Device authentication management method and system based on electromagnetic fingerprint

Abstract

The present invention discloses a device authentication management method and system based on electromagnetic fingerprint, and the method includes electromagnetic fingerprint collection, data enhancement, electromagnetic fingerprint management, electromagnetic fingerprint comparison, dynamic fingerprint update and device authentication management. The present invention relates to the technical field of electromagnetic device authentication management, and specifically refers to a device authentication management method and system based on electromagnetic fingerprint. The present invention adopts a method combining signal domain capture, calibration environment optimization and frequency band distinction enhancement to collect and enhance signals, and further construct electromagnetic fingerprints; adopts a convolutional long-term and short-term neural network combined with adaptive weighting improvement to optimize fingerprint comparison, and also optimizes the influence of environmental noise in the data collection stage; adopts a reinforcement learning method combined with convolution feature optimization to update dynamic fingerprints.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic device authentication management, and in particular to an electromagnetic fingerprint-based device authentication management method and system. Background Art

[0002] The device authentication management method based on electromagnetic fingerprint is a technical solution to identify and authenticate the device by collecting and analyzing the electromagnetic field characteristics (i.e., electromagnetic fingerprint) generated by the device in normal working state. Each electronic device will emit unique electromagnetic signals during use. These signals have a certain stability and uniqueness, similar to fingerprints. Therefore, the electromagnetic fingerprint can be used as the "digital identity" of the device to ensure the authenticity and security of the device.

[0003] The core function of this method is to achieve secure authentication of devices and prevent device forgery and tampering through electromagnetic fingerprint recognition technology. It can be widely used in the fields of identity authentication, network security protection, device tracking management, etc. of IoT devices to ensure that devices are protected from unauthorized access or attacks during connection and communication. In addition, electromagnetic fingerprint can provide a contactless and non-invasive authentication method to improve the efficiency and security of device management.

[0004] However, in the existing device authentication management methods based on electromagnetic fingerprints, there is a problem that the existing electromagnetic fingerprint construction process often uses machine learning methods or manual methods to extract features, and then constructs electromagnetic fingerprints to characterize electromagnetic devices. However, the electromagnetic fingerprints obtained by traditional methods contain many interference factors, and it is difficult to effectively standardize storage, and it is also unable to adapt to subsequent fine comparison and management updates. Technical problems; in the existing electromagnetic fingerprint comparison methods, there are traditional algorithm electromagnetic fingerprint comparison or matching methods, which often find it difficult to solve the difference problems caused by noise sensitivity and environmental change sensitivity. At the same time, due to the poor matching performance of traditional comparison algorithms for global comparison, it also lacks adaptability to dynamic changes, which leads to the technical problem that it is difficult to meet the intelligent needs of this solution; in the existing dynamic fingerprint update, there is a new idea for enhancing security in the electromagnetic fingerprint device authentication process, and the existing technology mainly focuses on updating the content of the fingerprint itself for dynamic fingerprint update, but the single update of electromagnetic fingerprint data itself is difficult to effectively support the intelligent needs of the integrated device authentication process, but may lead to data redundancy and reduce the accuracy and efficiency of device authentication. Technical problems. Summary of the invention

[0005] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a device authentication management method and system based on electromagnetic fingerprint. In the existing device authentication management method based on electromagnetic fingerprint, the existing electromagnetic fingerprint construction process often uses machine learning methods or manual methods to extract features, and then constructs electromagnetic fingerprints to characterize electromagnetic equipment. However, the electromagnetic fingerprint obtained by the traditional method itself contains many interference factors, and it is difficult to effectively standardize storage, and it is also unable to adapt to the subsequent fine comparison and management update technical problems. This solution creatively adopts a method combining signal domain capture, calibration environment optimization and frequency band distinction enhancement to collect and enhance signals, and further constructs electromagnetic fingerprints, and updates the parameter combination of electromagnetic fingerprints, which can not only improve the data quality of the electromagnetic fingerprint itself, but also provide good data support and foundation for subsequent comparison, update and management; in the existing electromagnetic fingerprint comparison method, there is a traditional algorithm electromagnetic fingerprint comparison or matching method, which often finds it difficult to solve the difference problems caused by noise sensitivity and environmental change sensitivity. At the same time, due to the poor matching performance of the traditional comparison algorithm for global comparison, it also lacks adaptability to dynamic changes. This leads to the technical problem that it is difficult to meet the intelligent needs of this solution. This solution creatively adopts a convolutional long-term and short-term neural network combined with adaptive weighting to optimize fingerprint matching, and also optimizes the influence of environmental noise in the data collection stage, thereby improving the basic accuracy of electromagnetic fingerprint matching, and also making good preparations for the adaptability of subsequent dynamic updates; in the existing dynamic fingerprint update, there is a new idea for enhancing security in the electromagnetic fingerprint device authentication process, and the existing technology mainly focuses on updating the content of the fingerprint itself for dynamic fingerprint update, but the single update of the electromagnetic fingerprint data itself is difficult to effectively support the intelligent needs of the integrated device authentication process, but may lead to data redundancy and reduce the accuracy and efficiency of device authentication. The technical problem, this solution creatively adopts a reinforcement learning method combined with convolution feature optimization to perform dynamic fingerprint updates, and simultaneously updates the electromagnetic fingerprint library and the electromagnetic fingerprint itself through reinforcement learning, which can not only optimize the management automation of the electromagnetic fingerprint library, but also optimize the intelligence and security of the entire electromagnetic fingerprint authentication process, and provides a good idea exploration and feasibility practice research for the device authentication management method based on electromagnetic fingerprint.

[0006] The technical solution adopted by the present invention is as follows: The device authentication management method based on electromagnetic fingerprint provided by the present invention comprises the following steps:

[0007] Step S1: electromagnetic fingerprint collection;

[0008] Step S2: data enhancement;

[0009] Step S3: electromagnetic fingerprint management;

[0010] Step S4: electromagnetic fingerprint comparison;

[0011] Step S5: dynamic fingerprint update;

[0012] Step S6: Device authentication management.

[0013] Furthermore, in step S1, the electromagnetic fingerprint collection is used to collect the original data set required for device authentication management, specifically by collecting the electromagnetic radiation signal of the device through the electromagnetic collection device to obtain the electromagnetic fingerprint original data set, which specifically includes the following steps:

[0014] Step S11: starting the device, specifically starting the target device according to the operation mode of the working state, and collecting data under normal operating conditions to obtain an original motor signal data set;

[0015] The target device specifically refers to a permanent magnet synchronous linear motor and a motor driver;

[0016] Step S12: acquisition calibration, specifically using a signal acquisition tool, and setting the sampling rate, frequency range and sensitivity, performing acquisition calibration, and obtaining a calibration motor signal acquisition parameter set;

[0017] Step S13: signal domain capture, specifically, capturing the electromagnetic radiation signal of the target device from the original motor signal data set according to the calibration motor signal acquisition parameter set, and performing domain signal capture to obtain a domain signal data set;

[0018] The sub-domain signal capture specifically includes time domain signal capture and frequency domain signal capture;

[0019] The frequency domain signal capture is specifically performed by using a spectrum analyzer to capture frequency distribution data to obtain frequency band signal data, wherein the frequency distribution specifically includes a low frequency band and a high frequency band;

[0020] Step S14: environmental isolation, specifically, when executing the operations of step S11 and step S13, using a shielding box to isolate the external electromagnetic environment interference, and performing interference filtering operations on the collected domain signal data sets to obtain basic optimized signal data;

[0021] Step S15: initial electromagnetic fingerprint construction, specifically, initializing and constructing electromagnetic fingerprint data by combining the sub-domain signal data sets to obtain an electromagnetic fingerprint original data set;

[0022] The electromagnetic fingerprint original data set is used to represent the original signal characteristics, random noise characteristics and repeatable stable characteristics of the electromagnetic signal.

[0023] Further, in step S2, the data enhancement is used to enhance the original data, specifically using a frequency band distinction enhancement method to enhance the original electromagnetic fingerprint data set to generate an enhanced electromagnetic fingerprint data set, specifically including the following steps:

[0024] Step S21: enhancing recognition, specifically evaluating the resonant frequency of the data in the electromagnetic fingerprint original data set by minimizing the electromagnetic impedance, and obtaining a signal change recognition enhancement frequency parameter;

[0025] Step S22: state differentiation enhancement, specifically, performing dual-frequency analysis on the data in the electromagnetic fingerprint original data set by distinguishing low-frequency and high-frequency electromagnetic radiation to obtain state differentiation electromagnetic impedance parameters;

[0026] Step S23: reactance combination enhancement, specifically, performing reactance combination enhancement on the data in the electromagnetic fingerprint original data set by removing the idealized electromagnetic response generated by the electromagnetism antigen component, so as to obtain the reactance elimination enhanced electromagnetic fingerprint combination data;

[0027] Step S24: resistance combination enhancement, specifically, performing resistance combination enhancement on the data in the electromagnetic fingerprint combination data by eliminating the reactance and removing the characteristic distortion in the resistance signal to obtain the resistance elimination enhanced electromagnetic fingerprint combination data;

[0028] Step S25: mutual inductance normalization enhancement, specifically, performing mutual inductance elimination enhancement on the reactance elimination enhancement electromagnetic fingerprint combination data and the resistance elimination enhancement electromagnetic fingerprint combination data to obtain mutual inductance normalization enhancement electromagnetic fingerprint combination data;

[0029] Step S26: electromagnetic fingerprint enhancement, specifically, performing electromagnetic fingerprint enhancement through the recognition enhancement, state differentiation enhancement, reactance combination enhancement, resistance combination enhancement and mutual inductance normalization enhancement to obtain an enhanced electromagnetic fingerprint data set.

[0030] Further, in step S3, the electromagnetic fingerprint management is used to construct a structured electromagnetic fingerprint library, specifically, based on the enhanced electromagnetic fingerprint data set, an enhanced electromagnetic fingerprint parameter set is constructed, and the electromagnetic fingerprint library is structuredly stored and managed to obtain a structured electromagnetic fingerprint library;

[0031] Further, in step S4, the electromagnetic fingerprint comparison is used to read the electromagnetic fingerprint in the electromagnetic fingerprint library and perform device authentication matching, specifically, based on the enhanced electromagnetic fingerprint data set, a convolutional long-term and short-term neural network combined with adaptive weighting improvement is used to perform electromagnetic fingerprint comparison to obtain electromagnetic fingerprint comparison result reference data;

[0032] The improved convolutional long short-term neural network combined with adaptive weighting specifically includes a convolution extraction subnet, a long short-term memory subnet, an adaptive weighted difference operator and a comparison classifier;

[0033] The convolutional extraction subnet is used to extract spatial features in the enhanced electromagnetic fingerprint dataset;

[0034] The long short-term memory subnet is used to extract the time dimension features in the enhanced electromagnetic fingerprint dataset;

[0035] The adaptive weighted difference operator is used to adaptively weight the differences between different features;

[0036] The comparison classifier is used to perform electromagnetic fingerprint comparison and classification in combination with the result of adaptive weighting;

[0037] The step of using the convolutional long-term and short-term neural network improved by combining adaptive weighting to perform electromagnetic fingerprint comparison and obtain electromagnetic fingerprint comparison result reference data includes:

[0038] Step S41: constructing a convolution extraction subnet, specifically constructing a standard convolution extraction subnet including a convolution layer and a pooling layer to obtain electromagnetic fingerprint convolution feature data;

[0039] Step S42: constructing a long short-term memory subnet, specifically constructing a standard long short-term memory neural network as a long short-term memory subnet, and extracting the time dimension features in the enhanced electromagnetic fingerprint data set to obtain electromagnetic fingerprint time series feature data;

[0040] Step S43: construct an adaptive weighted difference operator, specifically, perform feature fusion on the electromagnetic fingerprint convolution feature data and the electromagnetic fingerprint time series feature data, and perform feature weighting by constructing an adaptive weighted difference operator to obtain adaptive feature weight data. The calculation formula is:

[0041]

[0042] In the formula, Fe is the fusion feature data, y cnn is the electromagnetic fingerprint convolution feature data, y lstm is the electromagnetic fingerprint time series feature data, [·;·] is the feature concatenation operator, AWD(·) is the output of the adaptive weighted difference operator, which is used to represent the adaptive weighted difference value of the feature, Fe 1 is the first electromagnetic fingerprint weighted feature, Fe 2 is the weighted feature of the second electromagnetic fingerprint, I is the total number of features, i is the feature index, and w i is the weighting factor;

[0043] Step S44: construct a comparison classifier, specifically select a standard support vector machine model as a classifier, and determine whether the two electromagnetic fingerprints match by comparison and classification, and obtain comparison classification data

[0044] Step S45: electromagnetic fingerprint comparison model training, specifically, performing electromagnetic fingerprint comparison model training through the convolution extraction subnet, the long short-term memory subnet, the adaptive weighted difference operator and the comparison classifier to obtain the electromagnetic fingerprint comparison model Model LCA ;

[0045] Step S46: electromagnetic fingerprint comparison, specifically using the electromagnetic fingerprint comparison model Model LCA , performing electromagnetic fingerprint comparison according to the enhanced electromagnetic fingerprint data set to obtain electromagnetic fingerprint comparison result reference data;

[0046] The electromagnetic fingerprint comparison result reference data specifically includes the comparison result type and the comparison quantitative matching degree.

[0047] Further, in step S5, the dynamic fingerprint update is used to dynamically update the electromagnetic fingerprint library for optimization management, specifically, based on the enhanced electromagnetic fingerprint parameter set, a reinforcement learning method combined with convolution feature optimization is used to perform dynamic fingerprint update to obtain an updated electromagnetic fingerprint management data set, including the following steps:

[0048] Step S51: initializing the environment state, specifically initializing the structured electromagnetic fingerprint library and the enhanced electromagnetic fingerprint parameter set to obtain an initialized electromagnetic fingerprint update parameter set;

[0049] Step S52: state feature convolution modeling, specifically constructing a standard convolutional neural network including a convolution layer, a pooling layer and a fully connected layer, extracting features from the structured electromagnetic fingerprint library and the initialized electromagnetic fingerprint update parameter set, and obtaining initialized electromagnetic fingerprint state feature vector data;

[0050] Step S53: Initializing the reinforcement learning model, specifically, constructing an electromagnetic fingerprint library interactive update reinforcement learning initial model based on the initialized electromagnetic fingerprint state feature vector data, specifically including state parameters, action parameters, reward parameters and objective function;

[0051] The state parameter is used to indicate the electromagnetic fingerprint state of the device, specifically the current value of the electromagnetic fingerprint parameter;

[0052] The action parameter is used to indicate the update action of the battery fingerprint library, specifically including the update fingerprint operation, the delete fingerprint operation and the add new fingerprint operation;

[0053] The update fingerprint operation is used to update the fingerprint library by changing the existing electromagnetic fingerprint data to improve the comparison accuracy; the delete fingerprint operation is used to delete the electromagnetic fingerprint that has not been updated for a long time or has a small contribution to the comparison result; the add new fingerprint operation is used to update the fingerprint library by adding new electromagnetic fingerprint data to improve the comparison accuracy;

[0054] The reward parameter is used to calculate the reward according to the comparison accuracy of the updated electromagnetic fingerprint library;

[0055] The objective function is used to calculate the maximum accumulated reward parameter value;

[0056] The electromagnetic fingerprint library interactively updates the initial reinforcement learning model, which is used to represent the parameter combination for executing reinforcement learning model training. The calculation formula is:

[0057] RL = {s t ,a t ,r t ,R};

[0058] Where RL is the initial reinforcement learning model for interactive updating of electromagnetic fingerprint library, s t is the state parameter, a t is the action parameter, r t is the reward parameter, R is the maximum accumulated reward parameter value;

[0059] Step S54: construct a dynamic fingerprint update strategy, specifically, construct a state update output based on the state transfer function, and perform deep Q value learning based on the state update output, and perform management strategy update of the structured electromagnetic fingerprint library and update parameter optimization of the electromagnetic fingerprint update parameter set by combining the state feature convolution modeling and the reinforcement learning model initialization, and obtain the dynamic fingerprint update model Model by performing iterative training. CR ;

[0060] Step S55: Dynamic fingerprint adaptive update optimization, specifically using the dynamic fingerprint to update the model Model CR , performing dynamic fingerprint update according to the enhanced electromagnetic fingerprint parameter set to obtain an updated electromagnetic fingerprint management data set;

[0061] The updated electromagnetic fingerprint management data set specifically includes structured electromagnetic fingerprint library management reference data and electromagnetic fingerprint update parameter set data;

[0062] The structured electromagnetic fingerprint library manages reference data, specifically including updating, deleting or optimizing the data in the electromagnetic fingerprint library, and realizing dynamic update management of the structured electromagnetic fingerprint library by managing reference data through the structured electromagnetic fingerprint library;

[0063] The electromagnetic fingerprint update parameter set data specifically refers to the electromagnetic fingerprint parameter set after optimizing the enhanced electromagnetic fingerprint parameter set. The dynamic update of the electromagnetic fingerprint is achieved based on the optimized electromagnetic fingerprint parameter set.

[0064] Further, in step S6, the device authentication management is used to perform comprehensive device authentication management in combination with the electromagnetic fingerprint library and the device matching result, specifically by performing the electromagnetic fingerprint management and the electromagnetic fingerprint comparison according to the enhanced electromagnetic fingerprint data set, and by performing dynamic fingerprint update according to the updated electromagnetic fingerprint management data set, and continuing to perform comprehensive device authentication management according to the result of the dynamic fingerprint update, to obtain electromagnetic fingerprint device authentication reference data;

[0065] The electromagnetic fingerprint device authentication reference data specifically includes electromagnetic fingerprint comparison and matching reference data and electromagnetic fingerprint update management reference data.

[0066] The electromagnetic fingerprint-based device authentication management system provided by the present invention includes a collection and processing module, a database building management module, a comparison and authentication module, a dynamic update module and an authentication management module;

[0067] The acquisition and processing module is used for electromagnetic fingerprint acquisition and data enhancement, obtains an enhanced electromagnetic fingerprint data set through electromagnetic fingerprint acquisition and data enhancement, and sends the enhanced electromagnetic fingerprint data set to the database construction management module, the comparison and authentication module and the dynamic update module;

[0068] The library building management module is used for electromagnetic fingerprint management, and obtains an enhanced electromagnetic fingerprint parameter set and a structured electromagnetic fingerprint library through electromagnetic fingerprint management, and sends the enhanced electromagnetic fingerprint parameter set to the dynamic update module, and performs dynamic fingerprint update according to the structured electromagnetic fingerprint library;

[0069] The comparison and authentication module is used for electromagnetic fingerprint comparison, obtains electromagnetic fingerprint comparison result reference data through electromagnetic fingerprint comparison, and sends the electromagnetic fingerprint comparison result reference data to the authentication management module;

[0070] The dynamic update module is used for dynamic fingerprint update, obtains an updated electromagnetic fingerprint management data set through dynamic fingerprint update, and sends the updated electromagnetic fingerprint management data set to the authentication management module;

[0071] The authentication management module is used for device authentication management, and obtains electromagnetic fingerprint device authentication reference data through device authentication management.

[0072] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0073] (1) In the existing device authentication management methods based on electromagnetic fingerprints, the existing electromagnetic fingerprint construction process often uses machine learning methods or manual methods to extract features, and then constructs electromagnetic fingerprints to characterize electromagnetic devices. However, the electromagnetic fingerprints obtained by traditional methods contain many interference factors, and are difficult to store in an effective standardized manner. At the same time, they cannot adapt to the technical problems of subsequent fine comparison and management updates. This solution creatively adopts a method that combines signal domain capture, calibration environment optimization, and frequency band distinction enhancement to collect and enhance signals, and further constructs electromagnetic fingerprints, as well as updates the parameter combination of electromagnetic fingerprints. This can not only improve the data quality of the electromagnetic fingerprint itself, but also provide good data support and foundation for subsequent comparison, update, and management.

[0074] (2) In the existing electromagnetic fingerprint comparison methods, there are traditional algorithm electromagnetic fingerprint comparison or matching methods, which often have difficulty solving the difference problems caused by noise sensitivity and environmental change sensitivity. At the same time, due to the poor matching performance of traditional comparison algorithms for global comparison and the lack of adaptability to dynamic changes, it is difficult to meet the technical problems of the intelligent needs of this solution. This solution creatively adopts a convolutional long-term and short-term neural network combined with adaptive weighting to optimize fingerprint comparison. At the same time, it also optimizes the influence of environmental noise in the data collection stage, improves the basic accuracy of electromagnetic fingerprint comparison, and also makes good preparations for the adaptability of subsequent dynamic updates;

[0075] (3) In the existing dynamic fingerprint update, dynamic update is a new idea for enhancing security in the electromagnetic fingerprint device authentication process. The existing technology for dynamic fingerprint update mainly focuses on updating the content of the fingerprint itself. However, the single update of the electromagnetic fingerprint data itself is difficult to effectively support the intelligent needs of the integrated device authentication process. On the contrary, it may lead to data redundancy and reduce the accuracy and efficiency of device authentication. Technical problems. This scheme creatively adopts a reinforcement learning method combined with convolutional feature optimization to perform dynamic fingerprint update. Through reinforcement learning, the electromagnetic fingerprint library and the electromagnetic fingerprint itself are updated simultaneously, which can not only optimize the management automation of the electromagnetic fingerprint library, but also optimize the intelligence and security of the entire electromagnetic fingerprint authentication process, providing a good idea exploration and feasibility practice research for the device authentication management method based on electromagnetic fingerprint. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 A schematic diagram of the process of the device authentication management method based on electromagnetic fingerprint provided by the present invention;

[0077] Figure 2 A schematic diagram of the device authentication management system based on electromagnetic fingerprint provided by the present invention;

[0078] Figure 3This is a schematic diagram of the process of electromagnetic fingerprint collection in step S1;

[0079] Figure 4 Schematic diagram of the process of data enhancement in step S2;

[0080] Figure 5 This is a schematic diagram of the process of electromagnetic fingerprint comparison in step S4;

[0081] Figure 6 This is a schematic diagram of the process of dynamic fingerprint update in step S5.

[0082] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0083] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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 ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0084] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0085] Example 1, see Figure 1 The present invention provides a device authentication management method based on electromagnetic fingerprint, which comprises the following steps:

[0086] Step S1: electromagnetic fingerprint collection;

[0087] Step S2: data enhancement;

[0088] Step S3: electromagnetic fingerprint management;

[0089] Step S4: electromagnetic fingerprint comparison;

[0090] Step S5: dynamic fingerprint update;

[0091] Step S6: Device authentication management.

[0092] Example 2, see Figure 1 , Figure 2 and Figure 3 In step S1, the electromagnetic fingerprint collection is used to collect the original data set required for device authentication management, specifically by collecting the electromagnetic radiation signal of the device through the electromagnetic collection device to obtain the electromagnetic fingerprint original data set, which specifically includes the following steps:

[0093] Step S11: starting the device, specifically starting the target device according to the operation mode of the working state, and collecting data under normal operating conditions to obtain an original motor signal data set;

[0094] The target device specifically refers to a permanent magnet synchronous linear motor and a motor driver;

[0095] The normal operating conditions specifically refer to starting the linear motor under no-load and full-load conditions and collecting electromagnetic signals;

[0096] Step S12: acquisition calibration, specifically using a signal acquisition tool, and setting the sampling rate, frequency range and sensitivity, performing acquisition calibration, and obtaining a calibration motor signal acquisition parameter set;

[0097] The signal acquisition tool specifically refers to a spectrum analyzer;

[0098] The sampling rate is specifically set to 100kHz;

[0099] The frequency range is specifically set to 50kHz to 500kHz;

[0100] The sensitivity specifically refers to the sensitivity of the spectrum analyzer, which is manually set according to the signal strength;

[0101] Step S13: signal domain capture, specifically, capturing the electromagnetic radiation signal of the target device from the original motor signal data set according to the calibration motor signal acquisition parameter set, and performing domain signal capture to obtain a domain signal data set;

[0102] The sub-domain signal capture specifically includes time domain signal capture and frequency domain signal capture;

[0103] The time domain signal capture is specifically to capture the current and voltage waveforms through an oscilloscope, and to perform video analysis using a short-time Fourier transform method to obtain electromagnetic signal waveform data and video image data;

[0104] The frequency domain signal capture is specifically performed by using a spectrum analyzer to capture frequency distribution data to obtain frequency band signal data, wherein the frequency distribution specifically includes a low frequency band and a high frequency band;

[0105] The low frequency band is used to represent the basic operating frequency of the motor, and the specific value range is 50kHz to 100kHz;

[0106] The high frequency band is used to represent the switching frequency of the motor driver, and the specific value range is 100kHz to 500kHz;

[0107] Step S14: environmental isolation, specifically, when executing the operations of step S11 and step S13, using a shielding box to isolate the external electromagnetic environment interference, and performing interference filtering operations on the collected domain signal data sets to obtain basic optimized signal data;

[0108] Step S15: initial electromagnetic fingerprint construction, specifically, initializing and constructing electromagnetic fingerprint data by combining the sub-domain signal data sets to obtain an electromagnetic fingerprint original data set;

[0109] The electromagnetic fingerprint original data set is used to represent the original signal characteristics, random noise characteristics and repeatable stable characteristics of the electromagnetic signal.

[0110] Example 3, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S2, the data enhancement is used to enhance the original data. Specifically, the frequency band distinction enhancement method is used to enhance the original electromagnetic fingerprint data set to generate an enhanced electromagnetic fingerprint data set, and the enhanced electromagnetic fingerprint data set is obtained. Specifically, the following steps are included:

[0111] Step S21: Recognition enhancement, specifically, evaluating the resonant frequency of the data in the electromagnetic fingerprint original data set by minimizing the electromagnetic impedance, and obtaining the signal change recognition enhancement frequency parameter, the calculation formula is:

[0112]

[0113] In the formula, ω s is the signal change recognition enhancement frequency parameter, ω is the frequency independent variable, which is used to represent the response of the motor or drive at different frequencies, and Z R (·) is the electromagnetic impedance response function, which is used to represent the electromagnetic characteristics of the motor. is the calculation function of the minimized electromagnetic impedance amplitude corresponding to the frequency independent variable ω, and m is the load parameter, which is used to represent the influence of the motor load on the electromagnetic signal;

[0114] Step S22: state differentiation enhancement, specifically, performing dual-frequency analysis on the data in the electromagnetic fingerprint original data set by distinguishing low-frequency and high-frequency electromagnetic radiation to obtain state differentiation electromagnetic impedance parameters, and the calculation formula is:

[0115]

[0116] In the formula, Z R (αωs , m) is the high frequency band parameter in the state-differentiated electromagnetic impedance parameter, is the low-frequency band parameter in the state-differentiated electromagnetic impedance parameter, α is the state-differentiated parameter, which is used to represent the ratio of the high-frequency band frequency to the low-frequency band frequency in the dual-frequency phenomenon, ω s is the signal change recognition enhancement frequency parameter, used to indicate the resonant frequency, m is the load parameter, r p is the resistance value, j is an imaginary unit used to represent the phase information in the electrical signal, l R is the inductance value, which is used to express the relationship between the current change and the voltage in the motor, σ(·) is the electromagnetic characteristic function, which is specifically used to express the frequency response characteristics;

[0117] Step S23: reactance combination enhancement, specifically, by removing the idealized electromagnetic response generated by the electromagnetism antigen component from the data in the electromagnetic fingerprint original data set, and performing reactance combination enhancement to obtain the reactance elimination enhanced electromagnetic fingerprint combination data, the calculation formula is:

[0118]

[0119] In the formula, F d (α, m) is the dual-frequency electromagnetic fingerprint, which is used to represent the combined electromagnetic fingerprint data of the high-frequency band and the low-frequency band after the reactance elimination is synthesized. α is the state distinction parameter, ω s is the signal change recognition enhancement frequency parameter, m is the load parameter, r p is the resistance value, σ(·) is the electromagnetic characteristic function;

[0120] Step S24: resistance combination enhancement, specifically, by eliminating the data in the electromagnetic fingerprint combination data of the reactance enhancement, and removing the characteristic distortion in the resistance signal, the resistance combination enhancement is performed to obtain the resistance elimination enhancement electromagnetic fingerprint combination data, and the calculation formula is:

[0121]

[0122] In the formula, F s (α, m) is the resistance elimination enhanced electromagnetic fingerprint combination data, F d (α, m) is the dual-frequency electromagnetic fingerprint, F d (βα, m) is the scaled dual-frequency electromagnetic fingerprint, where β is the frequency scanning range scaling factor, α is the state discrimination parameter, and ω s is the signal change recognition enhancement frequency parameter, m is the load parameter, and σ(·) is the electromagnetic characteristic function;

[0123] Step S25: mutual inductance normalization enhancement, specifically, performing mutual inductance elimination enhancement on the reactance elimination enhancement electromagnetic fingerprint combination data and the resistance elimination enhancement electromagnetic fingerprint combination data to obtain mutual inductance normalization enhancement electromagnetic fingerprint combination data, and the calculation formula is:

[0124]

[0125] In the formula, F o (α) is the mutual inductance normalized enhanced electromagnetic fingerprint combination data, Norm(·) is the normalization operation, F s (α, m) is the resistance elimination enhanced electromagnetic fingerprint combination data, β is the frequency scanning range scaling factor, α is the state distinction parameter, ω s is the signal change recognition enhancement frequency parameter, m is the load parameter, and σ(·) is the electromagnetic characteristic function;

[0126] Step S26: electromagnetic fingerprint enhancement, specifically, performing electromagnetic fingerprint enhancement through the recognition enhancement, state differentiation enhancement, reactance combination enhancement, resistance combination enhancement and mutual inductance normalization enhancement to obtain an enhanced electromagnetic fingerprint data set.

[0127] By performing the above operations, in the existing device authentication management method based on electromagnetic fingerprint, the existing electromagnetic fingerprint construction process often uses machine learning methods or manual methods to extract features, and then constructs electromagnetic fingerprints to characterize electromagnetic equipment. However, the electromagnetic fingerprint obtained by the traditional method itself contains many interference factors, and it is difficult to effectively standardize storage. At the same time, it cannot adapt to the subsequent technical problems of fine comparison and management update. This solution creatively adopts a method combining signal domain capture, calibration environment optimization and frequency band distinction enhancement to collect and enhance signals, and further constructs electromagnetic fingerprints, as well as updates the parameter combination of electromagnetic fingerprints, which can not only improve the data quality of the electromagnetic fingerprint itself, but also provide good data support and foundation for subsequent comparison, update and management.

[0128] Example 4, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S3, the electromagnetic fingerprint management is used to construct a structured electromagnetic fingerprint library. Specifically, based on the enhanced electromagnetic fingerprint data set, an enhanced electromagnetic fingerprint parameter set is constructed, and the electromagnetic fingerprint library is structured and stored and managed to obtain a structured electromagnetic fingerprint library. The calculation formula of the enhanced electromagnetic fingerprint parameter set is:

[0129] θ={m,ω s ,α,β};

[0130] Where θ is the enhanced electromagnetic fingerprint parameter set, m is the load parameter, α is the state distinction parameter, ω sis the signal change recognition enhancement frequency parameter, β is the frequency scanning range scaling factor;

[0131] The specific value range of the state distinction parameter α is [1,2];

[0132] The specific value range of the frequency scanning range scaling coefficient β is [1,2];

[0133] The signal change recognition enhancement frequency parameter ω s The initial value of is 200kHz, and the specific operating frequency range of the motor is 50kHz to 500kHz.

[0134] Example 5, see Figure 1 , Figure 2 and Figure 5 In step S4, the electromagnetic fingerprint comparison is used to read the electromagnetic fingerprint in the electromagnetic fingerprint library and perform device authentication matching. Specifically, based on the enhanced electromagnetic fingerprint data set, a convolutional long-term and short-term neural network combined with adaptive weighting is used to perform electromagnetic fingerprint comparison to obtain electromagnetic fingerprint comparison result reference data;

[0135] The improved convolutional long short-term neural network combined with adaptive weighting specifically includes a convolution extraction subnet, a long short-term memory subnet, an adaptive weighted difference operator and a comparison classifier;

[0136] The convolutional extraction subnet is used to extract spatial features in the enhanced electromagnetic fingerprint dataset;

[0137] The long short-term memory subnet is used to extract the time dimension features in the enhanced electromagnetic fingerprint dataset;

[0138] The adaptive weighted difference operator is used to adaptively weight the differences between different features;

[0139] The comparison classifier is used to perform electromagnetic fingerprint comparison and classification in combination with the result of adaptive weighting;

[0140] The step of using the convolutional long-term and short-term neural network improved by combining adaptive weighting to perform electromagnetic fingerprint comparison and obtain electromagnetic fingerprint comparison result reference data includes:

[0141] Step S41: constructing a convolution extraction subnet, specifically constructing a standard convolution extraction subnet including a convolution layer and a pooling layer to obtain electromagnetic fingerprint convolution feature data;

[0142] The calculation formula for constructing the convolution extraction subnetwork is:

[0143] y cnn =MaxPool(ReLU(W C ·x+b C ));

[0144] In the formula, y cnn is the electromagnetic fingerprint convolution feature data, MaxPool(·) is the maximum pooling operation function, and ReLU(·) is the fiber box activation function W C is the convolution weight, x is the convolution data input, used to represent the enhanced electromagnetic fingerprint dataset, b C is the convolution bias term;

[0145] Step S42: constructing a long short-term memory subnet, specifically constructing a standard long short-term memory neural network as a long short-term memory subnet, and extracting the time dimension features in the enhanced electromagnetic fingerprint data set to obtain electromagnetic fingerprint time series feature data;

[0146] Step S43: construct an adaptive weighted difference operator, specifically, perform feature fusion on the electromagnetic fingerprint convolution feature data and the electromagnetic fingerprint time series feature data, and perform feature weighting by constructing an adaptive weighted difference operator to obtain adaptive feature weight data. The calculation formula is:

[0147]

[0148] In the formula, Fe is the fusion feature data, y cnn is the electromagnetic fingerprint convolution feature data, y lstm is the electromagnetic fingerprint time series feature data, [·;·] is the feature concatenation operator, AWD(·) is the output of the adaptive weighted difference operator, which is used to represent the adaptive weighted difference value of the feature, Fe 1 is the first electromagnetic fingerprint weighted feature, Fe 2 is the weighted feature of the second electromagnetic fingerprint, I is the total number of features, i is the feature index, and w i is the weighting factor;

[0149] Step S44: construct a comparison classifier, specifically, select a standard support vector machine model as a classifier, and judge whether the two electromagnetic fingerprints match by comparison and classification, and obtain comparison classification data. The calculation formula is:

[0150] y=sig(W S AWD (Fe 1 , Fe 2 )+b S );

[0151] Where y is the comparison classification data, sig(·) is the S-type activation function, and W S is the classifier weight, AWD(·) is the adaptive weighted difference operator output, which is used to represent the adaptive weighted difference value of the feature, Fe 1 is the first electromagnetic fingerprint weighted feature, Fe 2is the second electromagnetic fingerprint weighted feature, b S is the classifier bias;

[0152] Step S45: electromagnetic fingerprint comparison model training, specifically, performing electromagnetic fingerprint comparison model training through the convolution extraction subnet, the long short-term memory subnet, the adaptive weighted difference operator and the comparison classifier to obtain the electromagnetic fingerprint comparison model Model LCA ;

[0153] Step S46: electromagnetic fingerprint comparison, specifically using the electromagnetic fingerprint comparison model Model LCA , performing electromagnetic fingerprint comparison according to the enhanced electromagnetic fingerprint data set to obtain electromagnetic fingerprint comparison result reference data;

[0154] The electromagnetic fingerprint comparison result reference data specifically includes the comparison result type and the comparison quantitative matching degree.

[0155] By performing the above operations, in the existing electromagnetic fingerprint comparison methods, there are traditional algorithm electromagnetic fingerprint comparison or matching methods, which often find it difficult to solve the difference problems caused by noise sensitivity and environmental change sensitivity. At the same time, since the traditional comparison algorithm has poor matching performance for global comparison and lacks adaptability to dynamic changes, it is difficult to meet the technical problem of the intelligent needs of this solution. This solution creatively adopts a convolutional long-short-term neural network combined with adaptive weighting to optimize fingerprint comparison, and also optimizes the influence of environmental noise in the data acquisition stage, thereby improving the basic accuracy of electromagnetic fingerprint comparison and making good preparations for the adaptability of subsequent dynamic updates.

[0156] Example 6, see Figure 1 , Figure 2 and Figure 6 This embodiment is based on the above embodiment. In step S5, the dynamic fingerprint update is used to dynamically update the electromagnetic fingerprint library for optimization management. Specifically, according to the enhanced electromagnetic fingerprint parameter set, a reinforcement learning method combined with convolution feature optimization is used to perform dynamic fingerprint update to obtain an updated electromagnetic fingerprint management data set, including the following steps:

[0157] Step S51: initializing the environment state, specifically initializing the structured electromagnetic fingerprint library and the enhanced electromagnetic fingerprint parameter set to obtain an initialized electromagnetic fingerprint update parameter set;

[0158] Step S52: state feature convolution modeling, specifically constructing a standard convolutional neural network including a convolution layer, a pooling layer and a fully connected layer, extracting features from the structured electromagnetic fingerprint library and the initialized electromagnetic fingerprint update parameter set, and obtaining initialized electromagnetic fingerprint state feature vector data;

[0159] Step S53: Initializing the reinforcement learning model, specifically, constructing an electromagnetic fingerprint library interactive update reinforcement learning initial model based on the initialized electromagnetic fingerprint state feature vector data, specifically including state parameters, action parameters, reward parameters and objective function;

[0160] The state parameter is used to indicate the electromagnetic fingerprint state of the device, specifically the current value of the electromagnetic fingerprint parameter;

[0161] The action parameter is used to indicate the update action of the battery fingerprint library, specifically including the update fingerprint operation, the delete fingerprint operation and the add new fingerprint operation;

[0162] The update fingerprint operation is used to update the fingerprint library by changing the existing electromagnetic fingerprint data to improve the comparison accuracy; the delete fingerprint operation is used to delete the electromagnetic fingerprint that has not been updated for a long time or has a small contribution to the comparison result; the add new fingerprint operation is used to update the fingerprint library by adding new electromagnetic fingerprint data to improve the comparison accuracy;

[0163] The reward parameter is used to calculate the reward according to the comparison accuracy of the updated electromagnetic fingerprint library. The specific calculation formula is:

[0164]

[0165] In the formula, r t is the reward parameter, t is the iteration index;

[0166] The objective function is used to calculate the maximum accumulated reward parameter value, and the calculation formula is:

[0167]

[0168] Where R is the maximum cumulative reward parameter value, T is the maximum number of iterations, t is the iteration index, and γ t is the discount factor at the tth iteration, obtained through adaptive learning, r t is the reward parameter;

[0169] The electromagnetic fingerprint library interactively updates the initial reinforcement learning model, which is used to represent the parameter combination for executing reinforcement learning model training. The calculation formula is:

[0170] RL = {s t ,a t ,r t ,R};

[0171] Where RL is the initial reinforcement learning model for interactive updating of electromagnetic fingerprint library, s t is the state parameter, a t is the action parameter, r t is the reward parameter, R is the maximum accumulated reward parameter value;

[0172] Step S54: construct a dynamic fingerprint update strategy, specifically, construct a state update output based on the state transfer function, and perform deep Q value learning based on the state update output, and perform management strategy update of the structured electromagnetic fingerprint library and update parameter optimization of the electromagnetic fingerprint update parameter set by combining the state feature convolution modeling and the reinforcement learning model initialization, and obtain the dynamic fingerprint update model Model by performing iterative training. CR ;

[0173] Step S55: Dynamic fingerprint adaptive update optimization, specifically using the dynamic fingerprint to update the model Model CR , performing dynamic fingerprint update according to the enhanced electromagnetic fingerprint parameter set to obtain an updated electromagnetic fingerprint management data set;

[0174] The updated electromagnetic fingerprint management data set specifically includes structured electromagnetic fingerprint library management reference data and electromagnetic fingerprint update parameter set data;

[0175] The structured electromagnetic fingerprint library manages reference data, specifically including updating, deleting or optimizing the data in the electromagnetic fingerprint library, and realizing dynamic update management of the structured electromagnetic fingerprint library by managing reference data through the structured electromagnetic fingerprint library;

[0176] The electromagnetic fingerprint update parameter set data specifically refers to the electromagnetic fingerprint parameter set after optimizing the enhanced electromagnetic fingerprint parameter set. The dynamic update of the electromagnetic fingerprint is achieved based on the optimized electromagnetic fingerprint parameter set.

[0177] By performing the above operations, in the existing dynamic fingerprint update, dynamic update is a new idea for enhancing security in the electromagnetic fingerprint device authentication process, and the existing technology mainly focuses on updating the content of the fingerprint itself for dynamic fingerprint update, but the single update of electromagnetic fingerprint data itself is difficult to effectively support the intelligent needs of the integrated device authentication process, but may lead to data redundancy and reduce the accuracy and efficiency of device authentication. Technical problems, this solution creatively adopts a reinforcement learning method combined with convolutional feature optimization to perform dynamic fingerprint updates, and simultaneously updates the electromagnetic fingerprint library and the electromagnetic fingerprint itself through reinforcement learning, which can not only optimize the management automation of the electromagnetic fingerprint library, but also optimize the intelligence and security of the entire electromagnetic fingerprint authentication process, and provides a good idea exploration and feasibility practice research for the device authentication management method based on electromagnetic fingerprint.

[0178] Embodiment 7, see Figure 1 and Figure 2This embodiment is based on the above embodiment. In step S6, the device authentication management is used to perform comprehensive device authentication management in combination with the electromagnetic fingerprint library and the device matching result, specifically, the electromagnetic fingerprint management and the electromagnetic fingerprint comparison are performed according to the enhanced electromagnetic fingerprint data set, and the dynamic fingerprint update is performed according to the updated electromagnetic fingerprint management data set, and the comprehensive device authentication management is continued according to the dynamic update result to obtain the electromagnetic fingerprint device authentication reference data;

[0179] The electromagnetic fingerprint device authentication reference data specifically includes electromagnetic fingerprint comparison and matching reference data and electromagnetic fingerprint update management reference data.

[0180] Embodiment 8, see Figure 1 and Figure 2 , this embodiment is based on the above embodiment, the device authentication management system based on electromagnetic fingerprint provided by the present invention includes a collection and processing module, a database management module, a comparison and authentication module, a dynamic update module and an authentication management module;

[0181] The acquisition and processing module is used for electromagnetic fingerprint acquisition and data enhancement, obtains an enhanced electromagnetic fingerprint data set through electromagnetic fingerprint acquisition and data enhancement, and sends the enhanced electromagnetic fingerprint data set to the database construction management module, the comparison and authentication module and the dynamic update module;

[0182] The library building management module is used for electromagnetic fingerprint management, and obtains an enhanced electromagnetic fingerprint parameter set and a structured electromagnetic fingerprint library through electromagnetic fingerprint management, and sends the enhanced electromagnetic fingerprint parameter set to the dynamic update module, and performs dynamic fingerprint update according to the structured electromagnetic fingerprint library;

[0183] The comparison and authentication module is used for electromagnetic fingerprint comparison, obtains electromagnetic fingerprint comparison result reference data through electromagnetic fingerprint comparison, and sends the electromagnetic fingerprint comparison result reference data to the authentication management module;

[0184] The dynamic update module is used for dynamic fingerprint update, obtains an updated electromagnetic fingerprint management data set through dynamic fingerprint update, and sends the updated electromagnetic fingerprint management data set to the authentication management module;

[0185] The authentication management module is used for device authentication management, and obtains electromagnetic fingerprint device authentication reference data through device authentication management.

[0186] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0187] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0188] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A device authentication management method based on electromagnetic fingerprint, characterized in that: The method comprises the following steps: Step S1: electromagnetic fingerprint collection, using electromagnetic collection equipment to collect electromagnetic radiation signals from the equipment to obtain an electromagnetic fingerprint original data set; Step S2: data enhancement, which is used to enhance the original data, specifically using a frequency band distinction enhancement method to enhance the electromagnetic fingerprint original data set to generate an enhanced electromagnetic fingerprint data set, specifically including the following steps: Step S21: recognition enhancement, specifically evaluating the resonant frequency of the data in the electromagnetic fingerprint original data set by minimizing the electromagnetic impedance, and obtaining a signal change recognition enhancement frequency parameter; Step S22: state differentiation enhancement, specifically, performing dual-frequency analysis on the data in the electromagnetic fingerprint original data set by distinguishing low-frequency and high-frequency electromagnetic radiation to obtain state differentiation electromagnetic impedance parameters; Step S23: reactance combination enhancement, specifically, performing reactance combination enhancement on the data in the electromagnetic fingerprint original data set by removing the idealized electromagnetic response generated by the electromagnetism antigen component, so as to obtain the reactance elimination enhanced electromagnetic fingerprint combination data; Step S24: resistance combination enhancement, specifically, performing resistance combination enhancement on the data in the electromagnetic fingerprint combination data by eliminating the reactance and removing the characteristic distortion in the resistance signal to obtain the resistance elimination enhanced electromagnetic fingerprint combination data; Step S25: mutual inductance normalization enhancement, specifically, performing mutual inductance elimination enhancement on the reactance elimination enhancement electromagnetic fingerprint combination data and the resistance elimination enhancement electromagnetic fingerprint combination data to obtain mutual inductance normalization enhancement electromagnetic fingerprint combination data; Step S26: electromagnetic fingerprint enhancement, specifically, performing electromagnetic fingerprint enhancement through the recognition enhancement, state differentiation enhancement, reactance combination enhancement, resistance combination enhancement and mutual inductance normalization enhancement to obtain an enhanced electromagnetic fingerprint data set; Step S3: electromagnetic fingerprint management, constructing an enhanced electromagnetic fingerprint parameter set, and performing structured storage and management of the electromagnetic fingerprint library to obtain a structured electromagnetic fingerprint library; Step S4: electromagnetic fingerprint comparison, which is used to read the electromagnetic fingerprint in the electromagnetic fingerprint library and perform device authentication matching. Specifically, based on the enhanced electromagnetic fingerprint data set, a convolutional long-term and short-term neural network combined with adaptive weighting improvement is used to perform electromagnetic fingerprint comparison to obtain electromagnetic fingerprint comparison result reference data; The electromagnetic fingerprint comparison result reference data specifically includes the comparison result type and the comparison quantitative matching degree; Step S5: Dynamic fingerprint update, using a reinforcement learning method combined with convolution feature optimization to perform dynamic fingerprint update to obtain an updated electromagnetic fingerprint management data set; Step S6: Device authentication management, obtaining electromagnetic fingerprint device authentication reference data.

2. The device authentication management method based on electromagnetic fingerprint according to claim 1 is characterized in that: In step S1, the electromagnetic fingerprint collection is used to collect the original data set required for device authentication management, which specifically includes the following steps: Step S11: starting the device, specifically starting the target device according to the operation mode of the working state, and collecting data under normal operating conditions to obtain an original motor signal data set; The target device specifically refers to a permanent magnet synchronous linear motor and a motor driver; Step S12: acquisition calibration, specifically using a signal acquisition tool, and setting the sampling rate, frequency range and sensitivity, performing acquisition calibration, and obtaining a calibration motor signal acquisition parameter set; Step S13: signal domain capture, specifically, capturing the electromagnetic radiation signal of the target device from the original motor signal data set according to the calibration motor signal acquisition parameter set, and performing domain signal capture to obtain a domain signal data set; The sub-domain signal capture specifically includes time domain signal capture and frequency domain signal capture; The frequency domain signal capture is specifically performed by using a spectrum analyzer to capture frequency distribution data to obtain frequency band signal data, wherein the frequency distribution specifically includes a low frequency band and a high frequency band; Step S14: environmental isolation, specifically, when executing the operations of step S11 and step S13, using a shielding box to isolate the external electromagnetic environment interference, and performing interference filtering operations on the collected domain signal data sets to obtain basic optimized signal data; Step S15: initial electromagnetic fingerprint construction, specifically, initializing and constructing electromagnetic fingerprint data by combining the sub-domain signal data sets to obtain an electromagnetic fingerprint original data set; The electromagnetic fingerprint original data set is used to represent the original signal characteristics, random noise characteristics and repeatable stable characteristics of the electromagnetic signal.

3. The device authentication management method based on electromagnetic fingerprint according to claim 2 is characterized in that: In step S3, the electromagnetic fingerprint management is used to construct a structured electromagnetic fingerprint library, specifically, based on the enhanced electromagnetic fingerprint data set, an enhanced electromagnetic fingerprint parameter set is constructed, and the electromagnetic fingerprint library is structuredly stored and managed to obtain a structured electromagnetic fingerprint library; In step S4, the improved convolutional long short-term neural network combined with adaptive weighting specifically includes a convolution extraction subnet, a long short-term memory subnet, an adaptive weighted difference operator and a comparison classifier; The convolutional extraction subnet is used to extract spatial features in the enhanced electromagnetic fingerprint dataset; The long short-term memory subnet is used to extract the time dimension features in the enhanced electromagnetic fingerprint dataset; The adaptive weighted difference operator is used to adaptively weight the differences between different features; The comparison classifier is used to perform electromagnetic fingerprint comparison and classification in combination with the result of adaptive weighting.

4. The device authentication management method based on electromagnetic fingerprint according to claim 3 is characterized in that: In step S4, the step of using the convolutional long-term and short-term neural network improved by combining adaptive weighting to perform electromagnetic fingerprint comparison and obtain electromagnetic fingerprint comparison result reference data includes: Step S41: constructing a convolution extraction subnet, specifically constructing a standard convolution extraction subnet including a convolution layer and a pooling layer to obtain electromagnetic fingerprint convolution feature data; Step S42: constructing a long short-term memory subnet, specifically constructing a standard long short-term memory neural network as a long short-term memory subnet, and extracting the time dimension features in the enhanced electromagnetic fingerprint data set to obtain electromagnetic fingerprint time series feature data; Step S43: construct an adaptive weighted difference operator, specifically, perform feature fusion on the electromagnetic fingerprint convolution feature data and the electromagnetic fingerprint time series feature data, and perform feature weighting by constructing an adaptive weighted difference operator to obtain adaptive feature weight data. The calculation formula is: Where Fe is the fusion feature data, y cnn is the electromagnetic fingerprint convolution feature data, y lstm is the electromagnetic fingerprint time series feature data, [·;·] is the feature concatenation operator, AWD(·) is the adaptive weighted difference operator output, used to represent the adaptive weighted difference value of the feature, Fe1 is the first electromagnetic fingerprint weighted feature, Fe2 is the second electromagnetic fingerprint weighted feature, I is the total number of features, i is the feature index, w i is the weighting factor; Step S44: constructing a comparison classifier, specifically selecting a standard support vector machine model as a classifier, and determining whether two electromagnetic fingerprints match by comparison and classification, and obtaining comparison classification data; Step S45: electromagnetic fingerprint comparison model training, specifically, performing electromagnetic fingerprint comparison model training through the convolution extraction subnet, the long short-term memory subnet, the adaptive weighted difference operator and the comparison classifier to obtain the electromagnetic fingerprint comparison model Model LCA ; Step S46: electromagnetic fingerprint comparison, specifically using the electromagnetic fingerprint comparison model Model LCA , performing electromagnetic fingerprint comparison according to the enhanced electromagnetic fingerprint data set to obtain electromagnetic fingerprint comparison result reference data; The electromagnetic fingerprint comparison result reference data specifically includes the comparison result type and the comparison quantitative matching degree.

5. The device authentication management method based on electromagnetic fingerprint according to claim 4 is characterized in that: In step S5, the dynamic fingerprint update is used to dynamically update the electromagnetic fingerprint library for optimization management. Specifically, according to the enhanced electromagnetic fingerprint parameter set, a reinforcement learning method combined with convolution feature optimization is used to perform dynamic fingerprint update to obtain an updated electromagnetic fingerprint management data set, including the following steps: Step S51: initializing the environment state, specifically initializing the structured electromagnetic fingerprint library and the enhanced electromagnetic fingerprint parameter set to obtain an initialized electromagnetic fingerprint update parameter set; Step S52: state feature convolution modeling, specifically constructing a standard convolutional neural network including a convolution layer, a pooling layer and a fully connected layer, extracting features from the structured electromagnetic fingerprint library and the initialized electromagnetic fingerprint update parameter set, and obtaining initialized electromagnetic fingerprint state feature vector data; Step S53: Initializing the reinforcement learning model, specifically, constructing an electromagnetic fingerprint library interactive update reinforcement learning initial model based on the initialized electromagnetic fingerprint state feature vector data, specifically including state parameters, action parameters, reward parameters and objective function; The state parameter is used to indicate the electromagnetic fingerprint state of the device, specifically the current value of the electromagnetic fingerprint parameter; The action parameter is used to indicate the update action of the battery fingerprint library, specifically including the update fingerprint operation, the delete fingerprint operation and the add new fingerprint operation; The update fingerprint operation is used to update the fingerprint library by changing the existing electromagnetic fingerprint data to improve the comparison accuracy; the delete fingerprint operation is used to delete the electromagnetic fingerprint that has not been updated for a long time or has a small contribution to the comparison result; the add new fingerprint operation is used to update the fingerprint library by adding new electromagnetic fingerprint data to improve the comparison accuracy; The reward parameter is used to calculate the reward according to the comparison accuracy of the updated electromagnetic fingerprint library; The objective function is used to calculate the maximum accumulated reward parameter value; The electromagnetic fingerprint library interactively updates the initial reinforcement learning model, which is used to represent the parameter combination for executing reinforcement learning model training. The calculation formula is: RL={s t ,a t ,r t ,R}; Where RL is the initial reinforcement learning model for interactive updating of the electromagnetic fingerprint library, s t is the state parameter, a t is the action parameter, r t is the reward parameter, R is the maximum accumulated reward parameter value; Step S54: construct a dynamic fingerprint update strategy, specifically, construct a state update output based on the state transfer function, and perform deep Q value learning based on the state update output, and perform management strategy update of the structured electromagnetic fingerprint library and update parameter optimization of the electromagnetic fingerprint update parameter set by combining the state feature convolution modeling and the reinforcement learning model initialization, and obtain the dynamic fingerprint update model Model by performing iterative training. CR ; Step S55: Dynamic fingerprint adaptive update optimization, specifically using the dynamic fingerprint to update the model Model CR , performing dynamic fingerprint update according to the enhanced electromagnetic fingerprint parameter set to obtain an updated electromagnetic fingerprint management data set; The updated electromagnetic fingerprint management data set specifically includes structured electromagnetic fingerprint library management reference data and electromagnetic fingerprint update parameter set data; The structured electromagnetic fingerprint library manages reference data, specifically including updating, deleting or optimizing the data in the electromagnetic fingerprint library, and realizing dynamic update management of the structured electromagnetic fingerprint library by managing reference data through the structured electromagnetic fingerprint library; The electromagnetic fingerprint update parameter set data specifically refers to the electromagnetic fingerprint parameter set after optimizing the enhanced electromagnetic fingerprint parameter set. The dynamic update of the electromagnetic fingerprint is achieved based on the optimized electromagnetic fingerprint parameter set.

6. The device authentication management method based on electromagnetic fingerprint according to claim 5 is characterized in that: In step S6, the device authentication management is used to perform comprehensive device authentication management in combination with the electromagnetic fingerprint library and the device matching result, specifically, by performing the electromagnetic fingerprint management and the electromagnetic fingerprint comparison according to the enhanced electromagnetic fingerprint data set, and by performing dynamic fingerprint update according to the updated electromagnetic fingerprint management data set, and continuing to perform comprehensive device authentication management according to the result of the dynamic fingerprint update, to obtain electromagnetic fingerprint device authentication reference data; The electromagnetic fingerprint device authentication reference data specifically includes electromagnetic fingerprint comparison and matching reference data and electromagnetic fingerprint update management reference data.

7. A device authentication management system based on electromagnetic fingerprint, used to implement the device authentication management method based on electromagnetic fingerprint as described in any one of claims 1 to 6, characterized in that: It includes an acquisition and processing module, a database building management module, a comparison and authentication module, a dynamic update module and an authentication management module.

8. The device authentication management system based on electromagnetic fingerprint according to claim 7 is characterized in that: The acquisition and processing module is used for electromagnetic fingerprint acquisition and data enhancement, obtains an enhanced electromagnetic fingerprint data set through electromagnetic fingerprint acquisition and data enhancement, and sends the enhanced electromagnetic fingerprint data set to the database construction management module, the comparison and authentication module and the dynamic update module; The library building management module is used for electromagnetic fingerprint management, and obtains an enhanced electromagnetic fingerprint parameter set and a structured electromagnetic fingerprint library through electromagnetic fingerprint management, and sends the enhanced electromagnetic fingerprint parameter set to the dynamic update module, and performs dynamic fingerprint update according to the structured electromagnetic fingerprint library; The comparison and authentication module is used for electromagnetic fingerprint comparison, obtains electromagnetic fingerprint comparison result reference data through electromagnetic fingerprint comparison, and sends the electromagnetic fingerprint comparison result reference data to the authentication management module; The dynamic update module is used for dynamic fingerprint update, obtains an updated electromagnetic fingerprint management data set through dynamic fingerprint update, and sends the updated electromagnetic fingerprint management data set to the authentication management module; The authentication management module is used for device authentication management, and obtains electromagnetic fingerprint device authentication reference data through device authentication management.

Images