A zhihui che lian key management system

Abstract

The application belongs to the field of intelligent key management, and particularly discloses a kind of wisdom converges car key management system, system includes: user terminal, intelligent key cabinet and examination unit. The scheme is submitted by user terminal online Car information, according to preset rule, automatically complete examination and approval, shorten examination and approval time;Multi-layer protection system is constructed, password is handled by hash when user registers, avoid plaintext password leakage, intrusion detection model is constructed by using long short term memory network, optimized by stochastic gradient descent method, user information database data is iteratively encrypted by using Blowfish encryption algorithm, to ensure the security of data access;Adopt double authentication method, combined with the uniqueness and dynamics of biological characteristics, effectively prevent identity fraud and forgery, improve the security and reliability of key management system.

Description

Technical Field

[0001] This invention relates to the field of smart key management, specifically to a smart vehicle key management system. Background Technology

[0002] Intelligent vehicle key management is an intelligent management system for vehicle keys. With the increasing number of vehicles and the growing demands for vehicle management, traditional manual key management methods are no longer sufficient for modern enterprises. Traditional key management involves cumbersome approval processes that rely on manual intervention, resulting in long approval cycles and susceptibility to errors or delays due to human error. Traditional key management systems often employ simple encryption methods or lack dynamic protection mechanisms, making them vulnerable to increasingly sophisticated cyberattacks, allowing sensitive information to be easily stolen and posing significant security risks during data storage and transmission. Traditional identity authentication often relies on single methods such as passwords or card swiping. Passwords can be forgotten, leaked, or cracked, while card swiping carries risks of lost or copied cards. Facial recognition accuracy drops significantly in low light conditions or when the user's posture changes, leading to misidentification or rejection, failing to meet the high reliability and security requirements of modern enterprises for identity authentication. Summary of the Invention

[0003] To address the aforementioned issues and overcome the shortcomings of existing technologies, this invention provides a smart vehicle key management system. Addressing the problems of cumbersome and manual approval processes in traditional key management, leading to long approval cycles and susceptibility to errors or delays due to human error, this solution allows users to submit vehicle usage information online via a user terminal. The system automatically completes the approval process according to preset rules and generates an approval form that is synchronized to the smart key cabinet. This automated process breaks down information barriers between departments, eliminates manual transmission and verification, shortens approval time, and significantly improves the efficiency of multi-department collaboration. Furthermore, addressing the issues of traditional key management systems often employing simple encryption methods or lacking dynamic protection mechanisms, making them vulnerable to increasingly sophisticated network attacks and posing significant security risks during data storage and transmission, this solution constructs a multi-layered protection system. It hashes passwords during user registration to prevent plaintext password leakage, utilizes a long short-term memory network to build an intrusion detection model, and optimizes the model parameters using stochastic gradient descent. Upon detecting abnormal access, an immediate warning is issued. Simultaneously, the Blowfish encryption algorithm is used to iteratively encrypt user information database data, ensuring data access security, effectively resisting external intrusion and unauthorized internal access, and protecting the confidentiality and integrity of user data. Addressing the issue that traditional identity authentication often relies on single password or card swipe methods, while facial authentication suffers significant accuracy drops in low light conditions and user posture changes, easily leading to false positives or rejections, this solution employs a dual authentication method. It collects the time-series of user online signatures and multi-pose facial images, uses linear discriminant analysis to reduce the dimensionality of features, constructs intra-class and inter-class scatter matrices, and uses exponential discriminant analysis to find the optimal projection matrix, converting the feature data into a one-dimensional fused feature vector. This combines the uniqueness and dynamism of biometrics, enabling accurate user identification through multi-dimensional feature matching even in complex environments, effectively preventing identity theft and forgery, and improving the security and reliability of the key management system.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a smart vehicle key management system, specifically including a user terminal, a smart key cabinet and an approval unit;

[0005] The user fills in the vehicle usage information on the user terminal, which includes the user's name, user ID, usage time, usage location and vehicle type, and submits an application. The approval unit automatically approves the application. If the approval is successful, an approval form is generated and sent to the smart key cabinet.

[0006] The smart key cabinet includes a data encryption module and a linkage authentication module. The data encryption module includes a cloud server, a user registration unit, an intrusion detection unit, and a data access control unit. The linkage authentication module extracts the user's online signature and facial features, and uses a dual authentication method to identify approved users.

[0007] In the user registration unit, the user sends a registration request to the cloud server. The registration request includes the user's employee ID and user terminal password. After receiving the registration request, the cloud server performs hash processing on the password and generates a hash password, which is stored in the cloud server's user information database.

[0008] The intrusion detection unit establishes and initializes a long short-term memory network as an intrusion detection model, randomly generates model parameter vectors, and optimizes the model parameters in the long short-term memory network using stochastic gradient descent, as shown in the following formula:

[0009] ;

[0010] In the formula, It is the first The model parameter vector in the short-term memory network during the next iteration. It is the first The model parameter vector in the short-term memory network during the next iteration. It's the learning rate. It is the objective function The gradient;

[0011] The data access control unit collects user access data and inputs it into the intrusion detection model for training. When abnormal access is detected, the intrusion detection unit issues an abnormal warning.

[0012] The Blowfish encryption algorithm is used to iteratively encrypt the data in the user information database on the cloud server. The formula used is as follows:

[0013]

[0014]

[0015] In the formula, and These are the left and right halves of the 64-bit plaintext after it has been grouped. It is the number of iterations. It is a subkey generated during the key expansion stage of the Blowfish encryption algorithm. It is a bitwise XOR operation. This represents the commutation operation. It is a nonlinear transformation function. , , and It is the S-box generated by the Blowfish encryption algorithm during the key expansion stage. , , and It is The index of each group after splitting;

[0016] When a user accesses the user information database, a login request containing the user's employee ID and the user's terminal secret is sent. The cloud server verifies the hash password. After successful verification, a one-time hash digest is generated, and data access permissions are dynamically authorized using the Blowfish encryption algorithm.

[0017] The linked authentication module uses a dual authentication method to identify users, which specifically includes the following steps:

[0018] Step S1: Data acquisition. Collect user online signatures and multi-pose face images. The online signatures are obtained through a digital tablet and stored in a time sequence format, including coordinates, pressure value, azimuth angle, altitude, and timestamp. The multi-pose face images are obtained through a high-resolution camera, including frontal face, left-angled 15° side face, right-angled 15° side face, and images under high and low light intensity. The multi-pose face images are uniformly scaled and grayscale normalized.

[0019] Step S2: Feature extraction, extracting signature dynamic features and face features respectively. The signature dynamic features include displacement, velocity and acceleration. The face features include face texture features and face gradient features. The face features are obtained through context-aware methods.

[0020] Step S3: Feature dimensionality reduction. Linear discriminant analysis (LDA) is used to reduce the dimensionality of the signature and face features, constructing a scatter matrix. The scatter matrix includes within-class scatter matrix and between-class scatter matrix, using the following formula:

[0021] ;

[0022] ;

[0023] In the formula, It is the within-class scatter matrix. It is the inter-class scatter matrix. It is the feature vector of the same type of sample. It is the feature vector of the outlier sample. and These are the sets of similar sample pairs and the sets of dissimilar sample pairs. Indicates the transpose operation;

[0024] Step S4: Using the exponential discriminant analysis method, the feature data is transformed into the projection space through the projection matrix, and the optimal projection matrix is ​​found to obtain a one-dimensional fused feature vector. The formula used is as follows:

[0025] ;

[0026] In the formula, It is the optimal projection matrix. It is a projection matrix. It is an element-level arctangent. Indicates the transpose operation;

[0027] Step S5: Identity authentication. Establish and initialize a deep neural network model. The deep neural network model uses ReLU as the activation function. Input a one-dimensional fused feature vector into the deep neural network model for training. The deep neural network model outputs online signature and face authentication results.

[0028] Furthermore, in step S2, facial features are obtained using a context-aware method, specifically including the following steps:

[0029] Step S21: Connect the center points of both eyes and calculate the rotation angle, which is the angle between the line connecting the centers of both eyes and the horizontal axis. Correct the face image through rotation transformation. The formula used is as follows:

[0030] ;

[0031] ;

[0032] In the formula, and These are the first two parts of the rotated face image. The horizontal and vertical coordinates of each pixel and These are the first two images of the face before rotation. The horizontal and vertical coordinates of each pixel and These are the horizontal and vertical coordinates of the rotation center, which is set to the center of the face image.

[0033] Step S22: Vertically divide the facial image into three parts: upper, middle and lower. Select the points corresponding to the maximum and minimum vertical coordinates of each part to generate a hexagonal region. Keep the core facial region and remove irrelevant regions at the edges of the hexagon to obtain the region of interest.

[0034] Step S23: Extract facial texture features from the region of interest using a local binary histogram. The formula used is as follows:

[0035] ;

[0036] In the formula, It is a local binary pattern value. It is a threshold comparison function. It is the number of neighboring pixels. It is the first The grayscale value of each neighboring pixel. It is the grayscale value of the center pixel;

[0037] Step S24: Using the gradient histogram algorithm, the face is divided into pixel blocks, each pixel block is divided into units, and 9-dimensional gradient direction features are calculated for each unit to obtain the face gradient features.

[0038] The beneficial effects achieved by adopting the above solution are as follows:

[0039] (1) In the traditional key management, the approval process is cumbersome and relies on manual labor, resulting in a long approval cycle and easy to cause approval errors or delays due to human negligence. This solution allows users to submit vehicle information online through the user terminal, automatically complete the approval according to preset rules, and generate an approval form that is synchronized to the smart key cabinet. This automated process breaks down the information barriers between departments, eliminates the manual transmission and verification process, shortens the approval time, and greatly improves the efficiency of multi-department collaboration.

[0040] (2) In view of the fact that traditional key management systems often use simple encryption methods or lack dynamic protection mechanisms, sensitive information is easily stolen in the face of increasingly complex network attack methods, and there are great security risks in the data storage and transmission process, this solution constructs a multi-layer protection system. When users register, the password is hashed to avoid plaintext password leakage. An intrusion detection model is constructed using a long short-term memory network. The model parameters are optimized by the stochastic gradient descent method. When abnormal access is detected, an immediate warning is issued. At the same time, the Blowfish encryption algorithm is used to iteratively encrypt the user information database data to ensure the security of data access, effectively resist external intrusion and internal illegal access, and protect the confidentiality and integrity of user data.

[0041] (3) In view of the fact that traditional identity authentication often uses a single password or card swipe method, while the accuracy of face authentication drops significantly in scenarios such as insufficient light and changes in user posture, which can easily lead to misidentification or rejection, this solution adopts a dual authentication method. It collects the time sequence of the user's online signature and multi-pose face images, uses the linear discriminant analysis algorithm to reduce the dimensionality of the features, constructs intra-class and inter-class scatter matrices, and uses the exponential discriminant analysis method to find the optimal projection matrix, converting the feature data into a one-dimensional fusion feature vector. This combines the uniqueness and dynamism of biometric features, and can accurately identify the user's identity through multi-dimensional feature matching even in complex environments, effectively preventing identity theft and forgery, and improving the security and reliability of the key management system. Attached Figure Description

[0042] Figure 1 This invention provides a module connection diagram for a smart vehicle key management system.

[0043] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0044] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0045] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0046] Example 1, see Figure 1 The present invention provides a smart vehicle key management system, which specifically includes a user terminal, a smart key cabinet and an approval unit;

[0047] The user fills in the vehicle usage information on the user terminal, which includes the user's name, user ID, usage time, usage location and vehicle type, and submits an application. The approval unit automatically approves the application. If the approval is successful, an approval form is generated and sent to the smart key cabinet.

[0048] The smart key cabinet includes a data encryption module and a linkage authentication module. The data encryption module includes a cloud server, a user registration unit, an intrusion detection unit, and a data access control unit. The linkage authentication module extracts the user's online signature and facial features, and uses a dual authentication method to identify approved users.

[0049] Example 2, see Figure 1 This embodiment is based on the above embodiment. In the user registration unit, the user sends a registration request to the cloud server. The registration request includes the user's employee number and the user terminal password. After receiving the registration request, the cloud server performs hash processing on the password and generates a hash password, which is stored in the user information database of the cloud server.

[0050] The intrusion detection unit establishes and initializes a long short-term memory network as an intrusion detection model, randomly generates model parameter vectors, and optimizes the model parameters in the long short-term memory network using stochastic gradient descent, as shown in the following formula:

[0051] ;

[0052] In the formula, It is the first The model parameter vector in the short-term memory network during the next iteration. It is the first The model parameter vector in the short-term memory network during the next iteration. It's the learning rate. It is the objective function The gradient;

[0053] The data access control unit collects user access data and inputs it into the intrusion detection model for training. When abnormal access is detected, the intrusion detection unit issues an abnormal warning.

[0054] The Blowfish encryption algorithm is used to iteratively encrypt the data in the user information database on the cloud server. The formula used is as follows:

[0055]

[0056]

[0057] In the formula, and These are the left and right halves of the 64-bit plaintext after it has been grouped. It is the number of iterations. It is a subkey generated during the key expansion stage of the Blowfish encryption algorithm. It is a bitwise XOR operation. This represents the commutation operation. It is a nonlinear transformation function. , , and It is the S-box generated by the Blowfish encryption algorithm during the key expansion stage. , , and It is The index of each group after splitting;

[0058] When a user accesses the user information database, a login request containing the user's employee ID and the user's terminal secret is sent. The cloud server verifies the hash password. After successful verification, a one-time hash digest is generated, and data access permissions are dynamically authorized using the Blowfish encryption algorithm.

[0059] Example 3, see Figure 1 This embodiment is based on the above embodiment. The linked authentication module uses a dual authentication method to identify the user. The dual authentication method specifically includes the following steps:

[0060] Step S1: Data acquisition. Collect user online signatures and multi-pose face images. The online signatures are obtained through a digital tablet and stored in a time sequence format, including coordinates, pressure value, azimuth angle, altitude, and timestamp. The multi-pose face images are obtained through a high-resolution camera, including frontal face, left-angled 15° side face, right-angled 15° side face, and images under high and low light intensity. The multi-pose face images are uniformly scaled and grayscale normalized.

[0061] Step S2: Feature extraction, extracting signature dynamic features and face features respectively. The signature dynamic features include displacement, velocity and acceleration. The face features include face texture features and face gradient features. The face features are obtained through context-aware methods.

[0062] Step S3: Feature dimensionality reduction. Linear discriminant analysis (LDA) is used to reduce the dimensionality of the signature and face features, constructing a scatter matrix. The scatter matrix includes within-class scatter matrix and between-class scatter matrix, using the following formula:

[0063] ;

[0064] ;

[0065] In the formula, It is the within-class scatter matrix. It is the inter-class scatter matrix. It is the feature vector of the same type of sample. It is the feature vector of the outlier sample. and These are the sets of similar sample pairs and the sets of dissimilar sample pairs. Indicates the transpose operation;

[0066] Step S4: Using the exponential discriminant analysis method, the feature data is transformed into the projection space through the projection matrix, and the optimal projection matrix is ​​found to obtain a one-dimensional fused feature vector. The formula used is as follows:

[0067] ;

[0068] In the formula, It is the optimal projection matrix. It is a projection matrix. It is an element-level arctangent. Indicates the transpose operation;

[0069] Step S5: Identity authentication. Establish and initialize a deep neural network model. The deep neural network model uses ReLU as the activation function. Input a one-dimensional fused feature vector into the deep neural network model for training. The deep neural network model outputs online signature and face authentication results.

[0070] Example 4, see Figure 1 This embodiment is based on the above embodiment. In step S2, facial features are obtained through a context-aware method, specifically including the following steps:

[0071] Step S21: Connect the center points of both eyes and calculate the rotation angle, which is the angle between the line connecting the centers of both eyes and the horizontal axis. Correct the face image through rotation transformation. The formula used is as follows:

[0072] ;

[0073] ;

[0074] In the formula, and These are the first two parts of the rotated face image. The horizontal and vertical coordinates of each pixel and These are the first two images of the face before rotation. The horizontal and vertical coordinates of each pixel and These are the horizontal and vertical coordinates of the rotation center, which is set to the center of the face image.

[0075] Step S22: Vertically divide the facial image into three parts: upper, middle and lower. Select the points corresponding to the maximum and minimum vertical coordinates of each part to generate a hexagonal region. Keep the core facial region and remove irrelevant regions at the edges of the hexagon to obtain the region of interest.

[0076] Step S23: Extract facial texture features from the region of interest using a local binary histogram. The formula used is as follows:

[0077] ;

[0078] In the formula, It is a local binary pattern value. It is a threshold comparison function. It is the number of neighboring pixels. It is the first The grayscale value of each neighboring pixel. It is the grayscale value of the center pixel;

[0079] Step S24: Using the gradient histogram algorithm, the face is divided into pixel blocks, each pixel block is divided into units, and 9-dimensional gradient direction features are calculated for each unit to obtain the face gradient features.

[0080] Example 5 is based on the above examples. In Example 4, the facial image is vertically divided into three parts: upper, middle and lower. The upper part is the upper 1 / 4 area of ​​the facial image, the middle part is the middle 1 / 2 area of ​​the facial image, and the lower part is the bottom 1 / 4 area of ​​the facial image.

[0081] Example 6 is based on the above examples. In Example 4, a deep neural network model is established and initialized. The deep neural network model uses ReLU as the activation function. A one-dimensional fused feature vector is input into the deep neural network model for training. The batch size during training is 300, the maximum number of epochs is 800, the gradient threshold is set to 1, Adam is used as the optimizer, and cross-entropy loss is used as the loss function.

[0082] Example 7 is based on the above examples. In Example 2, the objective function... This is the cross-entropy loss function for binary classification.

[0083] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0084] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0085] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A smart vehicle key management system, characterized in that, Includes user terminals, smart key cabinets, and approval units; The user fills in the vehicle usage information on the user terminal and submits the application. The approval unit automatically approves the application. If the approval is successful, an approval form is generated and sent to the smart key cabinet. The smart key cabinet includes a data encryption module and a linkage authentication module. The data encryption module includes a cloud server, a user registration unit, an intrusion detection unit, and a data access control unit. The linkage authentication module extracts the user's online signature and facial features, and uses a dual authentication method to identify approved users. In the user registration unit, the user sends a registration request to the cloud server. The registration request includes the user's employee ID and user terminal password. After receiving the registration request, the cloud server performs hash processing on the password and generates a hash password, which is stored in the cloud server's user information database. The intrusion detection unit establishes and initializes a long short-term memory network as an intrusion detection model, randomly generates a model parameter vector, and optimizes the model parameters in the long short-term memory network using the stochastic gradient descent method. The data access control unit collects user access data and inputs it into the intrusion detection model for training. When abnormal access is detected, the intrusion detection unit issues an abnormal warning. The Blowfish encryption algorithm is used to iteratively encrypt the data in the user information database on the cloud server. When a user accesses the user information database, a login request containing the user's employee ID and the user's terminal secret is sent. The cloud server verifies the hash password. After successful verification, a one-time hash digest is generated, and data access permissions are dynamically authorized using the Blowfish encryption algorithm.

2. The intelligent vehicle key management system according to claim 1, characterized in that, The linked authentication module uses a dual authentication method to identify users, which specifically includes the following steps: Step S1: Data acquisition, acquiring user online signatures and multi-pose face images. The online signatures are obtained through a digital tablet and stored in a time sequence format. The multi-pose face images are uniformly scaled and grayscale normalized. Step S2: Feature extraction, extracting signature dynamic features and face features respectively. The signature dynamic features include displacement, velocity and acceleration. The face features include face texture features and face gradient features. The face features are obtained through context-aware methods. Step S3: Feature dimensionality reduction. Use the linear discriminant analysis algorithm to reduce the dimensionality of the signature features and face features, and construct the divergence matrix. Step S4: Using the exponential discriminant analysis method, the feature data is transformed into the projection space through the projection matrix, the optimal projection matrix is ​​found, and a one-dimensional fused feature vector is obtained; Step S5: Identity authentication. Establish and initialize a deep neural network model. The deep neural network model uses ReLU as the activation function. Input a one-dimensional fused feature vector into the deep neural network model for training. The deep neural network model outputs online signature and face authentication results.

3. The intelligent vehicle key management system according to claim 2, characterized in that, In step S2, facial features are obtained using a context-aware method, specifically including the following steps: Step S21: Connect the center points of both eyes and calculate the rotation angle, which is the angle between the line connecting the centers of both eyes and the horizontal axis. Correct the face image through rotation transformation. Step S22: Vertically divide the facial image into three parts: upper, middle and lower. Select the points corresponding to the maximum and minimum vertical coordinates of each part to generate a hexagonal region. Keep the core facial region and remove irrelevant regions at the edges of the hexagon to obtain the region of interest. Step S23: Extract facial texture features from the region of interest using a local binary histogram; Step S24: Using the gradient histogram algorithm, the face is divided into pixel blocks, each pixel block is divided into units, and 9-dimensional gradient direction features are calculated for each unit to obtain the face gradient features.

Images