Power equipment remote monitoring system based on homomorphic encryption

Abstract

The invention discloses a power equipment remote monitoring system based on homomorphic encryption, and the system comprises a key management module which is used for generating multiple layers of key pairs and correspondingly distributing the key pairs to all sensor nodes and edge equipment; the data acquisition module is used for acquiring various state data and performing preliminary encryption on the acquired data by using a first-layer public key; the edge calculation module is used for performing preliminary data processing and anomaly detection on the preliminarily encrypted data, and performing secondary encryption on the detected data by using a second-layer public key; the monitoring center module is used for receiving the secondarily encrypted data, calculating and analyzing ciphertext data by using a homomorphic encryption technology, decrypting layer by layer by using a second-layer private key and a first-layer private key, performing deep analysis on the decrypted data through a machine learning model, identifying abnormal conditions and performing real-time prediction; according to the method, the data security, the calculation efficiency, the real-time performance and the intelligent level of the remote monitoring system of the power equipment are remarkably improved, and the normal operation and the security of the power equipment are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment monitoring, and more particularly to a remote monitoring system for power equipment based on homomorphic encryption. Background Art

[0002] With the continuous development of the power system, the remote monitoring of power equipment has become an important means to ensure the stable operation of the power grid. At present, traditional remote monitoring systems for power equipment mainly rely on traditional data encryption technologies and centralized data processing methods. The data collected by sensor nodes is transmitted to the monitoring center through a secure communication network. The monitoring center decrypts and analyzes the data, and then makes decisions based on the analysis results.

[0003] Although the existing remote monitoring systems for power equipment can achieve remote monitoring of power equipment, there are still the following disadvantages:

[0004] Insufficient data security: Traditional encryption technologies may be at risk of being intercepted and cracked during data transmission, especially in an open network environment;

[0005] Low computing efficiency: The centralized data processing method leads to an excessive computing burden on the monitoring center, affecting the real-time performance and response speed of the system;

[0006] Lack of intelligence: Existing systems usually rely on preset thresholds for anomaly detection, lacking the ability to identify complex data patterns and being difficult to handle sudden abnormal situations.

[0007] Complex key management: Fixed keys are easily stolen, and the key distribution and update mechanisms are not flexible enough.

[0008] Therefore, how to ensure the data security and privacy protection in the data transmission and processing of the remote monitoring system for power equipment, while simplifying key management to improve the computing efficiency, real-time performance and intelligence level of the system is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0009] In view of this, the present invention provides a remote monitoring system for power equipment based on homomorphic encryption to solve some of the technical problems mentioned in the background art.

[0010] To achieve the above object, the present invention adopts the following technical solutions:

[0011] A remote monitoring system for power equipment based on homomorphic encryption, comprising a data acquisition module, an edge computing module, a monitoring center module and a key management module;

[0012] A key management module, which is used to generate multi-layer key pairs and distribute the corresponding multi-layer public keys to all sensor nodes and edge devices;

[0013] A data acquisition module, which is used to collect various status data through various sensors installed on power equipment and preliminarily encrypt the collected data using the first-layer public key;

[0014] An edge computing module, which is used to receive the preliminarily encrypted device status data, perform preliminary data processing and anomaly detection, and perform secondary encryption on the detected data using the second-layer public key;

[0015] A monitoring center module, which is used to receive the secondarily encrypted data from the edge device, perform calculations and analyses on the ciphertext data using homomorphic encryption technology, decrypt the ciphertext data layer by layer using the second-layer private key and the first-layer private key, and perform in-depth analysis on the decrypted data through a machine learning model to identify abnormal situations and perform real-time predictions.

[0016] Preferably, the remote monitoring system for power equipment based on homomorphic encryption further includes a data transmission module, which is used to transmit the encrypted data from each sensor node to the edge device and the monitoring center using a secure communication network.

[0017] Preferably, the key management module includes a key generation unit, a key distribution unit, and a key revocation and recovery unit;

[0018] The key generation unit is used to regularly generate new multi-layer key pairs (P1, S1) and (P2, S2);

[0019] The key distribution unit is used to distribute the new keys to all sensor nodes and edge devices using a secure key distribution protocol;

[0020] The key revocation and recovery unit is used to immediately revoke the key permissions of a certain node or device when it fails or is invaded, and provide a key recovery mechanism.

[0021] Preferably, the data acquisition module includes sensor nodes and a preliminary encryption unit;

[0022] The sensor nodes are various sensors installed on power equipment and are used to collect data including but not limited to temperature, current, and voltage.

[0023] The preliminary encryption unit is an encryption module built into each sensor node and is used to preliminarily encrypt the collected data using the first-layer public key P1.

[0024] Preferably, the edge computing module is an edge device deployed near the sensor nodes, and the edge device is built with a secondary encryption unit and a preliminary processing unit;

[0025] A preliminary processing unit for performing preliminary statistical analysis and anomaly detection on the preliminarily encrypted data to reduce the burden on the central server;

[0026] A secondary encryption unit, which is an encryption module built into the edge device, for using the second-layer public key P2 to perform secondary encryption on the data after preliminary statistical analysis and anomaly detection.

[0027] Preferably, the monitoring center module includes a data receiving unit, a homomorphic computing unit, a decryption unit, and a machine learning model;

[0028] The data receiving unit for receiving the secondary encrypted data from the edge device;

[0029] The homomorphic computing unit for performing statistical analysis, anomaly detection, data preprocessing, and feature extraction on the ciphertext data using homomorphic encryption technology;

[0030] The decryption unit for using the second-layer private key S2 and the first-layer private key S1 to decrypt the ciphertext data layer by layer to restore the original data of the power equipment for prediction model training;

[0031] The machine learning model for constructing and training a prediction model based on the restored original data of the power equipment and the extracted features, identifying the normal operation mode and abnormal mode of the equipment, and performing real-time prediction and anomaly detection through the trained prediction model.

[0032] In this embodiment, when the trained prediction model performs real-time prediction, only the feature vector extracted by the homomorphic computing unit needs to be input into the model without decrypting the original data.

[0033] Preferably, the specific content of the homomorphic computing unit is as follows:

[0034] Statistical analysis: Performing statistical analysis of sum, average value, and variance in the ciphertext state;

[0035] Anomaly detection: Performing threshold comparison, clustering analysis, and trend analysis in the ciphertext state to preliminarily identify abnormal data points;

[0036] Data preprocessing: Performing normalization and data cleaning in the ciphertext state to prepare the data for input into the LSTM model;

[0037] Feature extraction: Extracting the basic statistical features and time series features of the data in the ciphertext state and decrypting the variance to calculate the standard deviation for input into the LSTM model.

[0038] Preferably, the monitoring center module further includes an alarm unit and a data storage unit;

[0039] An alarm unit, which is used to issue an alarm when the machine learning model detects an abnormal situation and provide possible causes and recommended measures.

[0040] A data storage unit, which is used to use distributed storage technology to disperse and store the encrypted data on multiple edge devices and central servers.

[0041] Preferably, the trained prediction model is an LSTM model, specifically:

[0042] Y pred = LSTM(X real-time )

[0043] Where Y pred is the predicted value of the model, that is, the device data at a future time point, and X real-time is the real-time input data at the current moment;

[0044] Anomaly detection is to compare the predicted value Y pred with the actual value X real-time to determine whether there is an anomaly;

[0045] The anomaly detection formula is:

[0046] if∣Y pred -X real-time ∣>threshold

[0047] Where threshold is a pre-set anomaly threshold, which is used to determine whether the difference between the predicted value and the actual value exceeds the allowable range. If the difference between the predicted value and the actual value exceeds the set threshold, the system triggers an alarm mechanism.

[0048] Preferably, the specific content of constructing and training the prediction model based on the historical data of the power equipment collection is:

[0049] Input feature data set: Obtain the decrypted and restored original data and perform preprocessing. Combine the extracted features and divide the feature data into a training set, a validation set and a test set;

[0050] Model construction: The LSTM model includes an input layer, an LSTM layer and an output layer. The LSTM layer includes a forget gate, an input gate, a candidate memory unit, a memory unit update, an output gate and an output at the current moment;

[0051] Model training: By establishing a mean squared error MSE loss function and an Adam optimizer, use the training set data to train the model; use the validation set data to evaluate the performance of the model and adjust the hyperparameters to optimize the model;

[0052] Model evaluation: Use the test set data to evaluate the performance metrics of the model, draw a comparison chart of the predicted value and the true value, and show the prediction effect of the model.

[0053] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a remote monitoring system for power equipment based on homomorphic encryption. By introducing a multi-layer encryption and decryption mechanism, homomorphic encryption, edge computing, an intelligent early warning system, and dynamic key management, the data security, computing efficiency, real-time performance, and intelligence level of the remote monitoring system for power equipment are significantly improved. It can not only effectively protect the security of power equipment data, but also timely detect and handle abnormal situations to ensure the normal operation and safety of power equipment. Brief Description of the Drawings

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0055] Figure 1 Schematic diagram of a remote monitoring system for power equipment based on homomorphic encryption provided by the present invention;

[0056] Figure 2 Schematic diagram of the monitoring center module of the system provided by the present invention;

[0057] Figure 3 Schematic diagram of the homomorphic computing unit of the monitoring center module provided by the present invention. Detailed Embodiments

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0059] An embodiment of the present invention discloses a remote monitoring system for power equipment based on homomorphic encryption, such as Figure 1 , including a data acquisition module, an edge computing module, a monitoring center module, and a key management module;

[0060] The key management module is used to generate multi-layer key pairs and distribute the corresponding multi-layer public keys to all sensor nodes and edge devices;

[0061] The data acquisition module is used to collect various state data through various sensors installed on power equipment and perform preliminary encryption on the collected data using the first-layer public key;

[0062] An edge computing module, which is used to receive the preliminarily encrypted device status data, perform preliminary data processing and anomaly detection, and perform secondary encryption on the detected data using the second-layer public key;

[0063] A monitoring center module, which is used to receive the doubly encrypted data from the edge devices, perform calculations and analyses on the ciphertext data using homomorphic encryption technology, decrypt the ciphertext data layer by layer using the second-layer private key and the first-layer private key, and perform in-depth analysis on the decrypted data through a machine learning model to identify anomalies and make real-time predictions.

[0064] To further implement the above technical solution, a remote monitoring system for power equipment based on homomorphic encryption further includes a data transmission module, which is used to transmit the encrypted data from each sensor node to the edge devices and the monitoring center using a secure communication network.

[0065] To further implement the above technical solution, the key management module includes a key generation unit, a key distribution unit, and a key revocation and recovery unit;

[0066] The key generation unit is used to periodically generate new multi-layer key pairs (P1, S1) and (P2, S2);

[0067] The key distribution unit is used to distribute the new keys to all sensor nodes and edge devices using a secure key distribution protocol;

[0068] The key revocation and recovery unit is used to immediately revoke the key permissions of a certain node or device when it fails or is invaded, and provide a key recovery mechanism.

[0069] To further implement the above technical solution, the data acquisition module includes sensor nodes and a preliminary encryption unit;

[0070] The sensor nodes are various sensors installed on the power equipment, which are used to collect data including but not limited to temperature, current, and voltage.

[0071] The preliminary encryption unit is an encryption module built into each sensor node, which is used to perform preliminary encryption E1(D) on the collected data using the first-layer public key P1.

[0072] To further implement the above technical solution, the edge computing module is an edge device deployed near the sensor nodes, and the edge device is built with a secondary encryption unit and a preliminary processing unit;

[0073] The preliminary processing unit is used to perform preliminary statistical analysis and anomaly detection on the preliminarily encrypted data to reduce the burden on the central server;

[0074] The secondary encryption unit is an encryption module built into the edge device, which is used to perform secondary encryption E2(D) on the data after preliminary statistical analysis and anomaly detection using the second-layer public key P2.

[0075] To further implement the above technical solution, as Figure 2 shown, the monitoring center module includes a data receiving unit, a homomorphic computing unit, a decryption unit, and a machine learning model;

[0076] The data receiving unit is used to receive the secondary encrypted data from the edge device;

[0077] The homomorphic computing unit is used to perform statistical analysis, anomaly detection, data preprocessing, and feature extraction on the ciphertext data using homomorphic encryption technology;

[0078] The decryption unit is used to perform layer-by-layer decryption on the ciphertext data using the second-layer private key S2 and the first-layer private key S1 to restore the original data;

[0079] The machine learning model is used to construct and train a prediction model based on the restored original data of the power equipment and the extracted features, identify the normal operation mode and abnormal mode of the equipment, and perform real-time prediction and anomaly detection through the trained prediction model.

[0080] To further implement the above technical solution, as Figure 3 shown, the specific content of the homomorphic computing unit is:

[0081] Statistical analysis: Perform statistical analysis of sum, average, and variance in the ciphertext state;

[0082] Anomaly detection: Perform threshold comparison, clustering analysis, and trend analysis in the ciphertext state to initially identify abnormal data points;

[0083] Data preprocessing: Perform normalization and data cleaning in the ciphertext state to prepare the data for input into the LSTM model;

[0084] Feature extraction: Extract the basic statistical features and time series features of the data in the ciphertext state, and decrypt the variance to calculate the standard deviation for input into the LSTM model, including time series features, trend features, etc.

[0085] Specifically:

[0086] Statistical analysis:

[0087] Utilize the additive homomorphic property of homomorphic encryption to directly sum multiple data points in the ciphertext state:

[0088]

[0089] Calculate the average value by multiplying by the reciprocal:

[0090]

[0091] Calculate the square of the difference between each data point and the mean:

[0092] E2((D i - μ) 2 ) = E2(D i - μ) × E2(D i - μ)

[0093] Accumulate all the squared terms and calculate the average to obtain the variance:

[0094]

[0095] Anomaly detection:

[0096] Use a specific homomorphic encryption scheme (such as BGN encryption) to achieve threshold comparison in the ciphertext state. Set an encrypted threshold E2(threshold), and compare it with each data point E2(D i ). Mark the data points that exceed the threshold as anomalies:

[0097] if E2(D i ) > E2(threshold) then flag as anomaly

[0098] Trend analysis:

[0099] Linear regression. Perform simple linear regression analysis in the ciphertext state to calculate the slope and intercept:

[0100]

[0101] Normalization: Calculate the minimum value E2(min(D)) and the maximum value E2(max(D)) in the ciphertext state, and use the linear transformation formula for normalization; Remove outliers: Perform preliminary anomaly detection in the ciphertext state, mark and remove outliers;

[0102] In feature extraction, the basic statistical features include the mean and variance calculated in the ciphertext state, the time series features include the moving average calculated in the ciphertext state, and the frequency components of the Fourier transform calculated in the decrypted state, i.e., the periodic features E2(FFT(D1, D2,..., D n ));

[0103] Decrypt the variance E2(Var(D)) calculated in the ciphertext to obtain the plaintext variance Var(D):

[0104]

[0105] Calculate the standard deviation after decryption

[0106] Calculating the standard deviation by decrypting the variance under the ciphertext ensures that the data remains encrypted throughout the calculation process. Only the final variance needs to be decrypted, reducing the security risk; the calculation of the variance can be efficiently completed in the ciphertext state, and the calculation of the standard deviation only requires taking the square root of a single value (the variance), reducing the computational overhead.

[0107] The homomorphic calculation unit performs data preprocessing and feature extraction in the ciphertext state to ensure that the data remains encrypted during transmission and processing. Finally, the decryption unit decrypts the ciphertext data into plaintext data and inputs it into the LSTM model for training and prediction. This design not only improves the security and privacy protection capabilities of the data but also enhances the real-time performance and intelligence level of the system.

[0108] To further implement the above technical solution, the monitoring center module also includes an alarm unit for issuing an alarm when the machine learning model detects an abnormal situation and providing possible causes and recommended measures.

[0109] To further implement the above technical solution, the monitoring center module also includes a data storage unit for using distributed storage technology to disperse and store the encrypted data on multiple edge devices and central servers.

[0110] To further implement the above technical solution, the trained prediction model is an LSTM model, specifically:

[0111] Y pred = LSTM(X real-time )

[0112] where Y pred is the predicted value of the model, i.e., the device data at a future time point, and X real-time is the real-time input data at the current moment;

[0113] Anomaly detection is to compare the predicted value Y pred with the actual value X real-time to determine whether there is an anomaly;

[0114] The anomaly detection formula is:

[0115] if ∣Y pred - X real-time ∣ > threshold

[0116] where threshold is a pre-set anomaly threshold used to determine whether the difference between the predicted value and the actual value exceeds the allowable range. If the difference between the predicted value and the actual value exceeds the set threshold, the system triggers an alarm mechanism.

[0117] To further implement the above technical solution, the specific content of constructing and training a prediction model based on the collected historical data of power equipment is as follows:

[0118] Input feature dataset: Obtain the decrypted and restored original data and perform preprocessing. Combine the extracted features and divide the feature data into a training set, a validation set, and a test set;

[0119] The data vector x at each time point t t , for example, x t = [T t , I t , V t , where T t is temperature, I t is current, and V t is voltage;

[0120] Clean and normalize the data to remove outliers and missing values;

[0121] Model construction: The LSTM model includes an input layer, an LSTM layer, and an output layer. The LSTM layer includes a forget gate, an input gate, a candidate memory cell, memory cell update, an output gate, and the output at the current moment;

[0122] Specifically:

[0123] Input layer:

[0124] Define the shape of the input data, such as the time step T and the number of features F; the input data shape X ∈ R T×F ;

[0125] LSTM layer:

[0126] Add one or more LSTM layers and set appropriate number of units N and return sequence parameter;

[0127] LSTM layer calculation formula:

[0128] Forget gate:

[0129] f t = σ(W f · [h t-1 , x t + b f )

[0130] The forget gate determines the information discarded from the memory cell c t-1 at the previous time step. In the monitoring of power equipment, it determines to forget the historical data that no longer affects the current equipment state;

[0131] Input gate:

[0132] i t = σ(W i · [h t-1 , x t + b i )

[0133] The input gate controls how new information is added to the memory cell. For power equipment, it determines whether the new data collected at the current moment is important and whether it should be updated to the device's historical status record;

[0134] Candidate memory cell:

[0135]

[0136] The candidate memory cell is new information calculated based on the current input x t and the previous hidden state ht-1. In power equipment monitoring, this is to predict the possible change trend of the equipment based on the current measurement value;

[0137] Memory cell update:

[0138]

[0139] The memory cell C t is the information retained after being filtered by the forget gate from the previous memory cell C t-1 , plus the new information screened by the input gate, representing a comprehensive evaluation of the device state, considering the influence of historical data and current data;

[0140] Output gate:

[0141] o t = σ(W o · [h t-1 , x t + b o )

[0142] The output gate determines how the information in the memory cell C t is transformed into the output h t at the current moment. In power equipment monitoring, it is used to determine which information is important for predicting future device behavior;

[0143] Current moment output:

[0144] h t = o t ⊙ tanh(C t ) The final output h t is the non - linear transformation of the memory cell C t regulated by the output gate o t . In power equipment monitoring, h tFor predicting the device state at the next time point or as part of anomaly detection;

[0145] where f t is the output of the forget gate, i t is the output of the input gate, is the candidate cell state, C t is the cell state, o t is the output of the output gate, h t is the hidden state, W f 、W i 、W C 、W o are weight matrices, b f 、b i 、b C 、b o are bias vectors, x t is the input data at time t, h t-1 is the hidden state at time t - 1, σ is the Sigmoid activation function, tanh is the hyperbolic tangent activation function, and ⊙ is the element-wise multiplication;

[0146] Output layer calculation formula:

[0147] Y = W o H t + b o

[0148] where Y is the final output of the model, is the predicted device state, W o is the weight matrix of the output layer, which determines how the hidden state H t affects the final output. The weight matrix W o learns how to extract useful features from the hidden state, H t is the hidden state at the current time, which contains the output information of the LSTM cell, b o is the bias term of the output layer, which helps the model better fit the data. The bias term b o helps the model better adjust the final output in the initial stage.

[0149] Model training: By establishing the mean squared error MSE loss function and the Adam optimizer, use the training set data to train the model; use the validation set data to evaluate the performance of the model and adjust the hyperparameters to optimize the model;

[0150] The loss function is:

[0151]

[0152] where Loss is the loss value of the model, which is used to measure the difference between the model prediction value and the actual value, N is the number of samples, Y iis the predicted value of the i-th sample, is the actual value of the i-th sample;

[0153] The optimizer is:

[0154]

[0155] where θ t+1 are the updated model parameters, and θ t are the current model parameters. α is the learning rate, which controls the step size of parameter update. is the gradient of the loss function with respect to the model parameters, indicating the direction of parameter update. is the exponentially weighted moving average of the squared gradients, used to adjust the learning rate. ∈ is a small constant to prevent division by zero errors and ensure numerical stability.

[0156] Model evaluation: Use the test set data to evaluate the performance metrics of the model, and draw a comparison graph of the predicted values and the true values to show the prediction effect of the model.

[0157] A remote monitoring system for power equipment based on homomorphic encryption proposed by the present invention has the following specific technical effects:

[0158] 1) Enhance data security:

[0159] Multi-layer encryption: Through preliminary encryption and secondary encryption, ensure that data is not stolen or tampered with during transmission;

[0160] Homomorphic encryption: Perform calculations on the ciphertext data at the monitoring center to ensure that the data remains encrypted during processing, further improving data security.

[0161] Key management: Regularly update the keys and use a secure key distribution protocol to ensure the security and effectiveness of the keys.

[0162] 2) Improve computational efficiency and real-time performance:

[0163] Edge computing: Edge devices perform preliminary data processing and anomaly detection, reducing the computational burden on the central server and improving the overall computational efficiency of the system.

[0164] Distributed storage: Disperse the data storage on multiple edge devices and the central server to improve the read / write speed and reliability of the data.

[0165] 3) Enhance intelligence and prediction ability:

[0166] LSTM model: Use the LSTM model to train historical data, identify the normal operation mode and abnormal mode of the device, perform real-time prediction and anomaly detection, and improve the intelligence level of the system.

[0167] 4) Implement dynamic key management:

[0168] Modular design: The system consists of multiple modules with clear functions for each module, making it easy to maintain and upgrade; Dynamic key management: Supports the dynamic generation, distribution, and update of keys to ensure the long-term security and stability of the system; Distributed architecture: Adopts a distributed storage and computing architecture to support the horizontal expansion of the system and meet the monitoring requirements of large-scale power equipment.

[0169] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same and similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and reference can be made to the description in the method part for the relevant parts.

[0170] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A remote monitoring system for power equipment based on homomorphic encryption, characterized in that: It includes data acquisition module, edge computing module, monitoring center module and key management module; The key management module is used to generate multi-layer key pairs and distribute the multi-layer public keys to all sensor nodes and edge devices; A data acquisition module is used to collect various status data through various sensors installed on the power equipment, and to preliminarily encrypt the collected data using the first layer public key; The edge computing module is used to receive the initially encrypted device status data and perform preliminary data processing and anomaly detection, and use the second-layer public key to re-encrypt the detected data; The monitoring center module is used to receive secondary encrypted data from edge devices, use homomorphic encryption technology to calculate and analyze the ciphertext data, use the second-layer private key and the first-layer private key to decrypt the ciphertext data layer by layer, and perform in-depth analysis of the decrypted data through machine learning models to identify abnormal situations and make real-time predictions.

2. According to claim 1, a remote monitoring system for electric power equipment based on homomorphic encryption is characterized in that: It also includes a data transmission module for transmitting encrypted data from each sensor node to the edge device and the monitoring center using a secure communication network.

3. According to claim 1, a remote monitoring system for electric power equipment based on homomorphic encryption is characterized in that: The key management module includes a key generation unit, a key distribution unit and a key revocation and recovery unit; A key generation unit, for periodically generating new multi-layer key pairs (P1, S1) and (P2, S2); A key distribution unit, used to distribute new keys to all sensor nodes and edge devices using a secure key distribution protocol; The key revocation and recovery unit is used to immediately revoke the key permissions of a node or device when it fails or is invaded, and provide a key recovery mechanism.

4. According to claim 1, a remote monitoring system for electric power equipment based on homomorphic encryption is characterized in that: The data acquisition module includes sensor nodes and preliminary encryption units; Sensor nodes are various sensors installed on power equipment to collect data including but not limited to temperature, current, and voltage; The preliminary encryption unit is a built-in encryption module in each sensor node, which is used to perform preliminary encryption on the collected data using the first-layer public key P1.

5. According to claim 1, a remote monitoring system for electric power equipment based on homomorphic encryption is characterized in that: The edge computing module is an edge device deployed close to the sensor node, with a built-in secondary encryption unit and a preliminary processing unit. A preliminary processing unit, which is used to perform preliminary statistical analysis and anomaly detection on the preliminary encrypted data to reduce the burden on the central server; The secondary encryption unit is an encryption module built into the edge device, which is used to perform secondary encryption on the data after preliminary statistical analysis and anomaly detection using the second-layer public key P2.

6. According to claim 1, a remote monitoring system for electric power equipment based on homomorphic encryption is characterized in that: The monitoring center module includes a data receiving unit, a homomorphic computing unit, a decryption unit, and a machine learning model; A data receiving unit, used for receiving secondary encrypted data from an edge device; Homomorphic computing unit, used to perform statistical analysis, anomaly detection, data preprocessing and feature extraction on ciphertext data using homomorphic encryption technology; A decryption unit, used to use the second-layer private key S2 and the first-layer private key S1 to decrypt the ciphertext data layer by layer to restore the original data; The machine learning model is used to build and train the prediction model based on the recovered raw data of the power equipment and the extracted features, identify the normal operation mode and abnormal mode of the equipment, and perform real-time prediction and anomaly detection through the trained prediction model.

7. The remote monitoring system for electric power equipment based on homomorphic encryption according to claim 6 is characterized in that: The specific contents of the homomorphic computing unit are: Statistical analysis: Perform statistical analysis of sum, mean and variance in encrypted state; Anomaly detection: Perform threshold comparison and trend analysis in the encrypted state to preliminarily identify abnormal data points; Data preprocessing: normalize and clean the data in ciphertext state to prepare data for input into the LSTM model; Feature extraction: Extract the basic statistical features and time series features of the data in the encrypted state, and decrypt the variance to calculate the standard deviation for the input of the LSTM model.

8. According to claim 6, a remote monitoring system for electric power equipment based on homomorphic encryption is characterized in that: The monitoring center module also includes an alarm unit and a data storage unit; An alarm unit, which is used to issue an alarm when the machine learning model detects an abnormal situation and provide possible causes and recommended measures; The data storage unit is used to use distributed storage technology to disperse and store encrypted data on multiple edge devices and central servers.

9. The remote monitoring system for electric power equipment based on homomorphic encryption according to claim 6 is characterized in that: The trained prediction model is an LSTM model, specifically: Y pred =LSTM(X real-time ) Among them, Y pred is the predicted value of the model, that is, the device data at a certain point in the future, X real-time Input data in real time at the current moment; Anomaly detection is to predict the value Y pred With the actual value X real-time Compare and determine whether there is any abnormality; The anomaly detection formula is: if∣Y pred -X real-time ∣>threshold Among them, threshold is a pre-set abnormal threshold, which is used to determine whether the difference between the predicted value and the actual value exceeds the allowable range. If the difference between the predicted value and the actual value exceeds the set threshold, the system triggers an alarm mechanism.

10. The remote monitoring system for electric power equipment based on homomorphic encryption according to claim 6, characterized in that: The specific contents of building and training the prediction model based on the historical data of power equipment are as follows: Input feature data set: obtain the original data recovered by decryption and preprocess it, combine the extracted features, and divide the feature data into training set, verification set and test set; Model construction: The LSTM model includes input layer, LSTM layer and output layer. The LSTM layer includes forget gate, input gate, candidate memory unit, memory unit update, output gate and current output; Model training: By establishing the mean square error (MSE) loss function and the Adam optimizer, the model is trained using the training set data; the performance of the model is evaluated using the validation set data, and the hyperparameters are adjusted to optimize the model; Model evaluation: Use the test set data to evaluate the performance indicators of the model, draw a comparison chart between the predicted value and the true value, and demonstrate the prediction effect of the model.

Images