; the mixed analysis module is used for generating an electrical appliance health score and a behavior normality probability; the safety boundary adjustment module is used for calculating a dynamic safety boundary according to the electrical appliance health score and the behavior normality probability; and the intervention module is used for executing an intervention strategy; and the network communication module is used for communicating with a mobile terminal.The electrical characteristic fingerprint is acquired by the electrical characteristic acquisition module, and the electrical appliance health score is generated by the mixed analysis module based on the long-term evolution trend, so that the performance decline of the electrical appliance caused by aging or loss can be quantitatively represented, the safety protection is no longer limited to post-response, potential risks can be identified and warned in advance, and the safety management is changed from post-disposal to pre-prevention.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent socket, in particular to a socket power data intelligent analysis and interaction system based on Internet of Things. BACKGROUND

[0002] With the popularity of Internet of Things technology and the deepening of the concept of smart home, the field of power safety management has also undergone significant technological evolution. Traditional power safety mainly relies on passive protection devices such as fuses and air circuit breakers, whose function is to cut off the circuit when the current in the circuit exceeds the preset fixed rated value, in order to prevent fire and other dangers caused by overload or short circuit.

[0003] In recent years, as the application of Internet of Things in home scenarios, intelligent sockets have been widely deployed. The existing intelligent sockets integrate network communication and basic power metering functions on the basis of traditional sockets, enabling users to remotely control the switching of electrical appliances, set timing tasks, and view real-time power consumption or cumulative power usage. Some advanced products also introduce basic overload protection functions, that is, when the monitored power exceeds a fixed threshold, they can automatically power off and send alarm information to the user through the network.

[0004] However, the safety protection mechanisms of both traditional protection devices and existing intelligent sockets have inherent limitations. These mechanisms generally rely on a single and fixed safety threshold (such as rated current or rated power), making the determination method unable to adapt to complex and variable actual power scenarios. It cannot distinguish the health status of electrical appliances, and a nearly retired, severely degraded electrical appliance shares the same safety boundary as a new electrical appliance when running, which makes it impossible to identify potential risks in advance, thus creating potential risks. SUMMARY

[0005] To overcome the shortcomings of the prior art, the present application provides a socket power data intelligent analysis and interaction system based on Internet of Things, which solves the problem that the fixed safety threshold used in the prior art cannot associate the actual health status of electrical appliances, and the same safety boundary makes it impossible to identify potential risks in advance, thus creating potential risks.

[0006] The first aspect of the present application provides a socket power data intelligent analysis and interaction system based on Internet of Things, comprising:

[0007] An electrical characteristic acquisition module, the input end of which is connected to the power grid, and the output end of which is connected to the electrical appliance, for acquiring the instantaneous current and voltage waveform of the electrical appliance on the socket, and extracting electrical characteristic fingerprints containing steady-state, distortion and transient characteristics based on the instantaneous current and voltage waveform;

[0008] a hybrid analysis module connected to the output of the electrical characteristic collection module, which generates an electrical appliance health score based on the electrical characteristic fingerprint to represent the health status of the electrical appliance, and generates a behavior normality probability based on the current electrical behavior and the electrical appliance context behavior of the electrical appliance to represent the degree of conformity between the electrical appliance behavior and the user habit;

[0009] a security boundary adjustment module connected to the output of the hybrid analysis module, which calculates and adjusts the dynamic security boundary of the electrical appliance in real time according to the electrical appliance health score and the behavior normality probability;

[0010] an intervention module connected to the output of the security boundary adjustment module and the electrical characteristic collection module, which compares the real-time electrical parameters of the electrical appliance with the dynamic security boundary, and executes an intervention strategy when the real-time electrical parameters exceed the dynamic security boundary;

[0011] a network communication module connected to the intervention module and an external mobile terminal, which sends warning and inquiry messages associated with the intervention strategy generated by the intervention module, and receives control instructions from the mobile terminal, and then sends the control instructions of the mobile terminal to the intervention module, so that the intervention module executes the control instructions of the mobile terminal.

[0012] As a further optimization of the technical scheme of the present application:

[0013] In one embodiment, the electrical characteristic fingerprint is a multi-dimensional vector, which specifically includes the effective value of current, the effective value of voltage, active power, power factor, current crest factor and total harmonic distortion.

[0014] In one embodiment, the hybrid analysis module is internally provided with an electrical appliance health degradation prediction model. The model can adopt a time convolution network structure, the input of which is the time series of electrical characteristic fingerprints generated by the electrical appliance in a period of time, and the output is the electrical appliance health score. The model learns the correlation between the long-term evolution trend of the electrical characteristic fingerprint and the health status of the electrical appliance through offline training.

[0015] In one embodiment, to obtain the electrical appliance context behavior, the hybrid analysis module is configured to identify and record the electrical events occurring on one or more sockets in the same user environment; wherein an electrical event is defined as a complete electrical life cycle. Subsequently, the hybrid analysis module constructs a cross-socket electrical event chain based on the time stamps and logical associations of multiple electrical events. The event chain is used as part of the context input required to generate the behavior normality probability.

[0016] In one embodiment, the safety boundary adjustment module implements its function through a specific calculation process. The module takes the preset reference safety threshold related to the electrical appliance specifications as the calculation basis, and adjusts it non-linearly according to the electrical appliance health score and the behavior normality probability to generate the dynamic safety boundary. The specific implementation formula is:

[0017] T dyn =T base ·g(H score ,P normal );

[0018] In the formula, T dyn is the dynamic safety boundary, T base is the reference safety threshold, H score is the electrical appliance health score, P normal is the behavior normality probability, and g(.) is an adjustment function whose function value monotonically decreases with the decrease of input variables H score or P normal . This structure ensures that the value of the dynamic safety boundary decreases accordingly with the decrease of the electrical appliance health score and the behavior normality probability.

[0019] In one embodiment, the intervention strategy is a hierarchical strategy. Specifically, when the electrical appliance health score is lower than a first preset threshold, a low-risk early warning inquiry is triggered; when the behavior normality probability is lower than a second preset threshold, but the real-time power consumption parameter does not exceed the dynamic safety boundary, a medium-risk inquiry is triggered; when the real-time power consumption parameter exceeds the dynamic safety boundary, the intervention module performs the operation of cutting off the power supply of the electrical appliance by the socket. The low-risk early warning inquiry and the medium-risk inquiry are specifically implemented by sending messages to the mobile terminal through the network communication module.

[0020] In one embodiment, the intervention module also contains a parallel protection mechanism. This mechanism directly compares the real-time extracted electrical feature fingerprint with a preset normal mode representing the absolute normal working state of the electrical appliance. When the deviation of the comparison result exceeds the preset range, it indicates that a serious electrical fault has occurred. At this time, the intervention module does not rely on the comparison result of the dynamic safety boundary, and directly executes the intervention strategy of cutting off the power supply.

[0021] In one embodiment, to protect user data, the training and running process of the hybrid analysis module can be carried out under a federated learning framework. Under this framework, the original electrical feature fingerprint data containing user power consumption information is kept in the local computing unit and is not uploaded through the network. Only the updated parameters (such as gradients or weights) of the internal models such as the electrical appliance health degradation prediction model are transmitted to the cloud server through the network communication module for encrypted transmission, for aggregation to generate a global model.

[0022] The second aspect of the application provides a socket power data intelligent analysis and interaction method based on Internet of Things, which is realized by the system of any of the above technical solutions, and comprises the following steps:

[0023] Step S1, collecting the instantaneous current and voltage waveform of the electrical appliance on the socket, and extracting the electrical characteristic fingerprint;

[0024] Step S2, generating an electrical appliance health score based on the electrical characteristic fingerprint, and generating a behavior normal probability based on the current electrical behavior of the electrical appliance and the electrical appliance context behavior;

[0025] Step S3, calculating and adjusting the dynamic safety boundary of the electrical appliance in real time according to the electrical appliance health score and the behavior normal probability;

[0026] Step S4, comparing the real-time electrical parameters of the electrical appliance with the dynamic safety boundary, and executing an intervention strategy when the real-time electrical parameters exceed the dynamic safety boundary;

[0027] Step S5, sending a warning and inquiry message associated with the intervention strategy, and receiving a control instruction from a mobile terminal, and then the intervention module receives and executes the control instruction of the mobile terminal.

[0028] The application provides a socket power data intelligent analysis and interaction system based on Internet of Things. The system has the following advantages:

[0029] 1. The electrical characteristic fingerprint is obtained by the electrical characteristic acquisition module, and the electrical appliance health score is generated by the mixed analysis module based on the long-term evolution trend, so that the performance degradation of the electrical appliance caused by aging or wear can be quantitatively represented, the safety protection is no longer limited to post-response, and potential risks can be identified and warned in advance, and the safety management is changed from post-disposal to pre-prevention.

[0030] 2. The dynamic safety boundary is adjusted in real time by the safety boundary adjustment module according to the electrical appliance health score and the behavior normal probability, so that the dynamic safety boundary can be automatically tightened when the health condition of the electrical appliance is poor or its use behavior deviates significantly from the user habit, and the reasonable range is maintained during normal use, compared with the traditional fixed threshold protection, the accuracy and reliability of the protection are significantly improved.

[0031] 3、The application effectively reduces false alarms caused by atypical use scenarios by introducing and analyzing the electrical context behavior of the electrical appliance, and the mixed analysis module understands the electrical context behavior of the electrical appliance by constructing a cross-socket electrical event chain, generates a normal behavior probability, and can distinguish between truly abnormal electricity and atypical but normal operation, avoids unnecessary intervention in safe but unconventional use scenarios, improves the accuracy of risk identification, and ensures that the execution of the intervention strategy is more effective. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The figure is a schematic diagram of the system architecture of the application.

[0033] Figure 2 The figure is a schematic diagram of the method flow of the application.

[0034] Among them, 100, electrical characteristic acquisition module; 200, mixed analysis module; 300, safety boundary adjustment module; 400, intervention module; 500, network communication module. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the specification of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0036] In order to better understand the application, the above content will be described in detail below in combination with specific embodiments.

[0037] Referring to the drawings Figure 1 , Figure 1 The figure is a system architecture diagram of the socket electricity data intelligent analysis and interaction system based on the Internet of Things of the embodiment of the application. A socket electricity data intelligent analysis and interaction system based on the Internet of Things is provided, which comprises an electrical characteristic acquisition module 100, a mixed analysis module 200, a safety boundary adjustment module 300, an intervention module 400 and a network communication module 500.

[0038] An electrical characteristic acquisition module 100, which is physically integrated in the smart socket terminal, has its circuit input end connected with the external power grid and its circuit output end connected with the electrical appliance. The module is internally integrated with an analog-to-digital converter with a high sampling rate, which is used to synchronously acquire the instantaneous current and voltage waveforms flowing through the socket. Meanwhile, the module also contains an edge computing unit, which is used to extract the electrical characteristic fingerprints containing the steady-state, distortion and transient characteristics based on the instantaneous current and voltage waveforms in real time. The data output end of the module is connected with a hybrid analysis module 200 for transmitting the extracted electrical characteristic fingerprints; the data output end of the module is connected with an intervention module 400 for transmitting the electrical parameters representing the real-time power consumption of the electrical appliance.

[0039] A hybrid analysis module 200, which has its data input end connected with the electrical characteristic acquisition module 100 for receiving the electrical characteristic fingerprints. The module can be internally composed of one or more processors deployed in a cloud server or a home gateway and algorithm models running thereon. The module generates an electrical appliance health score representing the health state of the electrical appliance based on the time series of the received electrical characteristic fingerprints; and the module also generates a behavior normal probability representing the conformity of the electrical appliance behavior and user habits based on the current electrical behavior of the electrical appliance and the context behavior of the electrical appliance derived from the analysis. The output end of the module is connected with a safety boundary adjustment module 300 for transmitting the generated electrical appliance health score and behavior normal probability to the safety boundary adjustment module 300.

[0040] A safety boundary adjustment module 300, which has its input end connected with the hybrid analysis module 200 for receiving the electrical appliance health score and behavior normal probability. The module calculates and adjusts the dynamic safety boundary for the electrical appliance in real time according to the two received input values.

[0041] An intervention module 400, which has three input ends connected with the electrical characteristic acquisition module 100, the safety boundary adjustment module 300 and a network communication module 500 respectively. The module receives the real-time electrical parameters from the electrical characteristic acquisition module 100 and the dynamic safety boundary from the safety boundary adjustment module 300. The module internally contains a comparator for comparing the real-time electrical parameters with the dynamic safety boundary, and when the real-time electrical parameters exceed the dynamic safety boundary, the module executes one or more preset intervention strategies. The output end of the module sends a control signal to the power control unit (such as a relay) in the electrical characteristic acquisition module 100 to execute the power supply cutoff operation, and the other output end is connected with the network communication module 500 for sending the warning and inquiry messages associated with the intervention strategies. The module also receives control instructions from the network communication module 500 and executes the corresponding intervention strategies according to the instruction content.

[0042] A network communication module 500, which is a bidirectional communication interface, such as a Wi-Fi module or a cellular communication module. One end of its internal data link is connected to the intervention module 400, and the other end is used to establish a wireless communication connection with an external mobile terminal. The function of this module is to encapsulate the early warning and inquiry messages received from the intervention module 400 and send them to the mobile terminal. At the same time, the module also receives control instructions from the mobile terminal, analyzes them, and sends the instruction content to the intervention module 400.

[0043] In a specific embodiment, the electrical feature acquisition module 100 is configured to perform high-fidelity electrical signal digitization and feature extraction. Through its internal voltage sampling channel in parallel with the circuit and the current sampling channel in series, the module uses a high-resolution analog-to-digital converter (ADC) to synchronously sample the instantaneous current i(t) and the instantaneous voltage v(t) within one or more power frequency cycles, thereby obtaining discrete digital sequences that accurately reflect the waveform details. To ensure that high-frequency components generated by nonlinear loads or transient events can be captured, the sampling frequency f s is set to be significantly higher than twice the highest harmonic frequency in the measured signal to meet the requirements of the Nyquist sampling theorem.

[0044] The edge computing unit within the module processes the collected digital waveform sequences within a preset time window T, which is usually an integer multiple of the power frequency cycle, to extract a multi-dimensional electrical feature fingerprint vector. This vector is a set of calculated parameters that comprehensively and quantitatively describe the electrical behavior of the electrical appliance within the time window, providing a data basis for subsequent health state assessment and behavior analysis.

[0045] In a specific embodiment, the electrical feature fingerprint vector specifically includes the following parameters:

[0046] The current effective value I rms and the voltage effective value V rms These two parameters represent the effective amplitude of the current and voltage within a cycle, respectively, and are basic indicators for judging the basic power consumption level of the electrical appliance and the stability of the power supply grid. The calculation formula is:

[0047]

[0048] In the formula, x(t) represents the instantaneous current i(t) or the instantaneous voltage v(t), and T is the integration time window, i.e., a power frequency cycle.

[0049] The active power P active This parameter represents the power that the electrical appliance actually consumes and converts into other forms of energy during operation, and is the core basis for identifying the working state of the electrical appliance (e.g., running, standby, shutdown). The calculation formula is:

[0050]

[0051] Power factor PF. This parameter is a dimensionless index representing the efficiency of electrical energy utilization, whose value is equal to the active power P active over the apparent power. The change of power factor can reflect the change of load properties, such as the aging or failure of inductive load (e.g. motor) or capacitive load (e.g. switching power supply).

[0052] Current crest factor CFI. This parameter is the ratio of the peak value I peak of the absolute value of instantaneous current over the effective value of current I rms , which is used to quantify the peakiness of current waveform. The current crest factor of pure resistive load is close to 1 , while the non-linear load containing rectifier circuit will produce significantly higher crest factor. The abnormal change of this parameter can indicate the performance degradation of electronic components. Its calculation formula is:

[0053]

[0054] In the formula, max(|i(t)|) is the maximum current;

[0055] Total harmonic distortion THDI. This parameter is used to quantify the distortion degree of current waveform relative to ideal sinusoidal wave, which is generated by non-linear load. Its calculation process is as follows: first, the fast Fourier transform (FFT) is performed on the sampling sequence of current waveform i(t), and the amplitude I1 of fundamental component and the amplitudes I k of each harmonic component are separated out, where (k = 2, 3,..., n) The total harmonic distortion is the ratio of the total effective value of all high-order harmonic components over the effective value of fundamental component, which is an important index to evaluate the pollution degree of electrical appliances to power grid and diagnose the health status of internal electronic circuit. Its calculation formula is:

[0056]

[0057] After the calculation of the above parameters is completed, the electrical characteristic acquisition module 100 sends the electrical characteristic fingerprint vector composed of these parameters to the hybrid analysis module 200 through the data interface at a preset period or when a significant change in the state of electrical appliances is detected.

[0058] In one specific embodiment, the mixed analysis module 200 is configured to perform quantitative assessment of the health status of the electrical appliance. The module continuously receives the electrical signature fingerprint vectors extracted at different time points from the electrical signature acquisition module 100, and organizes these vectors in chronological order into a time series. The module internally contains an electrical appliance health degradation prediction model, which is specifically designed to analyze the long-term evolution trend in the electrical signature fingerprint time series, and outputs a scalar value, i.e., the electrical appliance health score.

[0059] In one preferred embodiment, the electrical appliance health degradation prediction model adopts a temporal convolutional network structure. This network structure is specifically designed for processing time series data. The input of this model is an electrical signature fingerprint sequence S = (F1, F2, …, FL) of length L, which contains a multi-dimensional electrical signature fingerprint vector collected at time point t. L

[0060] The core of the temporal convolutional network is composed of stacked residual blocks, each of which internally contains two or more layers of one-dimensional dilated causal convolution. Among them, the causal convolution structure ensures that the output at time step t is only determined by the inputs F1, F2, …, Ft-1 at time steps t and before, without relying on future inputs, which conforms to the physical causality law of time series data processing. L

[0061] The dilated convolution structure introduces a dilation factor d between the elements of the convolution kernel, so that the network can exponentially expand its receptive field without increasing the amount of computation. As the network depth increases, the dilation factor d also increases, which enables the model to capture the dependence relationships spanning a long time span in the electrical signature fingerprint sequence, and thus identify subtle but persistent feature changes caused by slow aging or progressive failure of the electrical appliance.

[0062] The output of each residual block is added to the input of the block through a residual connection, and then subjected to nonlinear activation. This residual connection structure helps to solve the gradient vanishing problem in the training process of deep networks, so that a deeper network can be constructed to learn more complex time series patterns.

[0063] Therefore, the last layer of the temporal convolutional network integrates the high-dimensional time series features extracted by the previous network layers, and finally outputs a single numerical value through a fully connected layer or a global average pooling layer. This value, after normalization (e.g., mapped to the interval of 0-100), is the electrical appliance health score. A higher score represents that the electrical appliance is in a healthy state, while a lower score represents that the electrical appliance has performance degradation or potential failure. This model is obtained by offline training on a large labeled dataset covering the entire life cycle of the electrical appliance.

[0064] ​​In addition to assessing the long-term health status of the electrical appliance, the hybrid analysis module 200 is configured to analyze the short-term usage behavior of the electrical appliance. The module identifies electrical events, constructs the associations between events, and calculates the normal probability of the current usage behavior, i.e. the behavior normal probability, based on the historical usage habits.

[0065] First, the module performs real-time analysis on the electrical characteristic fingerprint stream received from the electrical characteristic acquisition module 100 to identify independent electrical events. An electrical event is defined as the complete electrical process from the start state (e.g. the active power jumps from the standby value to the working value) to the end state (e.g. the active power falls back to the standby value or zero) of the electrical appliance. The module accurately segments the start and end time points and the electrical characteristics during each electrical event by setting the power change threshold and the duration condition.

[0066] Second, the hybrid analysis module 200 is configured to identify and construct the cross-socket electrical event chain. In the user environment, the module is able to aggregate all identified electrical events from one or more smart socket terminals. Based on the timestamps of these events and the categories of the electrical appliances involved, the events that are adjacent in time and logically related are connected into an ordered sequence, i.e. the cross-socket electrical event chain. The event chain constitutes the electrical appliance usage context behavior, for example, recording the sequence of "electric kettle boiling event" followed by "microwave heating event".

[0067] Then, the module calculates the behavior normal probability based on the electrical appliance usage context behavior and the current usage behavior. In one specific embodiment, the calculation is achieved through a probabilistic graphical model, such as a Bayesian network. The Bayesian network constructs the conditional probability table of the mutual associations between different electrical appliance events by learning the long-term electrical event chain data of the user in the offline phase.

[0068] When a new electrical event (current event E current ) occurs, the module combines it with the context event chain (E context ) consisting of one or more events that occurred before it to form a complete observation event sequence. The Bayesian network calculates the probability of the current event occurring given the context event chain has occurred according to the learned conditional probability table. The calculation result is the behavior normal probability P normal :

[0069] P normal = P(E current |E context );

[0070] where P normalis the conditional probability of the current event, whose value is normalized to the interval [0, 1]. A higher probability value indicates that the current power usage behavior is consistent with the established user usage habit, while a lower probability value indicates that the current behavior deviates from the established habit.

[0071] Finally, the module will calculate the behavior normality probability P normal and the appliance health score H score together to the safety boundary adjustment module 300.

[0072] In one specific embodiment, the safety boundary adjustment module 300 is used to generate a dynamic safety boundary that adapts to the state of the electrical appliance and the usage behavior. The input end of this module is connected to the output end of the hybrid analysis module 200, used to receive the real-time updated appliance health score H score and the behavior normality probability P normal .

[0073] Specifically, the core function of this module is to dynamically adjust the preset reference safety threshold value according to the two input values received. The reference safety threshold value T base is a fixed value associated with the factory-rated parameters (such as rated power or rated current) of a specific appliance category or model. It represents the upper limit of safe operation of the electrical appliance under ideal health status and standard usage.

[0074] In one specific embodiment, the expression formula of the dynamic safety boundary is:

[0075] T dyn = T base · g(H score , P normal );

[0076] In the formula, T dyn is the dynamic safety boundary, T base is the reference safety threshold value, H score is the appliance health score, P normal is the behavior normality probability, and g(.) is the adjustment function, whose function value monotonically decreases with the decrease of the input variable H score or P normal . This structure ensures that the value of the dynamic safety boundary decreases accordingly with the decrease of the appliance health score and the behavior normality probability.

[0077] For further illustration, one specific implementation of the adjustment function g(.) can be a weighted combination function:

[0078]

[0079] where a is a preset weight factor, whose value ranges between [0, 1], used to adjust the relative importance of the appliance health score and the probability of normal behavior in calculating the dynamic safety boundary.

[0080] Through the above calculation, when the appliance health status is good (H score value is high) and its usage behavior conforms to the historical habit (P normal value is high), the output value of the adjustment function g(.) is close to 1, so that the dynamic safety boundary T dyn is close to the baseline safety threshold T base . Conversely, when the appliance performance degrades (H score value decreases) or its usage behavior is abnormal (P normal value decreases), the output value of the adjustment function g(.) will be reduced accordingly, resulting in the dynamic safety boundary T dyn being tightened to a level lower than the baseline threshold.

[0081] After the calculation is completed, the safety boundary adjustment module 300 transmits the generated dynamic safety boundary value T dyn to the intervention module 400 for use as a reference for subsequent safety comparison.

[0082] In one specific embodiment, the intervention module 400 is configured to execute a hierarchical intervention strategy according to different input conditions. This module continuously receives the appliance health score H score and the probability of normal behavior P normal from the hybrid analysis module 200, the dynamic safety boundary T dyn from the safety boundary adjustment module 300, and the real-time electrical parameters P realtime from the electrical feature acquisition module 100.

[0083] The logic control unit inside this module judges the above inputs according to a set of preset rules and triggers the corresponding intervention operations. These rules specifically include:

[0084] When the appliance health score H score is lower than a preset first preset threshold, the module determines that there is a long-term, low-level health risk for the appliance. Under this condition, the module performs a low-risk warning inquiry. Specifically, the module generates a warning message containing the appliance health status information, and sends the message to the network communication module 500 through its data output end to notify the user.

[0085] When the probability of normal behavior P normal is lower than a preset second preset threshold, and at the same time, the real-time electrical parameters P realtime do not exceed the dynamic safety boundary T dynWhen the current usage behavior of the electrical appliance deviates significantly from the historical habits, the module determines that there is a medium level of potential risk. Under this condition, the module executes the medium risk inquiry. Specifically, the module generates an inquiry message containing the current electrical behavior anomaly information, and sends the message to the network communication module 500 through its data output end.

[0086] When the value of the real-time electrical parameter P realtime exceeds the dynamic safety boundary T dyn calculated by the safety boundary adjustment module 300, the module determines that the system is in a high level of imminent electrical risk. Under this condition, the module executes the highest priority intervention strategy. Specifically, the module sends a shutdown instruction to the power control unit (such as a solid-state relay or a mechanical relay) integrated in the smart socket through its control signal output end, to immediately interrupt the power supply to the electrical appliance, thereby preventing the occurrence or expansion of dangerous events.

[0087] In this grading strategy, the judgment of the real-time electrical parameter exceeding the dynamic safety boundary has the highest priority, and its triggering condition is independent of the results of the electrical appliance health score and the behavior normality probability.

[0088] In another specific embodiment, to deal with specific abnormal situations that do not cause the real-time electrical parameter to immediately exceed the dynamic safety boundary, but have shown serious fault characteristics, the intervention module 400 also contains a parallel protection mechanism. The execution logic of this mechanism is independent of the aforementioned grading intervention strategy based on the electrical appliance health score and the behavior normality probability.

[0089] The core of this mechanism is to directly compare the real-time acquired electrical feature fingerprint with the pre-set electrical normal mode that characterizes the absolute normal working state of the electrical appliance. The electrical normal mode is a reference benchmark, and is then stored as a multi-dimensional reference electrical feature fingerprint vector F ref . Each component of this vector (for example, the reference total harmonic distortion THDI, the reference current crest factor CFI) is experimentally calibrated or theoretically calculated when the electrical appliance is in a known good state.

[0090] During system operation, the intervention module 400 continuously receives the real-time electrical feature fingerprint vector F realtime from the electrical feature acquisition module 100. The internal logic unit of the module calculates the deviation degree D between the real-time fingerprint vector F realtime and the reference fingerprint vector F ref .

[0091] In one specific implementation, the deviation degree D is quantified by calculating the weighted distance between the two vectors:

[0092]

[0093] where D is the calculated deviation degree; n is the dimension of the electrical feature fingerprint vector; F realtime,i is the i-th component of the real-time fingerprint vector; F ref,i is the i-th component of the reference fingerprint vector; w i is the weight coefficient, used to adjust the importance of different feature components in the deviation calculation.

[0094] The module compares the calculated deviation degree D with a pre-set abnormality judgment threshold. If the value of D exceeds the threshold, it is determined that a serious electrical fault has occurred in the electrical appliance. In this case, the intervention module 400 does not rely on the comparison result of the dynamic safety boundary, but directly executes the highest level of intervention strategy, i.e. sends a shutdown instruction to the power control unit through its control signal output end, immediately interrupting the power supply to the electrical appliance. This mechanism ensures a fast response to specific fault modes.

[0095] In a specific embodiment, the network communication module 500 serves as a bidirectional data interface, responsible for establishing a communication link between the smart socket system and an external mobile terminal operated by the user. The module can be physically a Wi-Fi module, a Bluetooth module or a cellular network module (e.g. 4G or 5G), and follows standard communication protocols (e.g. TCP / IP protocol) for data exchange.

[0096] The data input end of the module is connected to the intervention module 400. When the intervention module 400 generates a warning or inquiry message according to its internal logic judgment, it sends a structured data packet containing the message type (e.g. low-risk warning), the target electrical appliance identifier and the specific information content (e.g. the specific value of the electrical appliance health score) to the network communication module 500.

[0097] After receiving the data packet, the network communication module 500 performs data packaging operations. This operation includes serializing the structured data packet into a standard format (e.g. a JSON format string), and can encrypt it to ensure transmission security. Subsequently, the module sends the packaged data to the cloud server through its wireless antenna, and the server forwards the message to the user application interface of the mobile terminal through the push service.

[0098] The data output end of the module is also connected to the intervention module 400. When the user responds through the user application interface of the mobile terminal, generating a control instruction (e.g. confirming to ignore the inquiry or instructing to immediately cut off the power), the instruction is sent to the cloud server, which forwards it to the network communication module 500.

[0099] The network communication module 500 performs a data parsing operation after receiving the instruction data from the network. This operation includes decrypting and deserializing the received data to restore it to a structured control instruction data packet. This data packet clearly defines the type of operation to be performed and the target device. Finally, the module accurately transmits the parsed control instruction data packet to the intervention module 400 through its data output, and the corresponding subsequent operation is performed by the intervention module 400.

[0100] Referring to the drawings Figure 2 , Figure 2 is a method flowchart of the socket power data intelligent analysis and interaction method based on the Internet of Things according to an embodiment of the application, providing a socket power data intelligent analysis and interaction method based on the Internet of Things, comprising the following steps:

[0101] Step S1, collecting the instantaneous current and voltage waveform of the electrical appliance on the socket, and extracting the electrical characteristic fingerprint;

[0102] In this embodiment, the instantaneous current and voltage waveform flowing through the socket is high-frequency synchronously sampled by the sampling circuit in the intelligent socket. Subsequently, on the local edge computing unit, the waveform data in one or more power frequency cycles is calculated to extract the multi-dimensional electrical characteristic fingerprint. The fingerprint vector includes but is not limited to current and voltage effective values, active power, power factor, current crest factor, and total harmonic distortion, etc.

[0103] Step S2, generating an electrical appliance health score based on the electrical characteristic fingerprint; and generating a behavior normal probability based on the current electrical behavior of the electrical appliance and the electrical appliance context behavior;

[0104] In this embodiment, the continuously collected electrical characteristic fingerprints are formed into a time series and input into an analysis model. The model analyzes the long-term trend in the fingerprint, such as the gradual increase in harmonic components or the continuous decline in power factor, thereby outputting a quantitative electrical appliance health score. At the same time, by identifying the start and stop of the electrical appliance, independent electrical events are defined, and the time-continuous, different-socket electrical events are constructed into a cross-socket electrical event chain, which is the electrical appliance context behavior. By comparing the current electrical behavior with the probability model formed by the historical event chain data, the behavior normal probability is generated.

[0105] Step S3, real-time calculation and adjustment of the dynamic safety boundary of the electrical appliance according to the electrical appliance health score and the behavior normal probability;

[0106] In this embodiment, the calculation of this step is based on the reference safety threshold related to the out-of-factory specification of the electrical appliance. Then, according to the appliance health score and the probability of normal behavior generated in the previous step, an adjustment coefficient is calculated through a weighting function. The reference safety threshold is multiplied by the adjustment coefficient to obtain the final dynamic safety boundary. When the appliance health score or the probability of normal behavior decreases, the boundary is correspondingly tightened to a lower threshold.

[0107] Step S4, comparing the real-time power consumption parameters of the electrical appliance with the dynamic safety boundary, and executing the intervention strategy when the real-time power consumption parameters exceed the dynamic safety boundary;

[0108] In this embodiment, in this step, the key power consumption parameters such as real-time monitored active power or current of the electrical appliance are continuously compared with the dynamic safety boundary generated in the previous step. Once the value of the real-time parameters exceeds the dynamic boundary, the intervention strategy with the highest priority is triggered. In a specific embodiment, the intervention strategy is to send a control signal to the relay in the socket to immediately cut off the power output, thereby preventing the risk.

[0109] Step S5, sending a warning and inquiry message associated with the intervention strategy, and receiving a control instruction from the mobile terminal, and then the intervention module receives and executes the control instruction from the mobile terminal;

[0110] In this embodiment, the method further comprises generating different types of messages according to preset conditions associated with the risk level. For example, a warning message is generated when the appliance health score is below a certain threshold, or an inquiry message is generated when the power consumption behavior is abnormal but does not trigger power overrun. These messages are sent to the user's mobile terminal through a wireless network. The user can reply to the control instruction through the mobile terminal, such as confirming the operation or manually cutting off the power. The instruction is received and parsed by the system, and then the corresponding operation is executed by the intervention module, thereby completing the interactive closed loop.

[0111] Although embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and alterations can be made thereto without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An IoT-based intelligent analysis and interaction system for socket power consumption data, characterized in that: The electrical feature acquisition module is used to acquire the instantaneous current and voltage waveforms of electrical appliances on the socket, and extract electrical feature fingerprints containing steady-state, distortion and transient features based on the instantaneous current and voltage waveforms; The hybrid analysis module generates an appliance health score to characterize the health status of the appliance based on the electrical feature fingerprint; and generates a normal probability of behavior to characterize the degree to which the appliance behavior conforms to the user's habits based on the current power consumption behavior and the power consumption context behavior of the appliance. The safety boundary adjustment module is used to calculate and adjust the dynamic safety boundary of the appliance in real time based on the appliance health score and the probability of normal behavior. An intervention module is used to compare the real-time power consumption parameters of the electrical appliance with the dynamic safety boundary, and to execute an intervention strategy when the real-time power consumption parameters exceed the dynamic safety boundary; The network communication module is used to communicate with the mobile terminal to send warning and inquiry messages generated by the intervention module and associated with the intervention strategy, and to receive control instructions from the mobile terminal. The control instructions from the mobile terminal are then sent to the intervention module, at which point the intervention module executes the control instructions from the mobile terminal.

2. The IoT-based intelligent analysis and interaction system for socket power consumption data as described in claim 1, characterized in that, The electrical fingerprint specifically includes: RMS current value, RMS voltage value, active power, power factor, current crest factor, and total harmonic distortion.

3. The IoT-based intelligent analysis and interaction system for socket power consumption data according to claim 1, characterized in that, The hybrid analysis module specifically includes: An appliance health decline prediction model is proposed, which uses a temporal convolutional network to generate an appliance health score by learning the long-term evolution trend of the electrical feature fingerprint.

4. The IoT-based intelligent analysis and interaction system for socket power consumption data according to claim 1, characterized in that, The hybrid analysis module includes: identifying and constructing power consumption events from one or more smart socket terminals in the same user environment, and constructing a cross-socket electrical event chain based on the power consumption events; and using the cross-socket electrical event chain as the power consumption context behavior of the appliance to generate the normal probability of the behavior.

5. The IoT-based intelligent analysis and interaction system for socket power consumption data according to claim 1, characterized in that, In the intervention module, the intervention strategy specifically includes: When the health score of the electrical appliance is lower than a first preset threshold, a low-risk warning inquiry is triggered. A medium-risk inquiry is triggered when the probability of the behavior being normal is lower than the second preset threshold but the real-time power consumption parameters do not exceed the dynamic safety boundary. When the real-time power consumption parameters exceed the dynamic safety boundary, the power supply from the socket to the appliance is cut off.

6. The IoT-based intelligent analysis and interaction system for socket power consumption data according to claim 5, characterized in that, The low-risk early warning inquiry and the medium-risk inquiry are implemented by sending messages to the mobile terminal through the communication module.

7. The IoT-based intelligent analysis and interaction system for socket power consumption data according to claim 1, characterized in that, The intervention module further includes: comparing the electrical feature fingerprint with a preset electrical normal mode, and executing the intervention strategy directly without relying on the dynamic security boundary when the comparison result is abnormal.

8. The IoT-based intelligent analysis and interaction system for socket power consumption data according to claim 1, characterized in that, The hybrid analysis module operates within a federated learning framework and specifically includes: The original data of the electrical feature fingerprint is stored locally, and the updated parameters of the electrical appliance health decline prediction model in the hybrid analysis module are transmitted through the network communication module.

9. A smart analysis and interaction system for socket power consumption data based on the Internet of Things as described in claim 1, characterized in that, The safety boundary adjustment module includes: using the appliance's baseline safety threshold as the calculation basis, and adjusting it according to the appliance's health score and the probability of normal behavior to generate the dynamic safety boundary; The value of the dynamic safety boundary decreases as the appliance health score and the probability of normal behavior decrease.

10. A method for intelligent analysis and interaction of socket power consumption data based on the Internet of Things (IoT), based on the intelligent analysis and interaction system for socket power consumption data based on the IoT as described in any one of claims 1-9, characterized in that, Includes the following steps: Step S1: Collect the instantaneous current and voltage waveforms of the electrical appliances on the socket and extract the electrical feature fingerprint; Step S2: Based on the electrical feature fingerprint, generate an appliance health score; and based on the current power consumption behavior and the power consumption context behavior of the appliance, generate a normal behavior probability. Step S3: Calculate and adjust the dynamic safety boundary of the appliance in real time based on the appliance health score and the probability of normal behavior; Step S4: Compare the real-time power consumption parameters of the electrical appliance with the dynamic safety boundary, and execute an intervention strategy when the real-time power consumption parameters exceed the dynamic safety boundary; Step S5: Send warning and inquiry messages associated with the intervention strategy and receive control instructions from the mobile terminal. Subsequently, the intervention module receives and executes the control instructions from the mobile terminal.

Images