Intelligent household power utilization safety real-time monitoring and self-adaptive control system

Abstract

The invention discloses a smart home power utilization safety real-time monitoring and self-adaptive control system, which relates to the technical field of smart home power utilization safety monitoring and comprises a sensor acquisition module, an anomaly identification module, a local response module, a network modeling module, a risk prediction module, a control strategy module and a user interaction module. The sensor acquisition module is used for performing data acquisition on the operation state of the household electrical appliance through current, voltage, temperature and power sensors to generate an original sensing data set; the anomaly recognition module is used for constructing a classifier model based on a deep neural network, performing feature extraction on the original sensing data set, generating a feature vector, performing anomaly probability evaluation on the feature vector based on the classifier model, and outputting an equipment operation anomaly signal; and the local response module is used for executing a preset safety response rule on the edge calculation unit based on the equipment operation abnormal signal, and generating a local power-off or current-limiting control instruction.

Description

Technical Field

[0001] The present invention relates to the technical field of smart home electricity safety monitoring, and in particular to a smart home electricity safety real-time monitoring and adaptive control system. Background Art

[0002] Smart home electrical safety monitoring technology utilizes advanced sensing, communication, and data analysis technologies to provide 24 / 7 monitoring of various electrical appliances within a home. Its core goal is to promptly detect and prevent electrical safety hazards, ensuring the safety and reliability of household electricity use. Therefore, utilizing advanced technologies to enhance the intelligence and security of smart home electrical safety monitoring has become a pressing issue.

[0003] In the field of smart home electricity safety monitoring, existing systems often ignore the interactions and influences between electrical appliances and are unable to effectively predict the occurrence of cascading failures. Traditional anomaly detection methods are mainly based on simple threshold judgments, which are prone to false alarms or missed alarms, especially in the face of complex and changing home electricity environments. At the same time, many existing systems rely on cloud servers for data analysis and decision-making, resulting in large response delays and an inability to meet real-time requirements. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a real-time monitoring and adaptive control system for smart home electricity safety to solve the problem that existing systems often ignore the interactions and influences between electrical appliances and cannot effectively predict the occurrence of cascading failures. In addition, traditional anomaly detection methods are mainly based on simple threshold judgments, which are prone to false alarms or missed alarms, especially when facing complex and changeable household electricity environments. At the same time, many existing systems rely on cloud servers for data analysis and decision-making, resulting in large response delays and inability to meet real-time requirements.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a smart home electricity safety real-time monitoring and adaptive control system, which includes:

[0008] Sensor acquisition module, anomaly recognition module, local response module, network modeling module, risk prediction module, control strategy module and user interaction module;

[0009] The sensor acquisition module is used to collect data on the operating status of household appliances through current, voltage, temperature, and power sensors to generate an original sensor data set;

[0010] The anomaly recognition module is used to build a classifier model based on a deep neural network, extract features from the original sensor data set, generate feature vectors, and perform an anomaly probability assessment on the feature vectors based on the classifier model to output a device operation anomaly signal;

[0011] The local response module is used to execute preset safety response rules on the edge computing unit based on the device operation abnormality signal and generate a local power-off or current-limiting control instruction;

[0012] The network modeling module is used to construct a graph structure model based on the physical connection topology and usage dependency between electrical appliances and use a graph neural network to update node embeddings to generate an electrical appliance correlation representation matrix;

[0013] The risk prediction module is used to combine the equipment operation abnormality signal with the electrical appliance correlation representation matrix, simulate the fault propagation path, predict the potential risk level, and output a risk warning signal;

[0014] The control strategy module is used to dynamically adjust the operating parameters of non-abnormal electrical appliances based on local power-off or current-limiting control instructions and risk warning signals in combination with a load balancing algorithm to generate a power distribution optimization plan;

[0015] The user interaction module is used to receive the power distribution optimization plan, generate a visual power consumption status report and operation suggestions, and present them to the user through the human-computer interface, receive personalized configuration parameters input by the user, and feed them back to the control strategy module.

[0016] As a preferred solution of the real-time monitoring and adaptive control system for smart home electricity safety described in the present invention, wherein: the data of the operating status of household appliances is collected by current, voltage, temperature, and power sensors to generate the original sensor data set, the specific steps are as follows:

[0017] Install current sensors, voltage sensors, temperature sensors, and power sensors inside household distribution boxes, sockets, and electrical appliances;

[0018] Deploy the above-mentioned sensors at the node locations where each appliance is connected to the circuit;

[0019] The output signals of each sensor are continuously collected with a fixed sampling period Δt;

[0020] Get the current value, voltage value, temperature value and power value at timestamp t;

[0021] Convert the collected analog signal into a digital signal;

[0022] The four types of collected data are normalized separately, and the expressions are:

[0023]

[0024] Among them, x represents the original value at the current moment, x min Indicates the historical minimum value of the parameter, x max Indicates the historical maximum value of the parameter, x norm is the normalized x value;

[0025] The normalized four-dimensional parameters are combined into a structured raw sensing dataset in time sequence.

[0026] As a preferred solution of the real-time monitoring and adaptive control system for smart home electricity safety described in the present invention, the method comprises the following steps: constructing a classifier model based on a deep neural network, extracting features from the original sensor data set, generating feature vectors, and performing abnormality probability assessment on the feature vectors based on the classifier model to output device operation abnormality signals.

[0027] Convolutional neural network (CNN) is used to extract local features from the original sensor data set obtained from the sensor acquisition module.

[0028] By setting the convolution kernel and bias term, a sliding window convolution operation is performed on the data in the original sensor data set to obtain the primary feature matrix M1, which is expressed as:

[0029] M1=ReLU(W conv *D0+b conv );

[0030] Among them, ReLU(.) represents the activation function, W conv is the trainable convolution weight matrix, b conv is the bias term, D0 is the input tensor;

[0031] Long short-term memory network (LSTM) is used to model the temporal dependency of the primary feature matrix;

[0032] The LSTM network processes each row of M1 in turn, updating the hidden state and cell state;

[0033] Take the hidden state of the last time step as the semantic representation of the entire time series, recorded as the feature vector;

[0034] A nonlinear discriminant function is used to evaluate the abnormal probability of the feature vector and output the abnormal signal of the equipment operation. The discriminant function expression is:

[0035]

[0036] Among them, a is the abnormal probability, α is the global weight coefficient, β is the local feature weighting factor, F j is the jth component of the eigenvector F, w jis the weight corresponding to the jth feature, F∥2 is the L2 norm of the feature vector, indicating the overall fluctuation intensity, and e is the base of the natural logarithm;

[0037] Set the abnormality judgment threshold θ. If a≥a, it is determined that the equipment has an abnormal operation and the equipment operation abnormality signal A=1 is output; otherwise, the output signal A=0 is output.

[0038] Among them, θ is the abnormal discrimination threshold preset by the system, and A is the abnormal signal.

[0039] As a preferred solution of the real-time monitoring and adaptive control system for smart home electricity safety described in the present invention, the specific steps of executing preset safety response rules on the edge computing unit based on the abnormal device operation signal to generate local power-off or current-limiting control instructions are as follows:

[0040] Use the preset security response rule set to match and judge the equipment operation abnormality signal A obtained from the abnormality identification module to determine the type of security response action to be performed;

[0041] Among them, the equipment operation abnormality signal A is a Boolean variable, A indicates that the current electrical appliance has an abnormal operation; the safety response rule set includes multiple rule items R1 to R n , each rule item consists of two parts:

[0042] Exception level description conditions;

[0043] Corresponding action strategy;

[0044] After judging by the above rules, output action type t;

[0045] Adopting relay control algorithm to encapsulate the action type t into instructions and generate local power-off or current-limiting control instructions;

[0046] The local power-off or current-limiting control instruction is a structured data packet, including the appliance's unique identifier, action type, execution timestamp, and target current value during current limiting.

[0047] If the action type t is "power off", the control instruction C cmd The format is:

[0048] C cmd ={ID, type = power off, time = t exec};

[0049] If the action type t is "current limiting", the control instruction C cmd The format is:

[0050] C cmd ={ID, type = current limit, time = t exec,value=I limit};

[0051] Among them, ID is the appliance number, t exec is the current system timestamp, I limit is the target current limit value;

[0052] The target current limit value I limit The calculation expression is:

[0053] I limit =γ·I rated ;

[0054] Among them, γ is the current limiting coefficient, I rated is the rated working current of the appliance;

[0055] Use the communication protocol to control the local power off or current limiting control instruction C cmd Send it to the local controller, which drives the relay or smart switch to perform the corresponding operation;

[0056] Among them, the local controller receives C cmd Then, parse out the appliance ID and action type;

[0057] Find the corresponding physical control port address according to the appliance ID;

[0058] Depending on the action type, do one of the following:

[0059] If type = power off, a shutdown signal is sent to the relay to cut off the power supply to the appliance;

[0060] If type = current limiting, a current limiting instruction is sent to the controllable current regulator to adjust its current output to I limit .

[0061] As a preferred solution of the real-time monitoring and adaptive control system for smart home electricity safety described in the present invention, the steps of constructing a graph structure model based on the physical connection topology and usage dependency between electrical appliances and updating node embeddings using a graph neural network to generate an electrical appliance correlation representation matrix are as follows:

[0062] Model the power connection relationship and functional dependency relationship between household appliances, and then build a graph structure model;

[0063] The graph convolutional network (GCN) is used to learn the embedded representation of nodes in the graph structure and generate an appliance correlation representation matrix.

[0064] Among them, the initial node embedding vector By the attribute vector x v Obtained through linear transformation, the expression is:

[0065]

[0066] Where W0 is the trainable weight matrix. b0 is the bias term. The l-th layer graph convolution is calculated as follows:

[0067]

[0068] Among them, σ is the activation function, N(v) represents the set of neighbor nodes of node v, and W l is the trainable weight matrix of layer l, b l is the bias term of the lth layer, Represents the embedding vector of neighbor node u in the previous layer;

[0069] After l layers of propagation, the node embedding vector is obtained Embed all nodes into an electrical appliance relevance representation matrix:

[0070]

[0071] Where n is the total number of nodes, M rep It is the electrical appliance correlation representation matrix;

[0072] Electrical appliance correlation representation matrix M rep Perform analysis and extract the influence weights between appliances, including the similarity scores of appliances ij and the vulnerability factor f of the electrical appliance i , the expression is:

[0073]

[0074] in, and are the final embedding vectors of appliances i and j respectively, T represents the transpose operation, c i is the number of historical failures of appliance i, k is the adjustment coefficient, P rated,i is the rated power of electrical appliance i, P avg is the system average power value.

[0075] As a preferred solution of the real-time monitoring and adaptive control system for smart home electrical safety described in the present invention, the specific steps of combining the abnormal operation signal of the equipment with the electrical appliance correlation representation matrix, simulating the fault propagation path, predicting the potential risk level, and outputting the risk warning signal are as follows:

[0076] The real-time monitoring module is used to collect the operating status data of each electrical equipment and generate equipment operation abnormality signal S abn ;

[0077] Among them, S abnis a vector, and each element s i corresponds to the abnormal degree score of an electrical device, and the expression is:

[0078] s i = α·(P curr,i - P rated,i ) + β·f freq,i ;

[0079] where P curr,i is the current power of electrical appliance i, P rated,i is the rated power of electrical appliance i, f freq,i is the historical failure frequency of electrical appliance i, and α and β are adjustment coefficients used to balance the influence of current power overload and historical failures on the abnormal score;

[0080] Based on the device operation abnormal signal S abn and the electrical appliance correlation characterization matrix M rep , a fault propagation model is constructed;

[0081] By defining a fault diffusion function, calculate the fault propagation probability between each electrical appliance, and the expression is:

[0082]

[0083] where n represents the total number of electrical appliances, m ij is an element in the M rep matrix, and P fp (i, j) represents the probability that the fault propagates from electrical appliance i to electrical appliance j after an abnormality occurs in electrical appliance i;

[0084] Based on P fp (i, j), simulate the propagation path of the fault in the electrical appliance network and predict the potential risk level;

[0085] For each electrical appliance i, calculate its cumulative risk value R risk,i after being affected by other electrical appliances, and the expression is:

[0086]

[0087] According to the magnitude of R risk,i divide the risk level;

[0088] If R risk,i < T1, the risk level is low risk;

[0089] If T1 ≤ R risk,i < T2, the risk level is medium risk;

[0090] If R risk,i ≥ T2, the risk level is high risk, where T1 and T2 are preset thresholds;

[0091] Output corresponding risk warning signals according to the risk level of each electrical appliance.

[0092] As a preferred solution of the real-time monitoring and adaptive control system for smart home electricity safety described in the present invention, the method of dynamically adjusting the operating parameters of non-abnormal electrical appliances based on local power-off or current-limiting control instructions and risk warning signals in combination with a load balancing algorithm to generate a power distribution optimization plan is specifically carried out as follows:

[0093] Receive control instructions from the local response module and risk warning signals from the risk prediction module;

[0094] Get the current system total load and available capacity;

[0095] For appliances that are not abnormal and have not issued warnings, new operating parameters are assigned based on the user-set priority and energy consumption;

[0096] The constraints for solving the optimal scheduling problem using linear programming are:

[0097]

[0098]

[0099] Among them, p i is the original power, p i ′ is the adjusted power, The maximum power allowed for the device;

[0100] Get the power distribution optimization plan, including the new operating parameters of each electrical appliance.

[0101] As a preferred solution of the real-time monitoring and adaptive control system for smart home electricity safety described in the present invention, the specific steps of receiving the power distribution optimization plan, generating a visual power status report and operation suggestions, and presenting them to the user through the human-machine interface, receiving the personalized configuration parameters input by the user, and feeding them back to the control strategy module are as follows:

[0102] Convert the power distribution optimization plan into a visual graphical interface element, including electrical appliance operating status indicator lights, power change curves, and energy consumption statistics charts;

[0103] Provide operational suggestions in text form;

[0104] Users enter personalized configuration parameters through the touch screen or voice interface, including appliance priority adjustment, energy-saving preference settings, and manual recovery instructions;

[0105] The personalized configuration parameters are encapsulated as configuration signals;

[0106] The configuration signal is fed back to the control strategy module to optimize the next round of regulation decisions.

[0107] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the smart home electricity safety real-time monitoring and adaptive control system as described in the first aspect of the present invention is implemented.

[0108] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the smart home electricity safety real-time monitoring and adaptive control system as described in the first aspect of the present invention.

[0109] The beneficial effects of the present invention are as follows: by deploying multiple types of sensors at multiple locations and performing continuous data collection, real-time perception of the full-dimensional operating status of household electrical appliances is achieved; by extracting local features of the original sensor data through CNN, effective capture of instantaneous mutations and short-term anomalies is achieved; by modeling the long-term dependence of time series through LSTM, accurate identification of slowly degrading anomalies is achieved; by fusing the L2 norm and the nonlinear discriminant function of the weighted feature sum, the adaptability and interpretability of the model to different types of anomalies are improved; by formulating hierarchical safety response rules, differentiated handling of anomalies of different severity is achieved; by designing structured control instructions, the traceability and execution consistency of control commands are guaranteed; by replacing direct power outages with current limiting mechanisms, unnecessary interruptions are avoided while ensuring safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0110] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0111] Figure 1 Schematic diagram of the real-time monitoring and adaptive control system for smart home electricity safety in Example 1.

[0112] Figure 2 This is an architectural diagram of the real-time monitoring and adaptive control system for smart home electricity safety in Example 1.

[0113] Figure 3 Schematic diagram of fault propagation simulation of the risk prediction module in Example 1.

[0114] Figure 4 This is a schematic diagram of the network modeling module in Example 1. DETAILED DESCRIPTION

[0115] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0116] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0117] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0118] Example 1, with reference to Figures 1 to 4 , which is the first embodiment of the present invention, provides a smart home electricity safety real-time monitoring and adaptive control system, including:

[0119] Sensor acquisition module, anomaly recognition module, local response module, network modeling module, risk prediction module, control strategy module and user interaction module;

[0120] The sensor acquisition module is used to collect data on the operating status of household appliances through current, voltage, temperature, and power sensors to generate raw sensor data sets;

[0121] Going a step further, current sensors, voltage sensors, temperature sensors, and power sensors are installed inside household distribution boxes, sockets, and appliances;

[0122] Deploy the above-mentioned sensors at the node locations where each appliance is connected to the circuit;

[0123] The output signals of each sensor are continuously collected with a fixed sampling period Δt;

[0124] Get the current value, voltage value, temperature value and power value at timestamp t;

[0125] Convert the collected analog signal into a digital signal;

[0126] The four types of collected data are normalized separately, and the expressions are:

[0127]

[0128] Among them, x represents the original value at the current moment, x minIndicates the historical minimum value of the parameter, x max Indicates the historical maximum value of the parameter, x norm is the normalized x value;

[0129] The normalized four-dimensional parameters are combined into a structured raw sensing data set in time sequence;

[0130] It should be noted that the multi-location and multi-type sensor deployment method adopted in this embodiment can comprehensively capture the changing trends of key parameters under the operating status of the electrical appliance; continuous collection is performed by setting a fixed sampling period, thereby ensuring the temporal resolution and continuity of the data; the normalization processing method effectively eliminates the impact of dimensional differences of different sensors on subsequent modeling, thereby improving the model stability and generalization ability; the structured original sensor data set format facilitates efficient processing by the subsequent feature extraction and anomaly recognition modules, thereby building a high-quality data foundation to support the operation of the entire system.

[0131] The anomaly recognition module is used to build a classifier model based on a deep neural network, extract features from the original sensor data set, generate feature vectors, and evaluate the anomaly probability of the feature vectors based on the classifier model to output equipment operation anomaly signals;

[0132] Furthermore, a convolutional neural network (CNN) is used to extract local features from the original sensor data set obtained from the sensor acquisition module.

[0133] By setting the convolution kernel and bias term, a sliding window convolution operation is performed on the data in the original sensor data set to obtain the primary feature matrix M1, which is expressed as:

[0134] M1=ReLU(W conv * D0+b conv );

[0135] Among them, ReLU(.) represents the activation function, W conv is the trainable convolution weight matrix, b conv is the bias term, D0 is the input tensor;

[0136] Long short-term memory network (LSTM) is used to model the temporal dependency of the primary feature matrix;

[0137] The LSTM network processes each row of M1 in turn, updating the hidden state and cell state;

[0138] Take the hidden state of the last time step as the semantic representation of the entire time series, recorded as the feature vector;

[0139] A nonlinear discriminant function is used to evaluate the abnormal probability of the feature vector and output the abnormal signal of the equipment operation. The discriminant function expression is:

[0140]

[0141] Among them, a is the abnormal probability, α is the global weight coefficient, β is the local feature weighting factor, F j is the jth component of the eigenvector F, w j is the weight corresponding to the jth feature, F∥2 is the L2 norm of the feature vector, indicating the overall fluctuation intensity, and e is the base of the natural logarithm;

[0142] Set the abnormality judgment threshold θ. If a≥a, it is determined that the equipment has an abnormal operation and the equipment operation abnormality signal A=1 is output; otherwise, the output signal A=0 is output.

[0143] Among them, θ is the abnormal discrimination threshold preset by the system, and A is the abnormal signal;

[0144] It should be noted that this embodiment adopts a CNN-LSTM hybrid neural network structure, combined with a nonlinear discriminant function to perform anomaly probability assessment, which has higher detection accuracy and robustness than the traditional single model; the convolution layer is used to extract local fluctuation characteristics, which is suitable for detecting sudden current and voltage anomalies; the LSTM layer is used to model long-term dependencies, which can effectively identify progressive anomalies such as slow temperature rise and power decrease; the final output equipment operation abnormality signal A can be used to trigger a local response mechanism to achieve early warning and intervention of household electricity risks.

[0145] The local response module is used to execute preset safety response rules on the edge computing unit based on abnormal device operation signals and generate local power-off or current-limiting control instructions;

[0146] Furthermore, a preset set of security response rules is used to match and judge the device operation abnormality signal A obtained from the abnormality identification module to determine the type of security response action to be performed;

[0147] Among them, the equipment operation abnormality signal A is a Boolean variable, A indicates that the current electrical appliance has an abnormal operation; the safety response rule set includes multiple rule items R1 to R n , each rule item consists of two parts:

[0148] The abnormal level describes the condition;

[0149] Corresponding action strategy;

[0150] After judging by the above rules, output action type t;

[0151] Adopting relay control algorithm to encapsulate the action type t into instructions and generate local power-off or current-limiting control instructions;

[0152] The local power-off or current-limiting control instruction is a structured data packet, including the appliance's unique identifier, action type, execution timestamp, and target current value during current limiting.

[0153] If the action type t is "power off", the control instruction C cmd The format is:

[0154] C cmd ={ID, type = power off, time = t exec};

[0155] If the action type t is "current limiting", the control instruction C cmd The format is:

[0156] C cmd ={ID, type = current limit, time = t exec ,value=I limit};

[0157] Among them, ID is the appliance number, t exec is the current system timestamp, I limit is the target current limit value;

[0158] Target current limit value I limit The calculation expression is:

[0159] I limit =γ·I rated ;

[0160] Among them, γ is the current limiting coefficient, I rated is the rated working current of the appliance;

[0161] Use the communication protocol to control the local power off or current limiting control instruction C cmd Send it to the local controller, which drives the relay or smart switch to perform the corresponding operation;

[0162] Among them, the local controller receives C cmd Then, parse out the appliance ID and action type;

[0163] Find the corresponding physical control port address according to the appliance ID;

[0164] Depending on the action type, do one of the following:

[0165] If type = power off, a shutdown signal is sent to the relay to cut off the power supply to the appliance;

[0166] If type = current limiting, a current limiting instruction is sent to the controllable current regulator to adjust its current output to I limit ;

[0167] It should be noted that by setting hierarchical safety response rules, the power-off or current-limiting strategy is automatically selected according to the abnormality level, which improves the flexibility and safety of the system; the design of structured control instructions ensures the consistency and traceability of control commands executed at the edge; the current limiting coefficient can be dynamically adjusted according to the type of electrical appliance to avoid interruptions in use due to direct power outages, thereby improving user satisfaction; the local controller drives the relay or current regulator to perform operations based on the instructions, with a fast response speed and no reliance on the cloud, which enhances the real-time and reliability of the system.

[0168] The network modeling module is used to build a graph structure model based on the physical connection topology and usage dependency between appliances, and uses a graph neural network to update node embeddings to generate an appliance correlation representation matrix;

[0169] Furthermore, the power connection relationship and functional dependency between household appliances are modeled, and then a graph structure model is constructed;

[0170] The graph convolutional network (GCN) is used to learn the embedded representation of nodes in the graph structure and generate an appliance correlation representation matrix.

[0171] Among them, the initial node embedding vector By the attribute vector x v Obtained through linear transformation, the expression is:

[0172]

[0173] Where W0 is the trainable weight matrix. b0 is the bias term. The l-th layer graph convolution is calculated as follows:

[0174]

[0175] Among them, σ is the activation function, N(v) represents the set of neighbor nodes of node v, and W l is the trainable weight matrix of layer l, b l is the bias term of the lth layer, Represents the embedding vector of neighbor node u in the previous layer;

[0176] After l layers of propagation, the node embedding vector is obtained Embed all nodes into an electrical appliance relevance representation matrix:

[0177]

[0178] Where n is the total number of nodes, M rep It is the electrical appliance correlation representation matrix;

[0179] Electrical appliance correlation representation matrix M repPerform analysis and extract the influence weights between appliances, including the similarity scores of appliances ij and the vulnerability factor f of the electrical appliance i , the expression is:

[0180]

[0181] in, and are the final embedding vectors of appliances i and j respectively, T represents the transpose operation, c i is the number of historical failures of appliance i, k is the adjustment coefficient, P rated,i is the rated power of electrical appliance i, P avg is the average power value of the system;

[0182] It should be noted that by modeling the physical connections and functional dependencies between electrical appliances based on graph structures, and using GCN to update node embeddings to generate an appliance correlation representation matrix, graph neural networks were introduced into smart home electricity scenarios for the first time. This method can not only reflect the explicit connection relationships between electrical appliances, but also explore potential indirect impact paths. By calculating similarity scores and vulnerability factors, the transmission intensity and interference risk between devices are further quantified, providing an important basis for subsequent risk prediction modules and significantly improving the system's ability to predict chain failures.

[0183] The risk prediction module is used to combine equipment operation abnormality signals with the electrical appliance correlation representation matrix, simulate the fault propagation path, predict the potential risk level, and output risk warning signals;

[0184] Furthermore, a real-time monitoring module is used to collect the operating status data of each electrical device and generate an abnormal operation signal S abn ;

[0185] Among them, S abn is a vector, each element s i The expression for the abnormality score of an electrical device is:

[0186] s i =α·(P curr,i -P rated,i )+β·f freq,i ;

[0187] Among them, P curr,i is the current power of appliance i, P rated,i is the rated power of appliance i, f freq,i is the historical fault frequency of appliance i, α and β are adjustment coefficients, which are used to balance the impact of current power overload and historical faults on the anomaly score;

[0188] Based on the abnormal signal Sabn and the electrical appliance correlation characterization matrix M rep , construct a fault propagation model;

[0189] By defining a fault diffusion function, calculate the fault propagation probability between electrical appliances. The expression is:

[0190]

[0191] where, n represents the total number of electrical appliances, and m ij is an element in the M rep matrix, and P fp (i, j) represents the probability that the fault propagates to electrical appliance j after electrical appliance i has an abnormality;

[0192] Based on P fp (i, j), simulate the propagation path of the fault in the electrical appliance network and predict the potential risk level;

[0193] For each electrical appliance i, calculate its cumulative risk value R risk,i , and the expression is:

[0194]

[0195] According to the magnitude of R risk,i , divide the risk level;

[0196] If R risk,i < T1, the risk level is low risk;

[0197] If T1 ≤ R risk,i < T2, the risk level is medium risk;

[0198] If R risk,i ≥ T2, the risk level is high risk, where T1 and T2 are preset thresholds;

[0199] Output corresponding risk warning signals according to the risk levels of each electrical appliance;

[0200] It should be noted that by integrating the abnormal signals of equipment operation and the electrical appliance correlation characterization matrix, constructing a fault diffusion function, and simulating the fault propagation path, the dynamic assessment of potential risks is realized; the introduction of the cumulative risk value enables the system to accurately divide the risk level of each electrical appliance, and then output risk warning signals; this mechanism breaks through the limitation of only focusing on the abnormality of a single device in the traditional sense, has a global perspective, supports linkage warning and collaborative regulation, and greatly reduces the possibility of large-scale power outages caused by the abnormality of a single device.

[0201] The control strategy module is used to dynamically adjust the operating parameters of non-abnormal electrical appliances based on local power-off or current-limiting control instructions and risk warning signals, combined with the load balancing algorithm, to generate a power distribution optimization plan;

[0202] Furthermore, it receives control instructions from the local response module and risk warning signals from the risk prediction module;

[0203] Get the current system total load and available capacity;

[0204] For appliances that are not abnormal and have not issued warnings, new operating parameters are assigned based on the user-set priority and energy consumption;

[0205] The constraints for solving the optimal scheduling problem using linear programming are:

[0206]

[0207] Among them, p i is the original power, p i ′ is the adjusted power, The maximum power allowed for the device;

[0208] Get the power distribution optimization plan, including the new operating parameters of each appliance;

[0209] It should be noted that based on local power-off / current-limiting instructions and risk warning signals, combined with the load balancing algorithm, the operating parameters of non-abnormal electrical appliances are dynamically adjusted to realize the automatic generation of power distribution optimization plans; the optimal scheduling problem is solved by linear programming, and the system's available capacity is maximized and energy consumption is minimized while meeting the equipment power limit; the module has good scalability, supports multiple energy-saving goals and user priority configurations, and can flexibly adjust the control strategy according to different scenarios to improve energy utilization efficiency and system stability.

[0210] The user interaction module is used to receive the power distribution optimization plan, generate a visual power consumption status report and operation suggestions, and present them to the user through the human-machine interface. It also receives the personalized configuration parameters entered by the user and feeds them back to the control strategy module;

[0211] Furthermore, the power distribution optimization plan is converted into visual graphical interface elements, including electrical appliance operation status indicators, power change curves, and energy consumption statistics charts;

[0212] Provide operational suggestions in text form;

[0213] Users enter personalized configuration parameters through the touch screen or voice interface, including appliance priority adjustment, energy-saving preference settings, and manual recovery instructions;

[0214] Personalized configuration parameters are encapsulated as configuration signals;

[0215] The configuration signal is fed back to the control strategy module to optimize the next round of regulation decisions;

[0216] It should be noted that the power distribution optimization results and power consumption status reports are presented through a visual interface, allowing users to intuitively understand the current power consumption situation and system recommendations; graphical interface elements such as operating status indicators, power curves and energy consumption charts enhance the information communication effect; users can input personalized configuration parameters through the touch screen or voice interface to achieve independent adjustment of appliance priorities, energy-saving preferences, etc.; after feedback to the control strategy module, the system can optimize the next round of regulation decisions accordingly, truly realize human-machine collaboration and on-demand regulation, and improve user experience and system intelligence.

[0217] This embodiment also provides a computer device suitable for the real-time monitoring and adaptive control system of smart home electricity safety, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the real-time monitoring and adaptive control system of smart home electricity safety proposed in the above embodiment.

[0218] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0219] This embodiment also provides a storage medium having a computer program stored thereon. When the program is executed by the processor, the real-time monitoring and adaptive control system for smart home electricity safety as proposed in the above embodiment is implemented. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0220] In summary, the present invention realizes real-time perception of the full-dimensional operating status of household electrical appliances by deploying multiple types of sensors at multiple locations and performing continuous data collection. It extracts local features of the original sensor data through CNN to effectively capture instantaneous mutations and short-term anomalies. It models the long-term dependence of time series through LSTM modeling to accurately identify slowly degrading anomalies. It improves the adaptability and interpretability of the model to different types of anomalies by fusing the L2 norm and the weighted feature sum nonlinear discriminant function. It realizes differentiated handling of anomalies of different severity by formulating hierarchical safety response rules. It ensures the traceability and execution consistency of control commands through structured control instruction design. It replaces direct power off with a current limiting mechanism to avoid unnecessary interruptions while ensuring safety.

[0221] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. Real-time monitoring and adaptive control system for smart home electricity safety, characterized by: include: Sensor acquisition module, anomaly recognition module, local response module, network modeling module, risk prediction module, control strategy module and user interaction module; The sensor acquisition module is used to collect data on the operating status of household appliances through current, voltage, temperature, and power sensors to generate an original sensor data set; The anomaly recognition module is used to build a classifier model based on a deep neural network, extract features from the original sensor data set, generate feature vectors, and perform an anomaly probability assessment on the feature vectors based on the classifier model to output a device operation anomaly signal; The local response module is used to execute preset safety response rules on the edge computing unit based on the device operation abnormality signal and generate a local power-off or current-limiting control instruction; The network modeling module is used to construct a graph structure model based on the physical connection topology and usage dependency between electrical appliances and use a graph neural network to update node embeddings to generate an electrical appliance correlation representation matrix; The risk prediction module is used to combine the equipment operation abnormality signal with the electrical appliance correlation representation matrix, simulate the fault propagation path, predict the potential risk level, and output a risk warning signal; The control strategy module is used to dynamically adjust the operating parameters of non-abnormal electrical appliances based on local power-off or current-limiting control instructions and risk warning signals in combination with a load balancing algorithm to generate a power distribution optimization plan; The user interaction module is used to receive the power distribution optimization plan, generate a visual power consumption status report and operation suggestions, and present them to the user through the human-computer interface, receive personalized configuration parameters input by the user, and feed them back to the control strategy module.

2. The smart home electricity safety real-time monitoring and adaptive control system according to claim 1, characterized in that: The specific steps of collecting data on the operating status of household appliances through current, voltage, temperature, and power sensors to generate original sensor data sets are as follows: Install current sensors, voltage sensors, temperature sensors, and power sensors inside household distribution boxes, sockets, and electrical appliances; Deploy the above-mentioned sensors at the node locations where each appliance is connected to the circuit; The output signals of each sensor are continuously collected with a fixed sampling period Δt; Get the current value, voltage value, temperature value and power value at timestamp t; Convert the collected analog signal into a digital signal; The four types of collected data are normalized separately, and the expressions are: Among them, x represents the original value at the current moment, x min Indicates the historical minimum value of the parameter, x max Indicates the historical maximum value of the parameter, x norm is the normalized x value; The normalized four-dimensional parameters are combined into a structured raw sensing dataset in time sequence.

3. The smart home electricity safety real-time monitoring and adaptive control system according to claim 2, characterized in that: The method of constructing a classifier model based on a deep neural network, extracting features from the original sensor data set, generating a feature vector, and performing an abnormality probability assessment on the feature vector based on the classifier model to output a device operation abnormality signal is as follows: Convolutional neural network (CNN) is used to extract local features from the original sensor data set obtained from the sensor acquisition module. By setting the convolution kernel and bias term, a sliding window convolution operation is performed on the data in the original sensor data set to obtain the primary feature matrix M1, which is expressed as: M1=ReLU(W conv *D0+b conv ); Among them, ReLU(.) represents the activation function, W conv is the trainable convolution weight matrix, b conv is the bias term, D0 is the input tensor; Long short-term memory network (LSTM) is used to model the temporal dependency of the primary feature matrix; The LSTM network processes each row of M1 in turn, updating the hidden state and cell state; Take the hidden state of the last time step as the semantic representation of the entire time series, recorded as the feature vector; A nonlinear discriminant function is used to evaluate the abnormal probability of the feature vector and output the abnormal signal of the equipment operation. The discriminant function expression is: Among them, a is the abnormal probability, α is the global weight coefficient, β is the local feature weighting factor, F j is the jth component of the eigenvector F, w j is the weight corresponding to the jth feature, F∥2 is the L2 norm of the feature vector, indicating the overall fluctuation intensity, and e is the base of the natural logarithm; Set the abnormality judgment threshold θ. If a≥a, it is determined that the equipment has an abnormal operation and the equipment operation abnormality signal A=1 is output; otherwise, the output signal A=0 is output. Among them, θ is the abnormal discrimination threshold preset by the system, and A is the abnormal signal.

4. The smart home electricity safety real-time monitoring and adaptive control system according to claim 3, characterized in that: Based on the abnormal operation signal of the device, the preset safety response rules are executed on the edge computing unit to generate a local power-off or current-limiting control instruction. The specific steps are as follows: Use the preset security response rule set to match and judge the equipment operation abnormality signal A obtained from the abnormality identification module to determine the type of security response action to be performed; Among them, the equipment operation abnormality signal A is a Boolean variable, A indicates that the current electrical appliance has an abnormal operation; the safety response rule set includes multiple rule items R1 to R n , each rule item consists of two parts: The abnormal level describes the condition; Corresponding action strategy; After judging by the above rules, output action type t; Adopting relay control algorithm to encapsulate the action type t into instructions and generate local power-off or current-limiting control instructions; The local power-off or current-limiting control instruction is a structured data packet, including the appliance's unique identifier, action type, execution timestamp, and target current value during current limiting. If the action type t is "power off", the control instruction C cmd The format is: C cmd ={ID, type = power off, time = t exec }; If the action type t is "current limiting", the control instruction C cmd The format is: C cmd ={ID, type = current limit, time = t exec ,value=I limit }; Among them, ID is the appliance number, t exec is the current system timestamp, I limit is the target current limit value; The target current limit value I limit The calculation expression is: I limit =γ·I rated ; Among them, γ is the current limiting coefficient, I rated is the rated working current of the appliance; Use the communication protocol to control the local power off or current limiting control instruction C cmd Send it to the local controller, which drives the relay or smart switch to perform the corresponding operation; Among them, the local controller receives C cmd Then, parse out the appliance ID and action type; Find the corresponding physical control port address according to the appliance ID; Depending on the action type, do one of the following: If type = power off, a shutdown signal is sent to the relay to cut off the power supply to the appliance; If type = current limiting, a current limiting instruction is sent to the controllable current regulator to adjust its current output to I limit .

5. The smart home electricity safety real-time monitoring and adaptive control system according to claim 4, characterized in that: Based on the physical connection topology and usage dependency between electrical appliances, a graph structure model is constructed and a graph neural network is used to update the node embedding to generate an electrical appliance correlation representation matrix. The specific steps are as follows: Model the power connection relationship and functional dependency relationship between household appliances, and then build a graph structure model; The graph convolutional network (GCN) is used to learn the embedded representation of nodes in the graph structure and generate an appliance correlation representation matrix. Among them, the initial node embedding vector By the attribute vector x v Obtained through linear transformation, the expression is: Where W0 is the trainable weight matrix. b0 is the bias term. The l-th layer graph convolution is calculated as follows: Among them, σ is the activation function, N(v) represents the set of neighbor nodes of node v, and W l is the trainable weight matrix of layer l, b l is the bias term of the lth layer, Represents the embedding vector of neighbor node u in the previous layer; After l layers of propagation, the node embedding vector is obtained Embed all nodes into an electrical appliance relevance representation matrix: Where n is the total number of nodes, M rep It is the electrical appliance correlation representation matrix; Electrical appliance correlation representation matrix M rep Perform analysis and extract the influence weights between appliances, including the similarity scores of appliances ij and the vulnerability factor f of the electrical appliance i , the expression is: in, and are the final embedding vectors of appliances i and j respectively, T represents the transpose operation, c i is the number of historical failures of appliance i, k is the adjustment coefficient, P rated,i is the rated power of electrical appliance i, P avg is the average power value of the system.

6. The smart home electricity safety real-time monitoring and adaptive control system according to claim 5, characterized in that: The specific steps of combining the equipment operation abnormality signal with the electrical appliance correlation representation matrix, simulating the fault propagation path, predicting the potential risk level, and outputting the risk warning signal are as follows: The real-time monitoring module is used to collect the operating status data of each electrical equipment and generate equipment operation abnormality signal S abn ; Among them, S abn is a vector, each element s i The expression for the abnormality score of an electrical device is: s i =α·(P curr,i -P rated,i )+β·f freq,i ; Among them, P curr,i is the current power of appliance i, P rated,i is the rated power of appliance i, f freq,i is the historical fault frequency of appliance i, α and β are adjustment coefficients, which are used to balance the impact of current power overload and historical faults on the anomaly score; Based on the abnormal signal S abn and electrical appliance correlation representation matrix M rep , build a fault propagation model; By defining the fault diffusion function, the fault propagation probability between electrical appliances is calculated, and the expression is: Among them, n represents the total number of electrical appliances, m ij It's M rep Elements in the matrix, P fp (i, j) represents the probability that the fault propagates to appliance,j after an abnormality occurs in appliance,i; Based on P fp (i, j), simulate the propagation path of faults in the electrical network and predict the potential risk level; For each appliance i, calculate the cumulative risk value R after it is affected by other appliances riski , the expression is: According to R risk,i Risk levels are divided according to the size of the risk factor; If R risk,i <T1, the risk level is low risk; If T1 ≤ R risk,i < and T2, then the risk level is medium risk; If R risk,i ≥T2, the risk level is high risk, where T1 and T2 are preset thresholds; Output corresponding risk warning signals according to the risk level of each electrical appliance.

7. The smart home electricity safety real-time monitoring and adaptive control system according to claim 6, characterized in that: Based on the local power-off or current-limiting control instructions and risk warning signals, combined with the load balancing algorithm, the operating parameters of non-abnormal electrical appliances are dynamically adjusted to generate a power distribution optimization plan. The specific steps are as follows: Receive control instructions from the local response module and risk warning signals from the risk prediction module; Get the current system total load and available capacity; For appliances that are not abnormal and have not issued warnings, new operating parameters are assigned based on the user-set priority and energy consumption; The constraints for solving the optimal scheduling problem using linear programming are: min∑ i (p i -p i ′) 2 ; Among them, p i is the original power, p i ′ is the adjusted power, The maximum power allowed for the device; Get the power distribution optimization plan, including the new operating parameters of each electrical appliance.

8. The smart home electricity safety real-time monitoring and adaptive control system according to claim 7, characterized in that: The receiving power distribution optimization plan, generating a visual power consumption status report and operation suggestions, and presenting them to the user through the human-machine interface, receiving the personalized configuration parameters input by the user, and feeding them back to the control strategy module, specifically includes the following steps: Convert the power distribution optimization plan into a visual graphical interface element, including electrical appliance operating status indicator lights, power change curves, and energy consumption statistics charts; Provide operational suggestions in text form; Users enter personalized configuration parameters through the touch screen or voice interface, including appliance priority adjustment, energy-saving preference settings, and manual recovery instructions; The personalized configuration parameters are encapsulated as configuration signals; The configuration signal is fed back to the control strategy module to optimize the next round of regulation decisions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the smart home electricity safety real-time monitoring and adaptive control system according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the smart home electricity safety real-time monitoring and adaptive control system according to any one of claims 1 to 8 are implemented.

Images