Intelligent marketing terminal power data communication security protection method and system

Abstract

The application discloses a smart marketing terminal power data communication security protection method and system, and belongs to the field of power data security, which comprises the following steps: collecting electromagnetic noise spectrum data through a terminal built-in environment electromagnetic sensing module, extracting electromagnetic characteristic factors and constructing an associated model of the electromagnetic characteristic factors and the complexity of elliptic curve cryptography; then building a terminal twin image, combining a long short-term memory neural network to predict a communication instruction sequence and calculate a deviation coefficient; classifying and screening the power data collected by the terminal, integrating the desensitized privacy data into a hash segment of the unique identification of the terminal device, and generating a desensitized power data packet containing the terminal identity identification; generating a data digest according to the encryption parameters determined by the associated model and verifying the access authority, converting the data collection timestamp into a binary coding sequence and embedding the data digest, generating an encrypted data packet with time sequence traceability watermark, and uploading the block chain for notarization; the application realizes the whole-process safety control of power data, and improves the communication security and data credibility.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of electric power data security, and particularly relates to an intelligent marketing terminal electric power data communication security protection method and system. BACKGROUND

[0002] With the in-depth promotion of smart grid construction, the intelligent marketing terminal, as a key node connecting users and power grids, needs to collect and transmit a large amount of electric power data, including user power consumption information, identity information and other sensitive contents. These data face security threats such as electromagnetic interference attacks, instruction tampering, privacy leakage and data tampering in the communication process. In the prior art, the electric power data communication security management mainly adopts fixed encryption strategy, static permission control and single-dimensional anomaly detection method, which has the following disadvantages: the encryption parameters cannot be dynamically adjusted according to the environment, resulting in insufficient attack resistance; the accuracy of abnormal instruction detection is low, and it is difficult to identify new attack modes; the desensitized privacy data cannot be effectively restored in the authorized scenario; and the data storage lacks integrity protection in the whole life cycle. Therefore, there is an urgent need for an intelligent marketing terminal electric power data communication security protection method which can realize dynamic encryption adaptation, accurate anomaly detection, security privacy protection and whole-process trusted storage. SUMMARY

[0003] In view of the deficiencies of the prior art, the application provides an intelligent marketing terminal electric power data communication security protection method and system. An electromagnetic noise spectrum data is collected by an environment electromagnetic sensing module built-in the terminal, an electromagnetic characteristic factor is extracted, and an associated model of the electromagnetic characteristic factor and the complexity of the elliptic curve cryptosystem is constructed. A terminal twin image is built, a communication instruction sequence is predicted by combining a long short-term memory neural network, and a deviation coefficient is calculated. The electric power data collected by the terminal is classified and screened, the desensitized privacy data is fused with a hash segment of the unique identification of the terminal device, and a desensitized electric power data packet containing the terminal identity identification is generated. A data digest is generated according to the encryption parameters determined by the associated model, and the access permission is verified. The data acquisition timestamp is converted into a binary coding sequence and embedded in the data digest, and an encrypted data packet with time sequence traceability watermark is generated to upload the block chain storage. The application realizes the whole-process security management of electric power data, and improves the communication security and data credibility.

[0004] To achieve the above purpose, the application provides the following technical scheme:

[0005] The intelligent marketing terminal electric power data communication security protection method comprises the following steps:

[0006] The terminal collects electromagnetic noise spectrum data by the built-in environment electromagnetic sensing module, extracts the electromagnetic characteristic factor, and constructs an associated model of the electromagnetic characteristic factor and the complexity of the elliptic curve cryptosystem;

[0007] A terminal twin image is constructed. Based on a long short-term memory neural network, the terminal operation status synchronized with the terminal twin image is used to predict the communication instruction sequence. The deviation coefficient between the actual communication instructions generated by the terminal and the predicted communication instruction sequence is calculated.

[0008] Based on the comparison result between the deviation coefficient and the preset threshold, it is determined whether the terminal behavior is abnormal;

[0009] The power data collected by terminals that are deemed to be normal is classified and filtered. Privacy-related data is extracted and desensitized to obtain desensitized power data. The hash fragment of the unique identifier of the terminal device is then extracted as a reversible parsing factor. A preset embedding algorithm is used to integrate the reversible parsing factor into the desensitized power data to generate a desensitized power data package containing the terminal's identity identifier.

[0010] The encryption parameters are determined based on the association model. The encryption parameters are used to generate a data digest for the de-identified power data packet. Access permission verification is completed based on the terminal identity identifier in the de-identified power data packet.

[0011] Obtain the data collection timestamp corresponding to the desensitized power data packet after access permission verification and convert it into a binary encoding sequence. Embed the binary encoding sequence into the least significant bit of the data digest to generate an encrypted data packet with a time-tracing watermark and upload it to the blockchain for evidence storage.

[0012] Specifically, the original privacy data is restored by a reversible parsing algorithm only if the identifier of the terminal device requesting access matches the hash fragment in the de-identified power data packet.

[0013] Specifically, the terminal collects electromagnetic noise spectrum data through a built-in environmental electromagnetic sensing module and extracts electromagnetic feature factors, including:

[0014] The environmental electromagnetic sensing module monitors and collects the original electromagnetic signals in the preset frequency band in real time when the terminal performs elliptic curve cryptography.

[0015] The original electromagnetic signal is preprocessed to obtain digitized electromagnetic noise spectrum data;

[0016] The digitized electromagnetic noise spectrum data is subjected to a fast Fourier transform to convert it from the time domain to the frequency domain, thus obtaining the noise spectrum.

[0017] The electromagnetic characteristic factors are extracted from the noise spectrum; the electromagnetic characteristic factors include noise amplitude, spectral distribution shape, and dynamic change rate; the noise amplitude is the signal strength value of a preset frequency point or frequency band in the noise spectrum; the spectral distribution shape is determined by quantizing the skewness, kurtosis, and energy entropy of the spectrum; the dynamic change rate is the difference in spectral energy between adjacent sampling cycles.

[0018] Specifically, the correlation model between the electromagnetic characteristic factor and the elliptic curve cryptography complexity includes:

[0019] A multi-dimensional feature vector is established, and the electromagnetic characteristic factor is taken as a vector element;

[0020] A plurality of elliptic curve cryptography systems with different complexities are deployed in the terminal; the complexity is determined by the domain size of the selected elliptic curve, the base point order, and the point multiplication algorithm;

[0021] Under different environmental electromagnetic noise, a plurality of elliptic curve cryptography operations are performed, and the operation time, power consumption and error rate are recorded as complexity response indicators;

[0022] A machine learning regression algorithm is used, and the multi-dimensional feature vector is taken as the input and the complexity response indicators are taken as the output to train the correlation model; the correlation model predicts the optimal encryption complexity level in the current environment according to the real-time extracted electromagnetic characteristic factor.

[0023] Specifically, the terminal twin image is built, the long short-term memory neural network is used, the terminal running operation state synchronized with the terminal twin image is combined to predict the communication instruction sequence, and the deviation coefficient of the communication instruction actually generated by the terminal and the predicted communication instruction sequence is calculated, including:

[0024] A terminal twin image corresponding to each physical terminal is built in a local or edge security node; the twin image is synchronized with the running operation state by receiving the system call, process list and network connection state information of the terminal;

[0025] The long short-term memory neural network is used to train the historical communication instruction sequence of the terminal in a preset business scenario to generate a prediction model;

[0026] When communication occurs, the terminal twin image executes the prediction model to output a predicted communication instruction sequence;

[0027] The communication instruction actually generated by the terminal is compared with the predicted communication instruction sequence in terms of similarity to obtain a deviation coefficient.

[0028] Specifically, the hash segment of the terminal device unique identifier is extracted as a reversible analysis factor, and a preset embedding algorithm is used to integrate the reversible analysis factor into the desensitized power data, including:

[0029] The hardware unique identifier of the terminal is obtained;

[0030] The SM3 hash algorithm is used to calculate the hardware unique identifier to obtain a hash digest;

[0031] Truncating the first N bits of the hash digest to generate a hash segment, and taking it as a reversible parsing factor;

[0032] Converting the reversible parsing factor into a binary stream;

[0033] Selecting the reserved bit or extension bit of the field in the desensitized power data packet as an embedding area;

[0034] Embedding the binary stream into the embedding area using the least significant bit replacement algorithm to generate the desensitized power data packet containing the terminal identity.

[0035] Specifically, the encrypted parameter is determined according to the association model, and the encrypted parameter is used to generate a data digest of the desensitized power data packet, including:

[0036] Inputting the real-time extracted electromagnetic feature factor into the association model;

[0037] Obtaining the encryption parameter of the elliptic curve cryptosystem output by the association model and matching the current electromagnetic environment complexity; The encryption parameter includes elliptic curve type, domain parameter and key length;

[0038] Using the elliptic curve digital signature algorithm determined by the encryption parameter to perform digital signature operation on the desensitized power data packet containing the terminal identity, and generating a digital signature value as the data digest.

[0039] Specifically, the data acquisition timestamp corresponding to the desensitized power data packet after access permission verification is obtained and converted into a binary coded sequence, and the binary coded sequence is embedded in the least significant bit of the data digest, including:

[0040] Extracting the data acquisition timestamp from the metadata of the desensitized power data packet;

[0041] Convert the data acquisition timestamp in UTC time format into a binary coded sequence;

[0042] Obtain the binary representation of the data digest;

[0043] Replace the original data in the least significant bit of the binary representation of the data digest with the binary coded sequence, to generate a watermarked data digest.

[0044] Specifically, when uploading to the blockchain for notarization, including:

[0045] Pack the encrypted data packet with time sequence traceability watermark and the deviation coefficient into a transaction;

[0046] Call the blockchain smart contract and submit the transaction to the blockchain network;

[0047] After the blockchain node consensus, a new block containing the transaction is added to the chain, completing the evidence storage.

[0048] The intelligent marketing terminal power data communication security protection system comprises an electromagnetic characteristic sensing module, a state monitoring module, an identity embedding module, an authority management module and a blockchain evidence storage module.

[0049] The electromagnetic characteristic sensing module is used for establishing an association with an encryption system by sensing terminal environment electromagnetic characteristics.

[0050] The state monitoring module is used for monitoring whether the terminal operation is abnormal by synchronizing the terminal operation state and predicting the communication instruction sequence through terminal twin imaging.

[0051] The identity embedding module is used for privacy protection processing of power data and embedding terminal identity information.

[0052] The authority management module is used for encrypting desensitized power data and verifying access authority based on terminal identity.

[0053] The blockchain evidence storage module is used for adding time sequence traceability watermark to encrypted data and uploading the blockchain.

[0054] Compared with the prior art, the intelligent marketing terminal power data communication security protection system has the advantages of simple process, low investment and operation cost, and low production cost.

[0055] 1. The intelligent marketing terminal power data communication security protection system is proposed, and the architecture, operation steps and process are optimized and improved, and the system has the advantages of simple process, low investment and operation cost, and low production cost.

[0056] 2. The intelligent marketing terminal power data communication security protection method is proposed, which constructs an encryption system association model through environmental electromagnetic sensing, can dynamically adapt the encryption parameters to the electromagnetic environment, improves the anti-interference and anti-attack ability, and relies on terminal twin imaging and neural network to predict the communication instruction sequence, can accurately identify abnormal terminal operation, and prevent and control the communication instruction tampering risk in advance, and double safeguards the terminal data transmission safety from the communication source and the operation process.

[0057] 3. The intelligent marketing terminal power data communication security protection method is proposed, which classifies and desensitizes power privacy data and integrates terminal identification hash segments, protects user privacy, and relies on identification to efficiently verify access authority; in combination with the association model encryption, time stamp watermark and blockchain evidence storage, realizes the full-process traceability and tamper-proof of power data from collection, processing to storage, and balances data security and management efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1The application provides an intelligent marketing terminal power data communication security protection method schematic diagram;

[0059] Figure 2 The application provides an intelligent marketing terminal power data communication security protection method schematic diagram;

[0060] Figure 3 The application provides an intelligent marketing terminal power data communication security protection method schematic diagram. DETAILED DESCRIPTION

[0061] Embodiment 1:

[0062] Please refer to Figure 1 and Figure 2 An embodiment provided by the application is an intelligent marketing terminal power data communication security protection method, the method comprises S1-S5, and the steps include the following steps:

[0063] S1: the terminal collects electromagnetic noise spectrum data through the built-in environment electromagnetic perception module, extracts electromagnetic characteristic factors, and constructs an association model of the electromagnetic characteristic factors and the complexity of the elliptic curve cryptography system;

[0064] It should be noted that the encryption scheme of the traditional intelligent marketing terminal mostly adopts an elliptic curve cryptography system with fixed complexity, and the encryption parameters are always unchanged regardless of the change of the electromagnetic environment of the terminal. When the environmental electromagnetic noise is strong, the fixed low-complexity encryption is easy to be cracked by electromagnetic side channel attacks, and the fixed high-complexity encryption will cause the operation error rate to soar due to electromagnetic interference. When the environmental electromagnetic noise is weak, the fixed high-complexity encryption will also cause waste of terminal computing power and reduce data communication efficiency. The method is to establish a dynamic matching mechanism of the environmental electromagnetic state and the encryption complexity, which specifically includes the following steps: through real-time perception of the electromagnetic noise around the terminal, such as the radiation of substation equipment and industrial interference signals, electromagnetic characteristic factors that can accurately reflect the severity of the electromagnetic environment are extracted, and then the electromagnetic characteristic factors are bound to the complexity of the elliptic curve cryptography system, such as the domain size, the base point order and the point multiplication algorithm, so that the terminal can adjust the encryption parameters according to the real-time electromagnetic environment. For example, when the electromagnetic noise is strong and the risk of side channel attacks is high, the association model automatically selects a high-complexity encryption system to improve the attack resistance; when the electromagnetic noise is weak and the communication efficiency demand is high, the association model automatically selects a medium or low-complexity encryption system to balance safety and efficiency. The traditional fixed encryption scheme cannot realize such environmental adaptive adjustment and cannot meet the requirements of anti-interference, attack prevention and high efficiency at the same time.

[0065] S2: a terminal twin image is built, a long short-term memory neural network is used, a terminal running operation state synchronous to the terminal twin image is combined to predict a communication instruction sequence, and a deviation coefficient of a communication instruction actually generated by the terminal and the predicted communication instruction sequence is calculated;

[0066] Further, the terminal twin image continuously receives the running log, system call sequence and resource occupation state from the physical terminal through a secure communication link to keep synchronization.

[0067] It should be noted that once the intelligent marketing terminal is maliciously tampered with, such as implanting malicious programs or replacing core chips, the attacker may forge power data, such as artificially increasing power consumption and tampering with user power records, and generate false communication instructions. The traditional security scheme can only verify the legality of the terminal through identity authentication, and cannot identify and establish a real-time comparison mechanism between the physical terminal and the twin image. The terminal twin image is not a simple copy of the terminal, but a digital double that can synchronize the terminal system call, process list, and network connection state. Through the learning of the terminal historical communication instructions by the long short-term memory neural network, the terminal can accurately predict the instruction sequence that should be generated in a specific business scenario, such as user power purchase and power data upload. When the physical terminal is tampered with, the actual generated instruction sequence will deviate from the predicted instruction sequence of the terminal twin image. By calculating the deviation coefficient, the terminal anomaly can be identified at the first time. For example, the normal terminal instruction sequence when uploading power data should be data reading, format verification, encryption, and sending. The tampered terminal may insert data tampering and false data generation instructions. At this time, the deviation coefficient will exceed the preset threshold, and the system needs to immediately terminate the data transmission of the terminal. Therefore, the abnormal detection function based on instruction prediction in the present application cannot be realized by the traditional identity authentication scheme, which fundamentally eliminates the risk of data forgery of the tampered terminal and forms a terminal operation security protection barrier.

[0068] S3: determining whether the terminal behavior is abnormal according to the comparison result of the deviation coefficient and the preset threshold;

[0069] Further, if the deviation coefficient is lower than the preset warning threshold, it is determined that the terminal behavior is normal, and the subsequent process is executed;

[0070] If the deviation coefficient is higher than the preset warning threshold but lower than the preset fuse threshold, it is determined that the terminal behavior is suspicious, a security warning log is generated and is associatedly marked with the desensitized power data packet;

[0071] If the deviation coefficient is higher than the preset fuse threshold, it is determined that the terminal behavior is abnormal, and the current communication process is terminated and the security isolation mechanism is triggered.

[0072] Further, the warning threshold is used to trigger observation and marking, and the fuse threshold is used to trigger interruption and isolation. The warning threshold and the fuse threshold are set based on a comprehensive engineering method of historical behavior baseline analysis, statistical modeling and business risk tolerance evaluation.

[0073] S4: classifying and screening the power data collected by the terminal determined as normal, extracting the privacy data for desensitization processing to obtain desensitized power data, extracting the hash segment of the unique identifier of the terminal device as a reversible analysis factor, and using a preset embedding algorithm to integrate the reversible analysis factor into the desensitized power data to generate a desensitized power data packet containing the terminal identity identifier;

[0074] Further, the power data collected by the terminal determined as normal is classified and screened, including:

[0075] (1) a preset data classification rule library; the data classification rule library defines which data fields belong to user privacy data, device operation data and public metering data;

[0076] (2) matching the original power data stream collected by the terminal determined as normal with the data classification rule library, identifying and screening out the privacy data containing user identity, power habit and accurate geographic location information.

[0077] Further, the privacy data is extracted for desensitization processing, including:

[0078] (1) the screened privacy data is desensitized by using a hash encryption method;

[0079] (2) recombining the desensitized data with the original non-privacy device operation data and public metering data to obtain desensitized power data.

[0080] It should be noted that the electric power data collected by the intelligent marketing terminal contains a large amount of user privacy information, including but not limited to user name, address, power habit, and the traditional desensitization scheme can protect privacy, but the desensitized data cannot be traced back to the terminal, when the data is wrong, it is difficult to determine which terminal collects, and it is difficult to hold accountable, at the same time, the traditional scheme cannot prevent unauthorized terminals from accessing desensitized data, and there is a risk of secondary privacy leakage. In the method, the privacy data is accurately located by classification and screening, and the non-privacy data such as public power grid voltage data is not over-desensitized to ensure data usability; secondly, the embedded reversible analysis factor is not the terminal ID directly, but a segment after SM3 hash operation, which avoids terminal identity information leakage and restores terminal identity through algorithm, solves the problem of data attribution tracing; finally, the desensitized data packet is bound with the terminal identity identifier, which provides a basis for subsequent permission verification, only the terminal matching the identifier can access the data to prevent unauthorized terminals from obtaining desensitized data. The irreplaceability of this design lies in that it breaks the contradiction between privacy protection and data tracing, and the traditional scheme may abandon tracing for privacy protection or reserve privacy information for tracing, and the method realizes the perfect balance of the two through the embedding of reversible analysis factor, and provides accurate identity basis for permission management.

[0081] S5: determining an encryption parameter according to the association model, generating a data digest for the desensitized electric power data packet using the encryption parameter, and completing access permission verification according to the terminal identity identifier in the desensitized electric power data packet;

[0082] Further, the access permission verification according to the terminal identity identifier in the desensitized electric power data packet comprises:

[0083] (1) restoring the hardware unique identifier of the terminal from the reversible analysis factor embedded in the desensitized electric power data packet;

[0084] (2) querying a preconfigured permission mapping table; the permission mapping table stores data access permission levels corresponding to different terminal identifiers;

[0085] (3) judging whether the permission of an entity requesting to access the data packet matches the data access permission level, if the permissions match, the verification is passed, if the permissions do not match, the process is terminated and a security event is recorded.

[0086] It should be noted that once the encryption parameters of the traditional intelligent marketing terminal are determined, they remain unchanged for a long time, and attackers can implement cracking by analyzing the vulnerabilities of fixed encryption parameters. At the same time, traditional permission verification is mainly based on role authorization, such as administrators who can access all data. There is a problem of excessive allocation of permissions, even if the administrator does not need to access the data of a terminal, he still has access rights, which increases the risk of data leakage. In this method, a dynamic encryption-precise authorization mechanism is established: on the one hand, the encryption parameters are not fixed, but are dynamically determined according to real-time electromagnetic characteristics through the association model constructed in the first step. For example, when the terminal is near a substation and the electromagnetic noise is strong, the model selects an elliptic curve with a 256-bit domain to generate high-strength encryption parameters. When the terminal is in a residential area and the electromagnetic noise is weak, an elliptic curve with a 192-bit domain is selected to generate adaptive encryption parameters. This dynamic encryption makes it impossible for attackers to crack through fixed parameters, improving the security of encryption. On the other hand, permission verification is not based on roles, but on the terminal identity in the desensitized power data packet. Only when the terminal identity requesting access matches the identity in the data packet can the verification be passed, realizing precise authorization that the terminal can only access the data it collects, and avoiding the problem of excessive allocation of permissions in traditional role authorization.

[0087] S6: Obtain the data collection timestamp corresponding to the desensitized power data packet after access permission verification and convert it to a binary code sequence. Embed the binary code sequence in the least significant bits of the data digest to generate an encrypted data packet with a time sequence source watermark and upload it to the blockchain for notarization.

[0088] Further, the embedding process ensures that the overall length and encryption strength of the data digest are not changed.

[0089] The original privacy data is restored by the reversible analysis algorithm only when the terminal device identity requesting access matches the hash segment in the desensitized power data packet.

[0090] Further, generating an encrypted data packet with a time sequence source watermark includes:

[0091] (1) Re-associate and encapsulate the data digest with the time stamp watermark embedded with the original desensitized power data packet;

[0092] (2) Generate the final encrypted data packet with a time sequence source watermark that can be uploaded, wherein the data digest serves as an integrity check and identity verification credential, and the data generation time information is hidden in the least significant bits.

[0093] Further, before uploading to the blockchain, the encrypted data packet is also subjected to integrity verification:

[0094] (1) Separate the data digest from the encrypted data packet with a time sequence source watermark;

[0095] (2) Extract the binary coded sequence embedded in the least significant bits of the data digest and restore it to a timestamp;

[0096] (3) Recompute the data digest for the de-identified power data packet portion using the same encryption parameters and algorithm;

[0097] (4) Compare the newly computed data digest with the original data digest that has been separated from the de-identified power data packet and embedded with a watermark, and if they are consistent within an allowable error range, the verification is passed.

[0098] Further, before uploading to the blockchain, it also includes lightweight encryption packaging of the encrypted data packet with the time sequence traceability watermark, and the packaging key is dynamically derived from the deviation coefficient and the reversible resolution factor.

[0099] Further, the blockchain is a power data security evidence platform based on a consortium chain architecture, only authorized nodes can access the data on the chain, and the integrity and generation time of the data packet can be verified and traced using the data digest and embedded time sequence watermark.

[0100] It should be noted that in the present method, a time sequence watermark-blockchain double traceability and evidence storage mechanism is established: first, the time sequence watermark is not simply a time stamp added, but a timestamp converted into a binary coded sequence, embedded in the least significant bits of the data digest, and the data digest corresponds one-to-one with the de-identified power data packet. If the data is tampered with, the data digest will change, and the watermark will also be damaged. By verifying the integrity of the watermark, it can be directly judged whether the data has been tampered with; second, the encrypted data packet with the watermark is uploaded to the blockchain, and the distributed ledger and consensus mechanism of the blockchain are used to realize data tamper-proof storage. Any node that attempts to modify the data packet must tamper with the records of all nodes, which is extremely difficult, achieving full-process protection of traceable data collection time, verifiable data integrity, and tamper-proof data storage, ensuring the safety of power data from collection to storage throughout its life cycle.

[0101] The terminal collects electromagnetic noise spectrum data through an embedded environmental electromagnetic sensing module and extracts electromagnetic characteristic factors, including:

[0102] A1: Through the environmental electromagnetic sensing module, real-time monitoring and collection of original electromagnetic signals in a preset frequency band are performed when the terminal performs elliptic curve password operation;

[0103] It needs to be understood that the main processor or dedicated security coprocessor inside the terminal device will send a precise synchronization trigger signal to the environmental electromagnetic perception module before performing elliptic curve cryptographic operation, which indicates that a cryptographic operation is about to start and requires the environmental electromagnetic perception module to enter the highest preparation state.

[0104] Further, the core of the environmental electromagnetic perception module is a highly miniaturized and highly sensitive electromagnetic induction probe. The probe is usually composed of a specially designed loop antenna or a magnetic field effect transistor, and its physical characteristics are precisely calibrated to respond to extremely weak electromagnetic field changes in a specific space range.

[0105] Further, the specific steps of A1 include:

[0106] (1) After receiving the synchronization trigger signal, the analog front-end circuit inside the environmental electromagnetic perception module is immediately activated; the analog front-end circuit includes a low-noise amplifier and a programmable gain controller, and its primary task is to preliminarily amplify the original analog electromagnetic signal captured by the probe while trying to suppress the inherent thermal noise and shot noise of the circuit itself to ensure signal fidelity;

[0107] (2) The preliminarily amplified original analog electromagnetic signal is sent to an anti-aliasing filter; the anti-aliasing filter is a sharp-cut low-pass filter, and its cutoff frequency is strictly set according to the requirements of the sampling theorem, and the core is to completely filter out all frequency components in the original analog electromagnetic signal that are higher than half of the preset sampling frequency, wherein the working principle of the low-pass filter is the existing technology in the art, and is not described here as it is not the inventive concept of the present application;

[0108] Further, in the present application, sending the preliminarily amplified original analog electromagnetic signal to the anti-aliasing filter can prevent these high-frequency components from being distorted in the subsequent sampling process and being incorrectly mapped to the low-frequency band, causing confusion and distortion of the signal spectrum, thereby ensuring the authenticity and effectiveness of the collected data.

[0109] (3) After the analog signal passes through the anti-aliasing filter, it enters the analog-to-digital conversion link, and the analog-to-digital converter works at a pre-set fixed sampling frequency to obtain the original digital signal sequence; the fixed sampling frequency is at least twice the highest frequency of the preset monitoring frequency band to meet the Nyquist sampling theorem;

[0110] Further, the analog-to-digital converter works to measure the continuous analog voltage signal at each equally spaced time point and convert it into a long string of discrete digital quantization values, and the accuracy is determined by the resolution of the analog-to-digital converter.

[0111] (4) the original digital signal sequence is sent into a digital down-conversion processing unit; the digital down-conversion processing unit is composed of a digital control oscillator and a digital filter;

[0112] Further, the working steps of the digital down-conversion processing unit include: first, the digital control oscillator generates a sine wave and a cosine wave digital reference signal with the same center frequency as the preset frequency band; then, the original digital signal sequence is multiplied by the two reference signals, which is equivalent to shifting the entire spectrum in the frequency domain, and moving the center frequency of the preset frequency band to the vicinity of zero frequency; finally, through a high-performance low-pass digital filter, only the low-frequency component near zero frequency after shifting is retained, and this low-frequency component corresponds to the preset frequency band in the original digital signal, and the output of the filter is called the baseband signal.

[0113] (5) outputting multiple frames of original electromagnetic signal digital samples after preprocessing and frequency band screening.

[0114] A2: preprocessing the original electromagnetic signal to obtain digital electromagnetic noise spectrum data;

[0115] Further, the preprocessing includes signal amplification, filtering and analog-to-digital conversion.

[0116] A3: performing fast Fourier transform on the digital electromagnetic noise spectrum data to convert it from time domain to frequency domain to obtain a noise spectrum, wherein the fast Fourier transform is a prior art content in the field and is not the inventive scheme of the present application, and will not be described here.

[0117] A4: extracting the electromagnetic characteristic factor from the noise spectrum; the electromagnetic characteristic factor includes noise amplitude, spectral distribution form and dynamic change rate; the noise amplitude is the signal intensity value of a preset frequency point or frequency band in the noise spectrum; the spectral distribution form is determined by calculating the skewness, kurtosis and energy entropy; and the dynamic change rate is the difference value of the spectral energy between adjacent sampling periods.

[0118] The correlation model between the constructed electromagnetic characteristic factor and the complexity of the elliptic curve cryptography system includes:

[0119] B1: establishing a multi-dimensional feature vector, and taking the electromagnetic characteristic factor as a vector element;

[0120] B2: deploying multiple elliptic curve cryptography systems with different complexities in the terminal; the complexity is determined by the domain size, base point order and point multiplication algorithm of the selected elliptic curve;

[0121] Further, before performing B2, the terminal needs to integrate a set of strictly tested and verified cryptographic parameters in advance, which contains multiple sets of different elliptic curve domain names and their corresponding complete parameters, and the parameters must come from international cryptographic standards.

[0122] Further, complexity is a comprehensive evaluation index, which is determined by three core dimensions: the first dimension is algorithm complexity, which is directly determined by the size of the selected elliptic curve domain, and the size of the domain in bits is the most intuitive measurement standard; the second dimension is calculation complexity, which is reflected in the actual calculation time and processor resources required to perform a complete cryptographic operation, such as digital signature generation or verification, which needs to be obtained through actual measurement; the third dimension is error rate complexity, which refers to the probability of calculation errors occurring during algorithm operation under severe electromagnetic noise interference; these three dimensions together constitute a three-dimensional model for evaluating the complexity of a cryptographic system.

[0123] Further, the process of deploying multiple elliptic curve cryptographic systems with different complexities in the terminal includes:

[0124] (1) Construct a preset elliptic curve cryptographic parameter library;

[0125] (2) Define and quantify the complexity index of the cryptographic system;

[0126] (3) Establish a test benchmark and performance collection environment, which specifically includes: developing an automated test framework that sequentially loads each curve configuration in the preset elliptic curve cryptographic parameter library; for each curve configuration, the test framework simulates real business scenarios, generates test data samples, and drives the cryptographic coprocessor to perform thousands of cryptographic operations, such as repeatedly signing and verifying data; in this process, the test framework uses the terminal's performance counters and high-precision timers to closely monitor and record the precise time consumed by each operation, the CPU utilization rate, and the memory access frequency, etc.

[0127] (4) Data analysis and complexity label association, including: for each group of curve configuration, the system aggregates and analyzes all test running results, calculates the average and variance of all operation running time to measure its calculation efficiency and stability, and counts the number of verification errors in all test running, calculates the error rate, finally, according to the preset rules, the quantitative results of the three dimensions are weighted and integrated, and a final single complexity level label is assigned to each group of curve configuration, for example, a 256-bit curve may be marked as standard complexity due to its high efficiency and low error rate under low noise; while a 521 million bit curve is marked as high complexity due to its huge calculation amount and possibly higher error rate;

[0128] (5) When the elliptic curve password parameter library is complete and the complexity of all configurations is labeled, a dynamic selection interface is created, which provides a unified function call interface to the outside world; the core input parameter of the function call is the recommended complexity level calculated by the association model representing the current environment severity, when this interface is called, its internal logic will quickly retrieve and select all curve configuration sets matching this level from the parameter library according to the input recommended complexity level, if there are multiple matching configurations, a polling method can be used to finally determine a curve and its supporting algorithm, once selected, the interface will load all parameters of the configuration and initialize the corresponding password algorithm engine to make it ready for execution of actual data encryption or signature tasks.

[0129] B3: Under different environmental electromagnetic noise, perform various elliptic curve password operations, and record the operation time, power consumption and error rate as complexity response indicators, wherein the elliptic curve password operation, also known as elliptic curve password system, is a prior art in the field and is not part of the inventive concept of the present application, and will not be described here;

[0130] B4: Use machine learning regression algorithm to train the association model with the multi-dimensional feature vector as input and the complexity response indicator as output; the association model predicts the optimal encryption complexity level under the current environment according to the real-time extracted electromagnetic characteristic factor, wherein the machine learning regression algorithm is a prior art in the field and is not part of the inventive concept of the present application, and will not be described here.

[0131] The terminal twin image is built based on a long short-term memory neural network, and the terminal running operation state synchronized with the terminal twin image is combined to predict a communication instruction sequence, and a deviation coefficient of the communication instruction actually generated by the terminal and the predicted communication instruction sequence is calculated, including:

[0132] S2.1: Construct a terminal twin image corresponding to the physical terminal at the local or edge security node; the twin image is synchronized with the running operation state by receiving the terminal's system call, process list, and network connection state information;

[0133] Further, before constructing the terminal twin image, the physical terminal must first be uniquely identified and registered on the edge security node, which is achieved by generating a unique terminal asset fingerprint.

[0134] Further, constructing a terminal twin image corresponding to the physical terminal at the local or edge security node includes:

[0135] (1) Start the agent program on the physical terminal to collect its multi-dimensional, tamper-resistant hardware and software inherent information, including hardware identification, software identification, and environment configuration, wherein the hardware identification includes the central processing unit serial number, motherboard serial number, network card media access control address, and hard disk serial number, the software identification includes the operating system installation identifier, firmware version number, and pre-installed security certificate identifier, and the environment configuration includes the installed application program list and its version, and key configuration file hash value;

[0136] (2) The agent program uses the SHA-256 algorithm to normalize and calculate the collected hardware and software inherent information to generate a hash value, i.e., a terminal asset fingerprint;

[0137] (3) The terminal asset fingerprint is sent to the designated local or edge security node along with the terminal's basic model and organization metadata through a certificate-based two-way authentication channel for registration. The local or edge security node stores the terminal asset fingerprint in the trusted terminal asset library, thereby completing the initial trust establishment of the terminal identity;

[0138] (4) The edge security node initializes a corresponding twin image for the physical terminal after receiving the registration information, including template selection, image instantiation, and identity injection;

[0139] Template selection: According to the metadata reported by the physical terminal, select a most matching basic template from the pre-set image template library. This basic template is a clean, minimized virtual machine or container image that contains the operating system and basic software environment common to the terminal model;

[0140] Image instantiation: Create a new virtual machine or container instance based on the selected basic template to obtain an image instance;

[0141] Identity injection: Write the terminal asset fingerprint generated in the first step as the unique identifier of the image instance into its internal configuration file to generate the newly created twin image instance;

[0142] (5) Deploy the same agent on the newly created twin image instance as on the physical terminal;

[0143] (6) Configure the agent with the network address of the physical terminal it corresponds to and the credentials needed to establish a secure connection, such as a key derived from the terminal's asset fingerprint;

[0144] (7) The agent on the physical terminal initiates a secure communication link with the agent in the twin image instance on the edge security node, typically encrypted using Transport Layer Security and based on the certificates registered in the first step for mutual authentication;

[0145] (8) After the connection is established, the physical terminal agent sends a complete snapshot of the current system state to the mirror agent. This includes:

[0146] Process list: current information of all running processes;

[0147] System service status: the enablement and running status of all services;

[0148] Network connection status: all active network connections;

[0149] User session information: the currently logged-in user;

[0150] Key system configurations: the latest system configurations and policies;

[0151] (9) The mirror agent receives the complete state of the current system and completely resets the internal state of the mirror to match the physical terminal, resulting in a synchronized terminal twin image.

[0152] S2.2: Use a long short-term memory neural network to train the terminal's historical communication instruction sequence under a preset business scenario to generate a prediction model, wherein the long short-term memory neural network is prior art content in the field and is not the inventive scheme of the present application, and will not be described here;

[0153] S2.3: When communication occurs, execute the prediction model by the terminal twin image to output a predicted communication instruction sequence;

[0154] Further, based on the long short-term memory neural network to predict the communication instruction sequence, specifically comprising:

[0155] (1) Collect the historical communication instruction sequence obtained by the terminal twin image synchronization to form a training data set;

[0156] (2) taking the first A instructions in the historical communication instruction sequence as input and the subsequent B instructions as output, performing supervised learning training on the long short-term memory neural network, wherein the long short-term memory neural network is prior art in the field and is not the inventive scheme of the present application, and will not be described here;

[0157] (3) using the trained long short-term memory neural network, taking the current and historical communication instructions of the terminal as input, predicting the most likely generated communication instruction sequence in the next period.

[0158] S2.4: Comparing the actual communication instructions generated by the terminal with the predicted communication instruction sequence to obtain a deviation coefficient.

[0159] Further, the core of obtaining the actual communication instructions generated by the terminal is to establish a transparent data monitoring and analysis pipeline in the terminal, which is equivalent to deploying a collection agent at the bottom layer of the operating system of the terminal, or can also be understood as setting a high-definition camera at all network outlets. Whenever any software on the terminal wants to send data through the network, the collection agent will instantly intercept a copy of the original data and record which program, at what time, and what data is sent; then according to the communication protocol, such as MQTT commonly used in the power industry, the original data is parsed to decipher the real intention, for example, identifying whether it is a voltage reading reporting instruction or a device query response command; after parsing, the parsed instructions are queued in chronological order to form a complete and ordered actual instruction sequence.

[0160] Further, the specific steps of S2.4 include:

[0161] (1) aligning and comparing the actual communication instruction sequence generated by the terminal with the predicted communication instruction sequence;

[0162] (2) calculating the difference between the two sequences using the cosine similarity algorithm, wherein the cosine similarity algorithm is prior art in the field and is not the inventive scheme of the present application, and will not be described here;

[0163] (3) mapping the difference to the interval [0, 1] to obtain a deviation coefficient, wherein the closer the deviation coefficient is to 1, the greater the deviation, and the higher the possibility of abnormal terminal behavior.

[0164] The hash segment of the unique identifier of the terminal device is extracted as a reversible analysis factor, and a preset embedding algorithm is used to integrate the reversible analysis factor into the desensitized power data, including:

[0165] S4.1: Obtain the hardware unique identifier of the terminal;

[0166] S4.2: calculating the hardware unique identifier using the SM3 hash algorithm to obtain a hash digest, wherein the SM3 hash algorithm is prior art in the field and is not the inventive scheme of the present application, and will not be described here;

[0167] S4.3: intercepting the first N bits of the hash digest to generate a hash segment and using it as an invertible parsing factor;

[0168] S4.4: converting the invertible parsing factor into a binary stream;

[0169] S4.5: selecting a reserved bit or an extension bit of a specific field in the desensitized power data packet as an embedding area;

[0170] S4.6: embedding the binary stream into the embedding area using the least significant bit replacement algorithm to generate the desensitized power data packet containing the terminal identity, wherein the least significant bit replacement algorithm is prior art in the field and is not the inventive scheme of the present application, and will not be described here.

[0171] The determination of the encryption parameter according to the correlation model and the generation of the data digest of the desensitized power data packet using the encryption parameter include:

[0172] S5.1: inputting the real-time extracted electromagnetic characteristic factor into the correlation model;

[0173] S5.2: obtaining the encryption parameter of the elliptic curve cryptosystem output by the correlation model and matching the current electromagnetic environment complexity; the encryption parameter includes an elliptic curve type, a domain parameter and a key length;

[0174] S5.3: performing a digital signature operation on the desensitized power data packet containing the terminal identity using the elliptic curve digital signature algorithm determined by the encryption parameter to generate a digital signature value as the data digest.

[0175] Further, the specific steps of S5.3 include:

[0176] (1) obtaining the encryption parameter and initializing the cryptographic environment;

[0177] (2) retrieving the terminal static private key, after identity verification, the private key of the terminal is decrypted and loaded into the secure memory area of the cryptographic operation module from the secure non-volatile memory, the private key is a very large secret integer, which is randomly generated within the range determined by the order of the elliptic curve base point defined by the encryption parameter, and is paired with a public key;

[0178] (3) obtaining the desensitized power data packet containing the terminal identity, calculating the hash value of the desensitized power data packet using the SM3 hash algorithm to obtain a message digest;

[0179] (4) generating a random number with high cryptographic strength, and then performing an elliptic curve point multiplication operation, including: multiplying a public base point on an elliptic curve with the random number, the result of the point multiplication being the coordinates of another point on the elliptic curve, taking the horizontal coordinate value of the result point and performing a modulo operation with the order of the elliptic curve base point as the modulus, finally obtaining a numerical value, i.e. a commitment value;

[0180] (5) multiplying the terminal private key with the generated commitment value to obtain a first intermediate product, at the same time, multiplying the random number with the message digest to obtain a second intermediate product, then adding the message digest to the first intermediate product to obtain a first sum, and subtracting the first sum from the value of the second intermediate product to obtain a final result, calculating the modular inverse of the final result with the order of the elliptic curve base point as the modulus to obtain a proof value;

[0181] (6) connecting the commitment value and the proof value in order to form a complete digital signature value, i.e. a data digest.

[0182] The data acquisition timestamp corresponding to the desensitized power data packet after access permission verification is obtained and converted into a binary code sequence, and the binary code sequence is embedded in the least significant bit of the data digest, including:

[0183] C1: extracting the data acquisition timestamp from the metadata of the desensitized power data packet;

[0184] C2: converting the UTC time format data acquisition timestamp into a continuous binary sequence, i.e. a binary code sequence;

[0185] C3: obtaining a binary representation of the data digest;

[0186] C4: replacing the original data in the least significant bit of the binary representation of the data digest with the binary code sequence bit by bit to generate a watermarked data digest.

[0187] When uploading to the blockchain for notarization, including:

[0188] D1: packing the encrypted data packet with time sequence traceability watermark and the deviation coefficient together into a transaction;

[0189] D2: calling a blockchain smart contract to submit the transaction to the blockchain network;

[0190] D3: after consensus of the blockchain node, a new block containing the transaction is appended to the chain to complete the notarization.

[0191] Further, when the blockchain smart contract is called, it performs:

[0192] (1) reading the terminal identity and time sequence traceability watermark in the storage data;

[0193] (2) verifying whether the terminal identity is a legal member in the alliance chain;

[0194] (3) cross-verification of the timestamp and the block time on the blockchain to ensure the authenticity of the data collection time;

[0195] (4) comparing the deviation coefficient with a preset threshold value, if the deviation coefficient exceeds the threshold value, an abnormal terminal behavior warning event is automatically generated and recorded on the blockchain.

[0196] Embodiment 2:

[0197] Referring to Figure 3 , the present application provides another embodiment: an intelligent marketing terminal power data communication security protection system, comprising:

[0198] an electromagnetic feature perception module, a state monitoring module, an identity embedding module, a permission management module, and a blockchain storage module;

[0199] The electromagnetic feature perception module is used to associate with the encryption system by perceiving the electromagnetic features of the terminal environment.

[0200] The state monitoring module is used to monitor whether the terminal operation is abnormal by synchronizing the terminal operation state and predicting the communication instruction sequence through the terminal twin image, and to ensure the safety of the data collection source.

[0201] The identity embedding module is used for privacy protection processing of power data and embedding terminal identity information to provide a basis for permission management.

[0202] The permission management module is used for encrypting desensitized power data and verifying access permissions based on terminal identity to ensure data transmission and access security.

[0203] The blockchain storage module is used to add time sequence traceability watermark to encrypted data and upload it to the blockchain to realize traceability and tamper-proof storage of the whole life cycle of data.

[0204] The electromagnetic feature perception module includes an electromagnetic noise acquisition unit, a feature extraction unit, and an encryption association modeling unit.

[0205] The electromagnetic noise acquisition unit is used to acquire electromagnetic noise spectrum data in the terminal operating environment by relying on the built-in environmental electromagnetic perception module of the terminal.

[0206] The feature extraction unit is used to extract electromagnetic feature factors including noise amplitude, spectral distribution pattern, and dynamic change rate from the acquired electromagnetic noise spectrum data.

[0207] The encryption association modeling unit is configured to construct an association model of the electromagnetic characteristic factor and the elliptic curve cryptography complexity, and to realize the function of dynamically determining the encryption parameter according to the real-time electromagnetic characteristic.

[0208] The state monitoring module comprises a twin mirror construction unit, a communication instruction prediction unit and an abnormality determination unit.

[0209] The twin mirror construction unit synchronizes the running operation state of the terminal in real time through the terminal twin mirror, including the hardware load, the software process and the communication behavior.

[0210] The communication instruction prediction unit is configured to predict the communication instruction sequence to be generated by the terminal based on the long short-term memory neural network and in combination with the terminal state data synchronized by the twin mirror.

[0211] The abnormality determination unit is configured to calculate the deviation coefficient of the communication instruction actually generated by the terminal and the predicted communication instruction sequence, and to determine that the terminal is abnormal and trigger an early warning when the deviation coefficient exceeds a preset threshold.

[0212] The identity embedding module comprises a classification and screening unit, a privacy data desensitization unit, a factor generation unit and an identity embedding unit.

[0213] The classification and screening unit is configured to classify the original power data collected by the terminal, and to screen out the user privacy data, including the power consumption details and the personal identity association data.

[0214] The privacy data desensitization unit is configured to perform desensitization processing, such as anonymization and data deformation, on the screened privacy data, to generate desensitized power data and avoid privacy leakage.

[0215] The factor generation unit is configured to extract the hash segment of the unique identifier of the terminal device as a reversible analysis factor, wherein the unique identifier of the terminal device comprises a hardware serial number and a device code.

[0216] The identity embedding unit is configured to use a preset embedding algorithm to integrate the reversible analysis factor into the desensitized power data to generate a desensitized power data packet containing the terminal identity.

[0217] The permission management module comprises an encryption parameter determination unit, a data digest generation unit and a permission verification unit.

[0218] The encryption parameter determination unit determines the encryption parameter of the elliptic curve cryptography by calling the association model generated by the electromagnetic characteristic sensing and encryption parameter association module and in combination with the current electromagnetic characteristic factor.

[0219] The data digest generation unit generates a unique corresponding data digest by performing encryption operation on the desensitized power data packet using the determined encryption parameter.

[0220] The permission verification unit is configured to verify whether the permission of the access subject matches according to the terminal identity embedded in the desensitized power data packet.

[0221] The blockchain storage module comprises a timestamp encoding unit, a traceability watermark embedding unit and a blockchain storage unit.

[0222] The timestamp encoding unit is configured to obtain a data collection timestamp corresponding to the desensitized power data packet that has passed the permission verification, and convert the data collection timestamp into a binary code sequence.

[0223] The traceability watermark embedding unit is configured to embed the binary code sequence into the least significant bit of the data digest, and generate an encrypted data packet with a time sequence traceability watermark.

[0224] The blockchain storage unit is configured to upload the encrypted data packet with the time sequence traceability watermark to a blockchain, and utilize the distributed ledger and tamper-proof characteristics of the blockchain to realize safe storage and full-process traceability of the data.

[0225] The embodiments of the present application are described above in combination with the drawings, but the present application is not limited to the above specific embodiments, and the above specific embodiments are only illustrative rather than limiting, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments without departing from the purpose and the scope of protection of the present application, and these are all within the protection scope of the present application.

[0226] If the technical solution of the present disclosure involves personal information, the product applying the technical solution of the present disclosure has explicitly informed the personal information processing rules before processing the personal information and has obtained the personal independent consent. If the technical solution of the present disclosure involves sensitive personal information, the product applying the technical solution of the present disclosure has obtained the personal independent consent before processing the sensitive personal information, and at the same time meets the requirement of "explicit consent". For example, at the personal information collection device such as camera, a clear and prominent mark is set to inform that the personal information collection range has been entered and the personal information will be collected. If the individual voluntarily enters the collection range, it is considered to agree to collect the personal information. Or, on the device for processing personal information, the personal information processing rules are informed by using obvious marks / information, and the personal authorization is obtained by pop-up information or by asking the individual to upload his / her personal information. The personal information processing rules can include personal information processor, personal information processing purpose, processing method and personal information type, etc.

Claims

1. A method for protecting the security of power data communication in intelligent marketing terminals, characterized in that, include: The terminal collects electromagnetic noise spectrum data through its built-in environmental electromagnetic sensing module, extracts electromagnetic feature factors, and constructs a correlation model between electromagnetic feature factors and the complexity of elliptic curve cryptography. A terminal twin image is constructed. Based on a long short-term memory neural network, the terminal operation status synchronized with the terminal twin image is used to predict the communication instruction sequence. The deviation coefficient between the actual communication instructions generated by the terminal and the predicted communication instruction sequence is calculated. Based on the comparison result between the deviation coefficient and the preset threshold, it is determined whether the terminal behavior is abnormal; The power data collected by terminals that are deemed to be normal is classified and filtered. Privacy-related data is extracted and desensitized to obtain desensitized power data. The hash fragment of the unique identifier of the terminal device is then extracted as a reversible parsing factor. A preset embedding algorithm is used to integrate the reversible parsing factor into the desensitized power data to generate a desensitized power data package containing the terminal's identity identifier. The encryption parameters are determined based on the association model. The encryption parameters are used to generate a data digest for the de-identified power data packet. Access permission verification is completed based on the terminal identity identifier in the de-identified power data packet. Obtain the data collection timestamp corresponding to the desensitized power data packet after access permission verification and convert it into a binary encoding sequence. Embed the binary encoding sequence into the least significant bit of the data digest to generate an encrypted data packet with a time-tracing watermark and upload it to the blockchain for evidence storage. The terminal collects electromagnetic noise spectrum data through its built-in environmental electromagnetic sensing module and extracts electromagnetic feature factors, including: The environmental electromagnetic sensing module monitors and collects the original electromagnetic signals in the preset frequency band in real time when the terminal performs elliptic curve cryptography. The original electromagnetic signal is preprocessed to obtain digitized electromagnetic noise spectrum data; The digitized electromagnetic noise spectrum data is subjected to a fast Fourier transform to convert it from the time domain to the frequency domain, thus obtaining the noise spectrum. The electromagnetic characteristic factors are extracted from the noise spectrum; the electromagnetic characteristic factors include noise amplitude, spectral distribution shape, and dynamic change rate; the noise amplitude is the signal strength value of a preset frequency point or band in the noise spectrum; the spectral distribution shape is determined by quantizing the skewness, kurtosis, and energy entropy of the spectrum; the dynamic change rate is the difference in spectral energy between adjacent sampling cycles; The construction of the correlation model between electromagnetic feature factors and the complexity of elliptic curve cryptography includes: Establish a multidimensional feature vector, using the electromagnetic feature factors as vector elements; Multiple elliptic curve cryptosystems with different complexities are deployed on the terminal; the complexity is determined by the field size, base order, and dot product algorithm of the selected elliptic curve. Under different electromagnetic noise environments, perform various elliptic curve cryptography operations and record their operation time, power consumption, and error rate as complexity response indicators; Using a machine learning regression algorithm, with the multidimensional feature vector as input and the complexity response index as output, an association model is trained to obtain an association model; the association model predicts the optimal encryption complexity level in the current environment based on the electromagnetic feature factors extracted in real time.

2. The method for protecting the security of power data communication in intelligent marketing terminals as described in claim 1, characterized in that, The original privacy data is restored using a reversible parsing algorithm only if the identifier of the terminal device requesting access matches a hash fragment in the de-identified power data packet.

3. The method for protecting the security of power data communication in intelligent marketing terminals as described in claim 2, characterized in that, The construction of the terminal twin mirror, based on a long short-term memory neural network, combines the terminal's operating state synchronized with the terminal twin mirror to predict the communication command sequence, and calculates the deviation coefficient between the actual communication commands generated by the terminal and the predicted communication command sequence, including: A terminal twin image corresponding one-to-one with the physical terminal is constructed on a local or edge security node; the twin image is synchronized with the running operation status by receiving the terminal's system calls, process list and network connection status information; A prediction model is generated by training the terminal's historical communication command sequences under preset business scenarios using a long short-term memory neural network. When communication occurs, the prediction model is executed by the terminal twin mirror, and the predicted sequence of communication instructions is output. The actual communication commands generated by the terminal are compared with the predicted communication command sequence to obtain the deviation coefficient.

4. The method for protecting the security of power data communication in intelligent marketing terminals as described in claim 3, characterized in that, The hash fragment of the extracted terminal device's unique identifier is used as a reversible parsing factor. A preset embedding algorithm is used to integrate the reversible parsing factor into the de-identified power data, including: Obtain the terminal's unique hardware identifier; The SM3 hash algorithm is used to calculate the hash digest of the hardware unique identifier. Extract the first N bits of the hash digest to generate a hash fragment, and use it as a reversible parsing factor; Convert the reversible parsing factor into a binary stream; Select the reserved or extended bits of the fields in the desensitized power data packet as the embedding area; The binary stream is embedded into the embedding region using the least significant bit substitution algorithm to generate the desensitized power data packet containing the terminal identity identifier.

5. The method for protecting the security of power data communication in intelligent marketing terminals as described in claim 4, characterized in that, The step of determining encryption parameters based on the association model and using the encryption parameters to generate a data digest for the de-identified power data packet includes: The electromagnetic feature factors extracted in real time are input into the correlation model; Obtain the encryption parameters of the elliptic curve cryptosystem output by the correlation model that match the complexity of the current electromagnetic environment; the encryption parameters include elliptic curve type, domain parameters, and key length; The de-identified power data packet containing the terminal identity is digitally signed using an elliptic curve digital signature algorithm determined by the encryption parameters, and a digital signature value is generated as the data digest.

6. The method for protecting the security of power data communication in intelligent marketing terminals as described in claim 5, characterized in that, The process of obtaining the data acquisition timestamp corresponding to the de-identified power data packet after access permission verification and converting it into a binary encoded sequence, and embedding the least significant bit of the binary encoded sequence into the data digest, includes: Extract the data acquisition timestamp from the metadata of the de-identified power data packet; Convert the data acquisition timestamps in UTC time format into binary encoded sequences; Obtain the binary representation of the data digest; The original data in the least significant bit of the binary representation of the data digest is replaced bit by bit by the binary encoded sequence to generate a watermarked data digest.

7. The method for protecting the security of power data communication in intelligent marketing terminals as described in claim 6, characterized in that, The process of uploading to the blockchain for evidence storage includes: The encrypted data packet with time-series traceability watermark and the deviation coefficient are packaged together into a single transaction; Invoke the blockchain smart contract to submit the transaction to the blockchain network; After consensus is reached among the blockchain nodes, a new block containing the transaction is added to the chain to complete the notarization.

8. A smart marketing terminal power data communication security protection system, used to implement the smart marketing terminal power data communication security protection method according to any one of claims 1-7, characterized in that, include: Electromagnetic feature sensing module, status monitoring module, identity embedding module, access control module, and blockchain evidence storage module; The electromagnetic feature sensing module is used to establish an association with the encryption system by sensing the electromagnetic features of the terminal environment. The status monitoring module is used to synchronize the terminal's operating status through the terminal twin mirror and predict the communication command sequence to monitor whether the terminal is operating abnormally. The identity embedding module is used to perform privacy protection processing on power data and embed terminal identity information at the same time. The access control module is used to encrypt the de-identified power data and verify access permissions based on the terminal identity. The blockchain evidence storage module is used to add a time-tracking watermark to encrypted data and upload it to the blockchain.

Images