Abnormal access behavior detection method, device, storage medium and electronic device

By acquiring and training a security protection model based on historical traffic data, the problem of inaccurate abnormal access behavior detection in the prior art is solved, and the accurate abnormal access behavior judgment of website server traffic data is realized.

CN113949525BInactive Publication Date: 2024-11-12ZHONGYUNWANGAN TECHNOLOGY (BEIJING) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111044460.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-09-07
Publication Date
2024-11-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing abnormal access behavior detection methods are not accurate enough, making it difficult to effectively identify the attack behavior of the website server.

Method used

By obtaining the historical traffic data of the target application, performing feature extraction and security protection model training, and generating a security protection model corresponding to the target application, which is used to judge the abnormal access behavior in the current traffic data.

Benefits of technology

The accuracy of abnormal access behavior detection is improved, and abnormal access behavior in traffic data accessed to the target application can be more accurately identified and judged.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113949525B_ABST
    Figure CN113949525B_ABST
Patent Text Reader

Abstract

The present application discloses a method, device, storage medium and electronic device for detecting abnormal access behavior. The method includes: obtaining a number of target historical traffic data corresponding to a target application; performing feature extraction based on each of the target historical traffic data to obtain a number of feature information; training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application; judging abnormal access behavior of the current traffic data generated by accessing the target application based on the security protection model to obtain the detection result of abnormal access behavior. In the present application, by obtaining the target historical traffic data corresponding to the target application and then performing model training to obtain the security protection model corresponding to the target application, it is possible to more accurately judge abnormal access behavior of the traffic data accessing the target application, thereby improving the accuracy of abnormal access behavior detection.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and is characterized by a method, device, storage medium and electronic device for detecting abnormal access behavior. Background Art

[0002] With the rapid development of the Internet, the acceleration of new infrastructure construction, and the trend of digitalization sweeping the world, digital transformation has been applied in government, energy, finance, education, medical and other industries. Then, there are more and more attacks on website servers, so it is particularly important to detect abnormal access behavior in the traffic data of accessing server websites.

[0003] In the existing abnormal access behavior identification, a unified identification method is usually used to detect abnormal access behavior on the traffic data accessing each website server, and the detection result is not accurate enough. Summary of the invention

[0004] In view of this, the present invention provides a method, device, storage medium and electronic device for detecting abnormal access behavior, the main purpose of which is to solve the current problem of inaccurate detection of abnormal access behavior.

[0005] To solve the above problems, the present application provides a method for detecting abnormal access behavior, including:

[0006] Obtaining several target historical traffic data corresponding to the target application;

[0007] Extract features based on the historical traffic data of each target to obtain some feature information;

[0008] Training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0009] Based on the security protection model, abnormal access behavior is judged on the current traffic data generated by accessing the target application to obtain a detection result of the abnormal access behavior.

[0010] Optionally, the acquiring of several target historical traffic data corresponding to the target application specifically includes:

[0011] Obtain some historical traffic data corresponding to the target application;

[0012] The historical traffic data are preprocessed based on a convolutional neural network model to filter out a number of target historical traffic data from the historical traffic data; each target historical traffic data is the traffic data generated by normal access behavior.

[0013] Optionally, the feature extraction is performed based on the historical traffic data of each target to obtain some feature information, specifically including:

[0014] Based on the historical traffic data of each target, obtaining the access directory, access time, network address of the accessing party, request message and response information corresponding to the historical traffic data of each target;

[0015] Based on the access directory, access time, network address of the accessing party, request message and response information corresponding to the historical traffic data of each target, extract and obtain the first characteristic information corresponding to the underlying logic, the second characteristic information corresponding to the access logic and the third characteristic information corresponding to the business logic;

[0016] The plurality of feature information is obtained based on one or more of the first feature information, the second feature information and the third feature information.

[0017] Optionally, the extracting of first feature information corresponding to the underlying logic, second feature information corresponding to the access logic, and third feature information corresponding to the business logic based on the access directory, access time, network address of the accessing party, request message, and response information corresponding to each target historical traffic data specifically includes:

[0018] Extracting and obtaining directory structure features corresponding to the target application based on the access directory to obtain the first feature information;

[0019] Extracting and obtaining time features, network address features and request features corresponding to the target application based on the access time, the network address of the accessing party and the request message, and obtaining the second feature information based on one or more of the time features, network address features and request features;

[0020] Business feature extraction is performed based on the response information to obtain response features corresponding to the target application, so as to obtain the third feature information.

[0021] Optionally, the feature extraction is performed based on the historical traffic data of each target to obtain some feature information, specifically including:

[0022] Extract features based on the access type of historical traffic data of each target, and obtain a feature information set corresponding to the access type;

[0023] The security protection model is trained based on each of the characteristic information to obtain a security protection model corresponding to the target application, which specifically includes: training the model based on the characteristic information sets corresponding to each access type to obtain a security protection sub-model, so as to obtain the security protection model based on each of the security protection sub-models.

[0024] Optionally, the acquiring of some historical traffic data corresponding to the target application specifically includes:

[0025] Obtain historical traffic data corresponding to the target application based on the bypass reverse proxy method.

[0026] Optionally, when the detection result of abnormal access behavior is obtained, the method further includes: prompting the abnormal access behavior in a predetermined prompting manner.

[0027] In order to solve the above problems, the present application provides a detection device for abnormal access behavior, including:

[0028] An acquisition module, used to acquire a number of target historical traffic data corresponding to a target application;

[0029] An extraction module, used for extracting features based on the historical traffic data of each target to obtain a number of feature information;

[0030] A training module, used to train a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0031] The detection module is used to judge abnormal access behavior of current traffic data generated by accessing the target application based on the security protection model, and obtain the detection result of the abnormal access behavior.

[0032] To solve the above problems, the present application provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any of the above-mentioned methods for detecting abnormal access behaviors are implemented.

[0033] To solve the above problems, the present application provides an electronic device, comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any of the above-mentioned methods for detecting abnormal access behaviors when executing the computer program on the memory.

[0034] In the present application, by acquiring the target historical traffic data corresponding to the target application and then performing model training to obtain a security protection model corresponding to the target application, it is possible to more accurately judge abnormal access behavior of the traffic data accessing the target application, thereby improving the accuracy of detecting abnormal access behavior.

[0035] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0037] Figure 1 This is a flow chart of a method for detecting abnormal access behavior according to an embodiment of the present application;

[0038] Figure 2 A flowchart of detecting abnormal access behavior according to another embodiment of the present application;

[0039] Figure 3 This is a structural block diagram of a device for detecting abnormal access behavior according to another embodiment of the present application. DETAILED DESCRIPTION

[0040] Various aspects and features of the present application are described herein with reference to the accompanying drawings.

[0041] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present application will occur to those skilled in the art.

[0042] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0043] These and other characteristics of the present application will become apparent from the following description of a preferred form of embodiment given as a non-limiting example with reference to the accompanying drawings.

[0044] It should also be understood that although the present application has been described with reference to some specific examples, those skilled in the art will be able to readily implement many other equivalent forms of the present application.

[0045] The above and other aspects, features and advantages of the present application will become more apparent in view of the following detailed description when taken in conjunction with the accompanying drawings.

[0046] Specific embodiments of the present application are described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments applied for are merely examples of the present application, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that obscure the present application. Therefore, the specific structural and functional details applied for herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present application in a variety of ways with substantially any suitable detailed structure.

[0047] This specification may use the phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," all of which may refer to one or more of the same or different embodiments according to the present application.

[0048] The embodiment of the present application provides a method for detecting abnormal access behavior, which can be specifically applied to a server or platform device, where various applications or services are deployed for access by terminal devices. Figure 1 As shown, the method in this embodiment includes the following steps:

[0049] Step S101, obtaining a number of target historical traffic data corresponding to a target application;

[0050] In this step, the target traffic data is the traffic data generated by the normal access behavior to the target application. The normal access behavior refers to non-attack behavior. When obtaining the target historical traffic data, a bypass reverse proxy method can be used to obtain a number of historical traffic data, and then the traffic data generated by the normal access behavior of the predetermined protocol type is screened from the several historical traffic data to be used as the target historical traffic data, wherein the predetermined protocol type includes any one of the following: Hyper Text Transfer Protocol over Secure Socket Layer (HTTPS) and Hyper Text Transfer Protocol (HTTP).

[0051] Step S102, extracting features based on the historical traffic data of each target to obtain some feature information;

[0052] When extracting features in this step, supervised learning and unsupervised learning can be used to extract features from the three levels of underlying logic, access logic and business logic to obtain some feature information. The specific process of extracting feature information using supervised learning is: first determine whether the first information in each sample flow data is feature information or non-feature information, and mark it to obtain a training data set, train and obtain an extraction model based on the marking results of each first information in the training data set, and finally use the extraction model to identify each first feature information in the target historical flow data, thereby obtaining some feature information. The process of extracting features using unsupervised learning is: extract each first information in the sample flow data to obtain a sample set, then classify each first information based on the similarity of each first information to obtain a clustering result, obtain a classification model based on the clustering result, and then classify each first information of each target historical flow data according to the clustering model to obtain a classification result, that is, classify each first information into a feature information category or a non-feature information category. Finally, combine the feature extraction results of the supervised learning method and the classification results of the unsupervised learning method to obtain the final feature information.

[0053] Furthermore, the first information may include any one or more of the following: access time, network address (IP address), field information in the http header, uniform resource identifier (URI for short), Payload data information (actual valid data), non-valid data information, etc. Among them, the field information in the http header specifically includes any one or more of the following: field type, value composition, and distribution of value content. Payload data information specifically includes any one or more of the following: data type, data, and character distribution.

[0054] Step S103, training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0055] Step S104: Based on the security protection model, abnormal access behavior is determined for the current traffic data generated by accessing the target application to obtain a detection result of the abnormal access behavior.

[0056] In this application, by obtaining the target historical traffic data corresponding to the target application, and then performing model training to obtain the security protection model corresponding to the target application, it is possible to more accurately judge the abnormal access behavior of the traffic data accessing the target application, thereby improving the accuracy of detecting abnormal access behavior.

[0057] On the basis of the above embodiments, in order to more accurately detect abnormal access behavior, another embodiment of the present application provides a method for detecting abnormal access behavior, such as Figure 2 As shown, including:

[0058] Step S201, obtaining a number of historical traffic data corresponding to the target application; preprocessing each of the historical traffic data based on a convolutional neural network model to filter and obtain a number of target historical traffic data from the several historical traffic data; each of the target historical traffic data is traffic data generated by normal access behavior;

[0059] Step S202, based on the historical traffic data of each target, obtaining the access directory, access time, network address of the access party, request message and response information corresponding to the historical traffic data of each target; based on the access directory, access time, network address of the access party, request message and response information corresponding to the historical traffic data of each target, extracting the first feature information corresponding to the underlying logic, the second feature information corresponding to the access logic and the third feature information corresponding to the business logic; obtaining the plurality of feature information based on one or more of the first feature information, the second feature information and the third feature information;

[0060] In the specific implementation process of this step, extracting the first characteristic information includes: extracting the directory structure characteristics corresponding to the target application based on the access directory to obtain the first characteristic information; extracting the second characteristic information includes: extracting the time characteristics, network address characteristics and request characteristics corresponding to the target application based on the access time, the network address of the accessing party and the request message, and obtaining the second characteristic information based on one or more of the time characteristics, network address characteristics and request characteristics; extracting the third characteristic information includes: extracting the business characteristics based on the response information to obtain the response characteristics corresponding to the target application to obtain the third characteristic information.

[0061] Step S203, training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0062] After obtaining the security protection model in this step, the identification information of the security access model can also be configured according to the attribute information of the target application, and a mapping relationship between the attribute information and the identification information can be established. The attribute information can specifically be the name of the target application, etc. In this way, when the target application is accessed to generate current traffic data, the corresponding security protection model can be accurately and quickly found according to the name of the target application in the current traffic data, which provides a guarantee for the subsequent use of the security protection model to judge abnormal access behavior of the current traffic data.

[0063] Step S204: Based on the security protection model, abnormal access behavior is determined for the current traffic data generated by accessing the target application to obtain a detection result of the abnormal access behavior.

[0064] In this embodiment, when extracting features based on the target historical traffic data to obtain several feature information, it is also possible to specifically combine the access type of the target historical traffic data to obtain several feature information corresponding to the access type, so that in the subsequent security protection model, the model training can be performed separately for the feature information corresponding to each access type to obtain the security protection sub-model corresponding to the access type, and each security protection sub-model together constitutes the security protection model. At the same time, in this embodiment, each security protection model can be configured with a unique first identification information; each security protection sub-model can be further configured with a unique second identification information, and then a first mapping relationship between the first identification information and the application name is established, and a second mapping relationship between the second identification information and the access type is established. In this way, when the traffic data is generated by accessing the application later, the first identification information can be determined by searching the first mapping relationship for the application name of the target application in the traffic data, and then the target security protection model can be quickly found according to the first identification information, and then the corresponding second identification information can be further found according to the access type in the traffic data. The second mapping relationship is determined, and the target security protection sub-model is quickly found from the target security protection model through the second identification information, and finally the target security protection sub-model is used to judge the abnormal behavior of the corresponding traffic data. This embodiment generates / trains security protection models in a targeted manner for different applications and access types of traffic data, which can make the detection of abnormal access behaviors more accurate.

[0065] In this embodiment, after the detection result of abnormal access behavior is obtained, a prompt of abnormal access behavior can also be provided in a predetermined prompting manner. The specific prompting manner can be a voice playback manner, a screen display manner, etc., that is, when abnormal access behavior is detected, a specific melody sound, voice, etc. is output through voice playback to provide a prompt, and a predetermined symbol, text, picture, etc. is displayed through a screen display manner to provide a prompt. By providing prompt information for abnormal access behavior, the staff can more intuitively understand the abnormal access behavior.

[0066] In the present application, corresponding security protection models are established for different applications. Therefore, when traffic data is generated by subsequent access to the application, the security protection model corresponding to the application can be called to judge the traffic data to determine whether the access behavior is normal business access behavior or attack behavior / abnormal access behavior, thereby improving the accuracy of abnormal access behavior detection.

[0067] In order to further explain the above embodiment, a specific description is given below by taking the target application as an OA application as an example.

[0068] Step 1: Obtain some historical traffic data generated by accessing the OA application by bypassing the reverse proxy. Specifically, normal access to the OA application can be established by bypassing the reverse proxy to obtain the historical traffic data of accessing the OA application.

[0069] Step 2: After obtaining some historical traffic data, each historical traffic data can be processed in a small data format to extract the traffic data generated by normal access behavior of a predetermined protocol type, and use this as the target historical traffic data. For example, extract the traffic data generated by normal business access regarding the HTTPS / HTTP protocol to obtain the target historical traffic data.

[0070] Step 3: Use supervised learning and unsupervised learning methods to extract features from the target historical traffic data to obtain several access features corresponding to the OA application, that is, to obtain some feature information.

[0071] More specifically, feature information can be extracted from different dimensions such as the underlying logic, business logic, and access logic. For example, the layer can upgrade the two-dimensional access data through multiple parameters such as the access time, IP address, http header (more than 60 parameters and parameter values), URI, and Payload, and then describe the different dimensions through a matrix model to establish associations. The result of this association is the underlying logic, business logic, and access logic. The three logics have different characteristics. The underlying logic includes the directory structure. The business logic is based on the underlying logic, and it is the relevance of a request to the directory within a certain time range, and the relevance of the response data. The access logic is based on the business logic, and it is the relevance and classification of access from different sources in the business logic.

[0072] Step 4: Perform model training based on the obtained feature information, thereby obtaining a security protection model corresponding to the OA application. Before model training, the feature information can be further converted to convert the underlying features into higher-level, more abstract feature representations. For example, the obtained feature information can be converted into a word vector file, and then the model training can be performed based on the word vector file.

[0073] In this embodiment, by obtaining several target historical traffic data corresponding to the OA application, and then performing feature extraction on the several target historical traffic data, a number of feature information under normal access is obtained, and then model training is performed based on each feature information, thereby obtaining a security protection model corresponding to the OA application. In this way, when the current traffic data is generated by accessing the OA subsequently, the security protection model can be called to judge the abnormal access behavior of the current traffic data, so as to accurately detect whether the access behavior to the OA application is normal business access or attack behavior.

[0074] Another embodiment of the present application provides a detection device for abnormal access behavior, such as Figure 3 As shown, including:

[0075] Acquisition module 1, used to acquire several target historical traffic data corresponding to the target application;

[0076] Extraction module 2, used for extracting features based on the historical traffic data of each target to obtain a number of feature information;

[0077] A training module 3 is used to train a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0078] The detection module 4 is used to judge abnormal access behavior of the current traffic data generated by accessing the target application based on the security protection model, and obtain the detection result of the abnormal access behavior.

[0079] The acquisition module is specifically used to: acquire a number of historical traffic data corresponding to the target application; pre-process each of the historical traffic data based on a convolutional neural network model to filter and obtain a number of target historical traffic data from the several historical traffic data; each of the target historical traffic data is the traffic data generated by normal access behavior.

[0080] The extraction module is specifically used to: obtain the access directory, access time, network address of the access party, request message and response information corresponding to each target historical traffic data based on the historical traffic data of each target; extract the first feature information corresponding to the underlying logic, the second feature information corresponding to the access logic and the third feature information corresponding to the business logic based on the access directory, access time, network address of the access party, request message and response information corresponding to each target historical traffic data; obtain the several feature information based on one or more of the first feature information, the second feature information and the third feature information.

[0081] The extraction module is also specifically used to: extract the directory structure characteristics corresponding to the target application based on the access directory to obtain the first characteristic information; extract the time characteristics, network address characteristics and request characteristics corresponding to the target application based on the access time, the network address of the accessing party and the request message, and obtain the second characteristic information based on one or more of the time characteristics, network address characteristics and request characteristics; extract the business characteristics based on the response information to obtain the response characteristics corresponding to the target application to obtain the third characteristic information.

[0082] Specifically, the extraction module is also used to: perform feature extraction based on the access type of each target historical traffic data to obtain a feature information set corresponding to the access type; train a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application, specifically including: performing model training based on the feature information sets corresponding to each access type to obtain a security protection sub-model, so as to obtain the security protection model based on each of the security protection sub-models.

[0083] Optionally, the acquisition module is specifically used to: acquire historical traffic data corresponding to the target application based on a bypass reverse proxy method.

[0084] The detection device for abnormal access behavior also includes a prompt module, and the prompt module is used to: prompt the abnormal access behavior in a predetermined prompting manner.

[0085] In the present application, by acquiring the target historical traffic data corresponding to the target application and then performing model training to obtain a security protection model corresponding to the target application, it is possible to more accurately judge abnormal access behavior of the traffic data accessing the target application, thereby improving the accuracy of detecting abnormal access behavior.

[0086] Another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the following steps of the abnormal access behavior detection method are implemented:

[0087] Step 1: Obtain some target historical traffic data corresponding to the target application;

[0088] Step 2: extracting features based on the historical traffic data of each target to obtain some feature information;

[0089] Step 3: training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0090] Step 4: Based on the security protection model, the current traffic data generated by accessing the target application is judged for abnormal access behavior to obtain a detection result of the abnormal access behavior.

[0091] The specific implementation process of the above method steps can refer to any of the above embodiments of the method for detecting abnormal access behavior, and this embodiment will not be repeated here.

[0092] In the present application, by acquiring the target historical traffic data corresponding to the target application and then performing model training to obtain a security protection model corresponding to the target application, it is possible to more accurately judge abnormal access behavior of the traffic data accessing the target application, thereby improving the accuracy of detecting abnormal access behavior.

[0093] Another embodiment of the present application provides an electronic device, comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the following steps of a method for detecting abnormal access behavior when executing the computer program on the memory:

[0094] Step 1: Obtain some target historical traffic data corresponding to the target application;

[0095] Step 2: extracting features based on the historical traffic data of each target to obtain some feature information;

[0096] Step 3: training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application;

[0097] Step 4: Based on the security protection model, the current traffic data generated by accessing the target application is judged for abnormal access behavior to obtain the detection result of abnormal access behavior

[0098] The specific implementation process of the above method steps can be found in the above-mentioned data reconciliation and repair method embodiment, and this embodiment will not be repeated here.

[0099] In the present application, by acquiring the target historical traffic data corresponding to the target application and then performing model training to obtain a security protection model corresponding to the target application, it is possible to more accurately judge abnormal access behavior of the traffic data accessing the target application, thereby improving the accuracy of detecting abnormal access behavior.

[0100] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. A method for detecting abnormal access behavior, characterized in that: include: Obtaining several target historical traffic data corresponding to the target application; Extract features based on the historical traffic data of each target to obtain some feature information; Training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application; the security protection model includes a plurality of security protection sub-models, each of which corresponds to an access type; Based on the security protection sub-model in the security protection model corresponding to the access type of the current traffic data generated by accessing the target application, the current traffic data is judged for abnormal access behavior to obtain a detection result of the abnormal access behavior; The feature extraction is performed based on the historical traffic data of each target to obtain a number of feature information, specifically including: Based on the historical traffic data of each target, obtaining the access directory, access time, network address of the accessing party, request message and response information corresponding to the historical traffic data of each target; Based on the access directory, access time, network address of the accessing party, request message and response information corresponding to the historical traffic data of each target, the first characteristic information corresponding to the underlying logic, the second characteristic information corresponding to the access logic and the third characteristic information corresponding to the business logic are extracted.

2. The method according to claim 1, characterized in that The obtaining of a number of target historical traffic data corresponding to the target application specifically includes: Obtain some historical traffic data corresponding to the target application; The historical traffic data are preprocessed based on a convolutional neural network model to filter out a number of target historical traffic data from the historical traffic data; each target historical traffic data is the traffic data generated by normal access behavior.

3. The method according to claim 1, characterized in that The extracting of first characteristic information corresponding to the underlying logic, second characteristic information corresponding to the access logic, and third characteristic information corresponding to the business logic based on the access directory, access time, network address of the accessing party, request message, and response information corresponding to each target historical traffic data specifically includes: Extracting and obtaining directory structure features corresponding to the target application based on the access directory to obtain the first feature information; Extracting and obtaining time features, network address features and request features corresponding to the target application based on the access time, the network address of the accessing party and the request message, and obtaining the second feature information based on one or more of the time features, network address features and request features; Business feature extraction is performed based on the response information to obtain response features corresponding to the target application, so as to obtain the third feature information.

4. The method according to claim 1, characterized in that The feature extraction is performed based on the historical traffic data of each target to obtain a number of feature information, specifically including: Extract features based on the access type of historical traffic data of each target, and obtain a feature information set corresponding to the access type; The security protection model is trained based on each of the characteristic information to obtain a security protection model corresponding to the target application, which specifically includes: training the model based on the characteristic information sets corresponding to each access type to obtain a security protection sub-model, so as to obtain the security protection model based on each of the security protection sub-models.

5. The method according to claim 2, characterized in that The obtaining of a number of historical traffic data corresponding to the target application specifically includes: Obtain historical traffic data corresponding to the target application based on the bypass reverse proxy method.

6. The method according to claim 1, characterized in that When the detection result of the abnormal access behavior is obtained, the method further includes: prompting the abnormal access behavior in a predetermined prompting manner.

7. A device for detecting abnormal access behavior, characterized in that: include: An acquisition module, used to acquire a number of target historical traffic data corresponding to a target application; An extraction module, used for extracting features based on the historical traffic data of each target to obtain a number of feature information; A training module, used for training a security protection model based on each of the feature information to obtain a security protection model corresponding to the target application; the security protection model includes a plurality of security protection sub-models, each of which corresponds to an access type; A detection module, configured to determine abnormal access behavior of the current traffic data based on the security protection sub-model in the security protection model corresponding to the access type of the current traffic data generated by accessing the target application, and obtain a detection result of the abnormal access behavior; The extraction module is specifically used for: Based on the historical traffic data of each target, obtaining the access directory, access time, network address of the accessing party, request message and response information corresponding to the historical traffic data of each target; Based on the access directory, access time, network address of the accessing party, request message and response information corresponding to the historical traffic data of each target, the first characteristic information corresponding to the underlying logic, the second characteristic information corresponding to the access logic and the third characteristic information corresponding to the business logic are extracted.

8. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for detecting abnormal access behavior described in any one of claims 1 to 6 are implemented.

9. An electronic device, characterized in that: The device at least comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method for detecting abnormal access behavior as described in any one of claims 1 to 6 when executing the computer program on the memory.