Wind power plant personnel identity verification system and method based on face recognition and biological characteristics
By adopting a multimodal identity verification system in wind farms, combining biological features such as face, fingerprints, and iris, and using deep feature extraction and combined Bayesian methods, the existing system is easily disturbed and misidentified, and realizes the problem of realizing the problem of interference and misidentification. Efficient and secure authentication.
Patent Information
- Application Number
- CN202411033996.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-12
AI Technical Summary
The identity verification system of existing wind farms is susceptible to external factors, resulting in misidentification and poses safety hazards.
Using a multimodal authentication system based on face recognition and biometric features, multiple biometric information is collected in real time through the front-end acquisition terminal, edge computing gateways quickly aggregate and process, and cloud management platform generates multi-layer authentication strategies, combining deep-level features Extract and combine Bayesian methods for verification.
It improves the accuracy and efficiency of identity verification, reduces the risk of misjudgment, enhances the robustness and security of the system, and adapts to complex environmental conditions.
Smart Images

Figure CN118940242A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of wind power plant security identification, and in particular to a wind farm personnel identity verification system and method based on face recognition and biometrics. Background Art
[0002] With the rapid development of the wind power industry, the scale of wind farms has continued to expand, which has put forward higher requirements for the safety management and personnel control of wind farms. At present, the identity authentication methods used in wind farms mainly use IC cards, passwords and other means. These authentication methods allow staff to enter the wind farm by swiping a card or entering a password. However, the use of IC cards and passwords has security risks such as easy copying, easy loss, and easy theft. When non-wind farm personnel obtain IC cards or passwords, they can enter the wind farm, which poses a great security risk. Therefore, in order to improve the security level of wind farms, it is urgent to introduce more reliable and efficient identity authentication technology.
[0003] In recent years, biometric technology has been widely used in the field of identity authentication due to its uniqueness and stability. Among them, face recognition, fingerprint recognition, iris recognition and other technologies have gradually become the mainstream methods of identity authentication due to their non-contact, fast and accurate characteristics. However, there are still some limitations in the application of single biometric recognition in wind farm identity authentication. For example, the influence of factors such as ambient lighting, occlusion, and forgery can cause the verification system to misidentify, allowing non-wind farm personnel to enter the wind farm, resulting in serious accidents. Summary of the invention
[0004] In order to solve the problem that the existing wind farm adopts single biometric authentication which is easily interfered by external factors and causes misidentification of the authentication system, the present invention provides a wind farm personnel identity authentication system and method based on face recognition and biometrics.
[0005] To achieve the above object, the present invention provides the following technical solutions:
[0006] The present invention discloses a wind farm personnel identity verification system based on face recognition and biometrics, comprising a front-end acquisition terminal, the front-end acquisition terminal data is connected to an edge computing gateway, and the edge computing gateway data is connected to a cloud management platform;
[0007] The front-end acquisition terminal is used to collect multiple biometric information, pre-process and extract multiple biometric features, and issue an identity authentication request;
[0008] The edge computing gateway is used to aggregate the biometric features after the preprocessing and feature extraction, as well as the identity authentication request and result;
[0009] The cloud management platform is used to process the biometric features obtained in the edge computing gateway, generate identity authentication results, and then transmit them to the edge computing gateway.
[0010] Preferably, the front-end acquisition terminal includes a face recognition camera, a fingerprint / palmprint collector, an iris collector and a liveness detection sensor, and the face recognition camera, the fingerprint / palmprint collector, and the iris collector are respectively connected to the edge computing gateway data.
[0011] Preferably, the liveness detection sensor comprises an infrared camera and a structured light projector for face liveness detection, a multispectral imager, a skin color and texture sensor, and an iris sensor;
[0012] The infrared camera and the structured light projector are respectively connected to the face recognition camera and the edge computing gateway;
[0013] The multispectral imager is respectively connected to the face recognition camera and the edge computing gateway by data;
[0014] The skin color and texture sensors are respectively connected to the face recognition camera and the edge computing gateway;
[0015] The iris meter is data-connected to the face recognition camera and the edge computing gateway.
[0016] Preferably, the front-end acquisition terminal also includes a biometric preprocessor, and the biometric preprocessor is respectively connected to the face recognition camera, the fingerprint / palmprint collector, and the iris collector.
[0017] Preferably, the cloud management platform includes a biometric feature comparison module, a heterogeneous feature fusion module, an adaptive strategy management module, a data security management module and a wind farm business module; the biometric feature comparison module is respectively connected to the edge computing gateway and the heterogeneous feature fusion module;
[0018] The heterogeneous feature fusion module is respectively connected with the biometric feature comparison module and the adaptive strategy management module;
[0019] The adaptive policy management module is respectively connected to the edge computing gateway and the data security management module;
[0020] The data security management module is data-connected to the wind farm business module, and the wind farm business module is communicatively connected to the edge computing gateway.
[0021] The present invention discloses a wind farm personnel identity verification method based on face recognition and biometrics, comprising the following steps:
[0022] Obtaining biometrics of the verification person to generate a biometric vector;
[0023] Generating a feature vector verification result and a threshold comparison result based on the biometric feature vector;
[0024] A modal verification result is formed based on the biometric feature vector, a feature verification strategy is established based on the modal verification result, the feature vector verification result and the threshold comparison result, the level area of the verification person is obtained and verified through the feature verification strategy to obtain the identity authentication result.
[0025] Preferably, the step of obtaining the biometrics of the verification personnel to generate a biometrics vector comprises:
[0026] Collect the facial image of the verification personnel, extract the features of the facial image, and obtain the facial feature vector F facefeature ;
[0027] Collect the fingerprint image and palm print image of the verification person, extract the features of the fingerprint image and the palm print image, and obtain the fingerprint feature vector F fingerprint And the palmprint feature vector F palmprint ;
[0028] Collecting an iris image of a verification person, performing feature extraction on the iris image, and obtaining an iris feature vector;
[0029] Collect the facial feature vector F facefeature , the fingerprint feature vector F fingerprint , the palmprint feature vector F palmprint , and the iris feature vector to obtain the biometric feature vector.
[0030] Preferably, the forming of the feature vector verification result and the threshold comparison result based on the biometric feature vector includes:
[0031] The face feature vector F facefeature , the fingerprint feature vector F fingerprint , the palmprint feature vector F palmprint Performing environmental feature extraction with the iris feature vector to sequentially obtain a first environmental feature, a second environmental feature, a third environmental feature, and a fourth environmental feature;
[0032] Acquire a current environment image, extract the current environment image, and obtain current environment features;
[0033] Compare the first environmental feature, the second environmental feature, the third environmental feature and the fourth environmental feature with the current environmental feature respectively, and obtain a face image discrimination threshold, a fingerprint image discrimination threshold, a palm print image discrimination threshold and an iris image discrimination threshold in sequence;
[0034] Compare the face image discrimination threshold, the fingerprint image discrimination threshold, the palm print image discrimination threshold, and the iris image discrimination threshold with the preset discrimination threshold to obtain the threshold comparison result
[0035] The face feature vector F facefeature , the fingerprint feature vector F fingerprint , the palmprint feature vector F palmprint , the iris feature vector is compared with a preset feature vector to obtain the feature vector verification result.
[0036] Preferably, forming a modality verification result based on a biometric feature vector comprises:
[0037] Performing deep feature extraction on the biometric feature vector to obtain modal features;
[0038] The modal features are mapped to a common feature subspace through a joint Bayesian method, and modal verification is performed on each modal feature preset in the common feature subspace to obtain a modal verification result.
[0039] Preferably, the feature verification strategy formed based on the modal verification result, the feature vector verification result and the threshold comparison result includes:
[0040] Performing verification score processing on the feature vector verification result, the threshold comparison result and the modal verification result to obtain a feature vector verification score vector, a threshold verification score vector and a modal verification score vector in sequence;
[0041] Based on the modality verification score vector, the feature vector verification score vector and the threshold verification score vector, adaptive fusion is performed to dynamically generate a fusion weight vector, obtain a final identity authentication result through weighted voting, and obtain a complete feature verification strategy;
[0042] Generate a final verification threshold for the modal verification score vector, the feature vector verification score vector and the threshold verification score vector, compare it with the preset final threshold, generate a regular identity authentication result, and obtain a regular verification strategy;
[0043] The conventional verification strategy and all feature verification strategies are combined to obtain the feature verification strategy.
[0044] Compared with the prior art, the present invention has the following beneficial technical effects:
[0045] The present invention discloses a wind farm personnel identity authentication system based on face recognition and biometrics, which combines a front-end acquisition terminal, an edge computing gateway and a cloud management platform to form an efficient and secure biometric identity authentication system. In the system, multiple biometric information, namely fingerprint, facial recognition, iris, etc., can be collected in real time through the front-end acquisition terminal, and the biometrics are preprocessed and feature extracted through the built-in algorithm, which reduces the bandwidth demand for data transmission and improves the security and privacy protection of the data. Secondly, the system uses the edge computing gateway as a bridge connecting the front-end acquisition terminal and the cloud management platform to quickly aggregate the biometric information after preprocessing and feature extraction sent by the front-end acquisition terminal, as well as the identity authentication request and results, thereby improving the response speed of the system. Finally, the system uses the cloud management platform as the core of the entire system, which is responsible for fusing and comparing the acquired biometrics, generating a multi-layer identity authentication strategy, and realizing efficient and secure biometric identity authentication in wind farms, and improving the accuracy and efficiency of identity authentication.
[0046] Furthermore, the front-end acquisition terminal in this system integrates a face recognition camera, a fingerprint / palmprint collector, an iris collector, and a liveness detection sensor, which can accurately and quickly capture the user's biometric information. Through the data connection with the edge computing gateway, the front-end acquisition terminal can send the collected biometric information to the gateway in real time for further processing, thereby improving the efficiency and accuracy of identity authentication.
[0047] Furthermore, the front-end acquisition terminal in the system also includes a biometric preprocessor. Preprocessing by the biometric preprocessor can effectively reduce the interference of invalid data and improve the accuracy of identity authentication.
[0048] Furthermore, the cloud management platform in this system uses the biometric matching module to accurately compare the biometric information collected by the front end with the information stored in the database to ensure the accuracy of identity authentication. Then, the heterogeneous feature fusion module realizes the effective integration of different biometric information, improving the robustness and adaptability of the system. The adaptive strategy management module can dynamically adjust the identity authentication strategy according to the actual application scenario and user needs, and provide personalized identity authentication services. The data security management module ensures the security of biometric information and the identity authentication process, and prevents data leakage and illegal access. As a module combined with specific business scenarios, the wind farm business module can provide targeted identity authentication services according to the actual needs of the wind farm to ensure the safe operation of the wind farm. At the same time, through the communication connection with the edge computing gateway, the cloud management platform can obtain the operating status and data of the wind farm in real time, providing strong support for the safety management of the wind farm.
[0049] The present invention also discloses a wind farm personnel identity authentication method based on face recognition and biometrics. The method achieves personalized and high-precision identity authentication by obtaining the biometrics of the authenticator and generating a biometric vector. Then, by forming a feature vector verification result and a threshold comparison result, a quantitative basis and judgment criteria are provided for identity authentication. Then, based on the modal verification result formed by the biometric vector, it is possible to comprehensively consider a variety of biometric information, thereby improving the accuracy and robustness of identity authentication. This multimodal verification method enables the system to more comprehensively evaluate the identity of the authenticator and reduce the risk of misjudgment and missed judgment. Finally, a feature verification strategy is formed in combination with the wind farm safety level area to achieve an effective combination of identity authentication and wind farm safety management. This targeted verification strategy can ensure the safe operation of the wind farm and reduce safety risks.
[0050] Furthermore, by extracting environmental features from biometric vectors such as face, fingerprint, palm print and iris, this method can capture subtle changes in these biometric features in a specific environment, such as light, temperature, humidity, etc. The introduction of such environmental features increases the accuracy and robustness of identity authentication, allowing the system to adapt to various complex environmental conditions. Secondly, the extracted environmental features are compared with the current environmental features to obtain the discrimination thresholds of different biometric images. This step effectively evaluates the credibility of the biometric image in the current environment and avoids misjudgment or missed judgment due to environmental factors. Finally, by comparing these discrimination thresholds with the preset discrimination thresholds, and comparing the biometric vectors with the preset feature vectors, the identity of the verification person is comprehensively judged, the accuracy of identity authentication is improved, and the security of the system is enhanced. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 The architecture diagram of the wind farm personnel identity verification system based on face recognition and biometrics provided by the present invention;
[0052] Figure 2 A schematic diagram of the DCNN face feature extraction network structure in the wind farm personnel identity verification system based on face recognition and biometrics provided by the present invention;
[0053] Figure 3 A decision flow chart of the adaptive identity authentication security strategy module in the wind farm personnel identity authentication system based on face recognition and biometrics provided by the present invention. DETAILED DESCRIPTION
[0054] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.
[0055] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.
[0056] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0057] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or a communication; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0058] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is lower in level than the second feature.
[0059] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.
[0060] The present invention discloses a wind farm personnel identity authentication system based on face recognition and biometrics. The system includes a front-end acquisition terminal, which is used to collect multiple biometric information, pre-process and extract features of multiple biometrics, and issue identity authentication requests; an edge computing gateway, which is used to aggregate and fuse the biometrics after pre-processing and feature extraction, and communicate with a cloud management platform to report identity authentication requests and results; and a cloud management platform, which is used to compare the biometrics transmitted by the edge computing gateway, fuse heterogeneous features, perform adaptive strategy management, and data security management.
[0061] The contents disclosed in the present invention are explained and illustrated below in combination with embodiments and drawings.
[0062] See also Figure 1 , the present invention discloses a wind farm personnel identity verification system based on face recognition and biometrics, the system includes a front-end acquisition terminal, an edge computing gateway and a cloud management platform;
[0063] Among them, the front-end acquisition terminal includes a face recognition camera, a fingerprint / palmprint collector, an iris collector and a liveness detection sensor.
[0064] The face recognition camera is used to collect the facial image of the verification person, and the facial image is used for face detection and feature extraction using a deep convolutional neural network (DCNN). That is, the facial image of the verification person is input into the DCNN for processing, such as Figure 2 As shown in FIG. 1 , the collected facial information is input into the convolution layer through the input layer for 2-layer convolution processing, and then input into the pooling layer for processing, and then input into the 2-layer convolution layer for processing, and then processed in the input pooling layer, and then processed by two layers of fully connected layers in sequence and output through the output layer to extract a 128-dimensional facial feature vector, and calculate the similarity through the Euclidean distance to obtain a 128-dimensional face feature vector F facefeature ; Among them, the loss function in DCNN processing adopts the weighted sum of cross entropy loss and center loss.
[0065] The expression of weighted sum is:
[0066] L=αL cross-entropy +βOL center-loss (1)
[0067] Among them, α is the weight coefficient of cross entropy loss, and β is the weight coefficient of center loss.
[0068] The fingerprint / palmprint collector uses an optical sensor to collect the fingerprint image and palmprint image of the verification person, and then extracts the fingerprint details, endpoints, bifurcation points and other features in the fingerprint image and the palmprint lines, base points and other features in the palmprint image, and processes the fingerprint details, endpoints, bifurcation points and other features to generate a 2048-dimensional fingerprint feature vector F fingerprint , the palmprint features such as lines and base points are processed to generate a 2048-dimensional palmprint feature vector F palmprint .
[0069] The iris collector uses a near-infrared camera to collect the iris image of the verification person in the 730-860nm band, extracts the Gabor texture features of the iris from the iris image, and generates a 2048-dimensional iris code C. iris , and obtain a 2048-dimensional iris feature vector.
[0070] The liveness detection sensor is used for liveness verification of biometric features, that is, receiving the face feature vector F facefeature 、Fingerprint feature vector F fingerprint , palmprint feature vector F palmprint and iris feature vector, obtain the current environment image, extract the current environment image, generate the current environment feature, extract the received face feature vector F facefeature 、Fingerprint feature vector F fingerprint , palmprint feature vector F palmprint and the environmental features in the iris feature vector to generate image environmental data, compare the image environmental data with the current environmental data, and determine the received face feature vector F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F palmprint The liveness detection sensor includes an infrared camera and structured light projector for face liveness detection, a multi-spectral imager for fingerprint liveness detection, a skin color and texture sensor for palm print liveness detection, and an iris detector for iris liveness detection. That is, the face of the verification person is detected by the infrared camera and structured light projector, and the face feature vector F is extracted. facefeature The first environmental feature in the multispectral imager detects the fingerprint of the verification person and extracts the fingerprint feature vector F fingerprint The second environmental feature in the skin color and texture sensor detects the palm print of the verification person and extracts the palm print feature vector F palmprint The iris meter measures the iris state of the verification person and extracts the fourth environmental feature in the iris feature vector.
[0071] The front-end acquisition terminal also includes a biometric feature preprocessor, which is used to compare and analyze the first environmental feature with the current environmental feature to generate a face feature vector Ffacefeature The face image discrimination threshold is calculated; the second environment feature is compared with the current environment feature to generate a fingerprint feature vector F fingerprint The fingerprint image discrimination threshold is calculated; the third environmental feature is compared with the current environmental feature to generate a palm print feature vector F palmprint palmprint image discrimination threshold; compare and analyze the fourth environmental feature with the current environmental feature to generate an iris image discrimination threshold of the iris feature vector.
[0072] Edge computing network management, used to aggregate facial feature vectors F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F palmprint , iris feature vector, face image discrimination threshold, fingerprint image discrimination threshold, palm print image discrimination threshold, iris image discrimination threshold; and transmit these data to the cloud management platform, and transmit the identity authentication request / result to the cloud management platform.
[0073] The cloud management platform is used to compare the biometric features transmitted by the edge computing gateway, fuse heterogeneous features, perform adaptive strategy management and data security management; it includes a biometric comparison module, a heterogeneous feature fusion module, an adaptive strategy management module, a data security management module and a wind farm business module;
[0074] Biometric comparison module, used to obtain the facial feature vector F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F palmprint and iris feature vector, and the face feature vector F facefeature Compare it with the preset face feature vector to obtain the face feature vector verification result; fingerprint Compare it with the preset fingerprint feature vector to obtain the fingerprint feature vector verification result; palmprintThe palm print feature vector is compared with the preset palm print feature vector to obtain the palm print feature vector verification result; the iris feature vector is compared with the preset iris feature vector to obtain the iris feature vector verification result; the face feature vector verification result, the fingerprint feature vector verification result, the palm print feature vector verification result and the iris feature vector verification result are transmitted to the heterogeneous feature fusion module; at the same time, the face image discrimination threshold, the fingerprint image discrimination threshold, the palm print image discrimination threshold and the iris image discrimination threshold are obtained, and the face image discrimination threshold is compared with the preset face threshold range to obtain the face threshold comparison result; the fingerprint image discrimination threshold is compared with the preset fingerprint threshold range to obtain the fingerprint threshold comparison result; the palm print image discrimination threshold is compared with the preset palm print threshold range to obtain the palm print threshold comparison result; the iris image discrimination threshold is compared with the preset iris threshold range to obtain the iris threshold comparison result; then the face threshold comparison result, the fingerprint threshold comparison result, the palm print threshold comparison result and the iris threshold comparison result are aggregated and processed by the data security management module.
[0075] The heterogeneous feature fusion module is used to obtain the face feature vector F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F palmprint and iris feature vector, face feature vector F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F palmprint The DCNN is used to perform deep feature extraction on the face modal features, fingerprint modal features, palmprint modal features and iris modal features respectively, and these modal features are mapped to a common feature subspace through the joint Bayesian method, so that the dimensions and scales of these modal features are unified, and the preset modal features therein are modally verified to obtain the confidence of each modal verification result, that is, the confidence of the face modal verification result, the confidence of the fingerprint modal verification result, the confidence of the palmprint modal verification result and the confidence of the iris modal verification result, and the confidence of the face modal verification result, the confidence of the fingerprint modal verification result, the confidence of the palmprint modal verification result and the confidence of the iris modal verification result are calculated. The confidence of the modal verification result and the confidence of the iris modal verification result are scored to obtain face verification score data, fingerprint verification score data, palm print verification score data and iris verification score data in turn, and a face modal verification score vector, a fingerprint modal verification score vector, a palm print modal verification score vector and an iris modal verification score vector are generated in turn through the face verification score data, the fingerprint verification score data, the palm print verification score data and the iris verification score data, and the face modal verification score vector, the fingerprint modal verification score vector, the palm print modal verification score vector and the iris modal verification score vector are aggregated to obtain a verification score vector for each modal verification result;
[0076] Processing the face threshold comparison result, the fingerprint threshold comparison result, the palm print threshold comparison result and the iris threshold comparison result to generate a verification score vector for each threshold;
[0077] The face feature vector verification results, the fingerprint feature vector verification results, the palm print feature vector verification results and the iris feature vector verification results are processed to generate a verification score vector for each feature vector, which is then transmitted to the adaptive strategy management module.
[0078] The data security management module is used to obtain the regional image of the wind farm, construct the regional distribution map of the wind farm, mark the regional security level on the regional distribution map of the wind farm, and set the biometric verification level according to the regional security level; that is, for the medium and high security areas of the wind farm, such as the control center, it is marked as a high-level verification area, and a high security verification level is set in this area; low security areas, such as offices, are marked as ordinary verification areas, and a low security verification level is set in this area, and the wind farm identity authentication level distribution model data is generated and transmitted to the adaptive strategy management module.
[0079] The adaptive strategy management module is used to receive the verification request of the edge computing gateway, obtain the high security verification level and the low security verification level in the wind farm identity authentication level distribution model data, generate a high pass threshold for the high security verification level, and use the high pass threshold to form all feature verification strategies for identity authentication, obtain high-level identity authentication results and record high-level abnormal verification; generate a conventional verification strategy for the low security verification level, form some feature verification strategies through the conventional threshold, obtain low-level identity authentication results and record conventional abnormal verification; the adaptive strategy management module processes the high-level abnormal verification and the conventional abnormal verification, that is, compares the high-level abnormal verification and the conventional abnormal verification with the abnormal verification data preset therein, and generates a verification log or an alarm log, such as Figure 3 shown.
[0080] Among them, the overall feature verification strategy is to adaptively fuse each feature vector verification score vector, each threshold verification score vector and each modality verification score vector, and dynamically generate a fusion weight vector w = [w1, w2, ..., w N ] T , the final identity authentication result y is obtained through weighted voting, and the final identity authentication result y is input into the wind farm business system.
[0081] The expression of the fusion weight vector w is:
[0082] w=f fuzzy (s, c, e)(2)
[0083] Where s = [s1, s2, ..., s N ] Tis the verification score vector of each modal feature, c is the context information of the current verification scene (such as lighting, occlusion, etc.), e is the system's risk assessment of the current environment, and f fuzzy is the fuzzy logic reasoning function.
[0084] The expression of the final authentication result y is:
[0085]
[0086] Among them, s is the verification score vector of each modal feature, w is the fusion weight vector, and N is the number.
[0087] The conventional verification strategy is to obtain each feature vector verification score vector, each threshold verification score vector and each modality verification score vector, generate a final verification threshold through each feature vector verification score vector, each threshold verification score vector and each modality verification score vector, compare it with the set preset final threshold, and generate a conventional identity authentication result.
[0088] The wind farm business module is used to obtain the routine final verification results and the final identity authentication results, transmit them to the edge computing gateway, and obtain the verification log or alarm log to issue a warning to the management personnel.
[0089] The present invention uses a variety of biometric technologies such as face recognition, fingerprint recognition, iris recognition, and palm print recognition, combined with innovative methods such as front-end liveness detection, edge computing, heterogeneous feature fusion, adaptive security strategy, and federated learning privacy protection, to build a complete, high-performance, safe and reliable wind farm personnel identity authentication system. The system solves the problems of low accuracy, susceptibility to environmental interference, and insufficient privacy protection of traditional single biometric recognition. At the same time, it fully adapts to the special working environment and management requirements of wind farms, and provides a solid foundation for the intelligent security and smart operation of wind farms.
[0090] The solution makes full use of cutting-edge technologies such as artificial intelligence, big data, edge computing, and privacy protection. Aiming at the special geographical environment and management needs of wind farms, the solution has been comprehensively optimized in terms of biometric recognition performance, system security and reliability, user privacy protection, and platform expansion flexibility, forming a set of advanced technology, complete functions, and strong practicality. The overall solution for wind farm personnel identity authentication has overcome the limitations and shortcomings of existing technologies, has significant innovation and practical value, and is suitable for promotion and application in wind farms and other similar special industries and scenarios.
[0091] The contents disclosed in the present invention are further explained and illustrated in combination with practical applications.
[0092] A large wind power operation company is building a new million-kilowatt wind farm project. The wind farm is located in a remote mountainous area with a harsh environment and difficult personnel management. In order to strengthen the safety precautions and personnel management of the wind farm, the company decides to deploy the wind farm personnel identity verification system of the present invention.
[0093] First, front-end data collection terminals were installed at key areas such as the main entrances and exits, control center, and engine room of the wind farm.
[0094] Secondly, multiple edge gateways are deployed in the wind farm, which are responsible for aggregating, preprocessing and fusing the data from each collection point, and communicating securely with the cloud management platform through the 5G private network.
[0095] In actual operation, wind farm staff only need to scan their faces or press their palms on the front-end collection terminal at the entrance and exit to quickly complete identity confirmation, which greatly improves the efficiency of passage.
[0096] Once, a technician needed to enter the wind turbine cabin for routine maintenance. When he took off his dust glasses in front of the front-end acquisition terminal at the cabin door, the system found that his iris features did not completely match the preset information, but his face and fingerprint features were verified. The cloud platform immediately started the secondary identity verification process and asked him to perform palm print verification on the front-end acquisition terminal device outside the door. Because the cabin is a high-security area, the system automatically sets a high pass threshold. In the end, the employee successfully passed the palm print verification and was allowed to enter the cabin.
[0097] After more than a year of continuous operation, the system has demonstrated excellent recognition performance and stability in harsh environments, effectively controlling the occurrence of illegal intrusions. At the same time, with a strict privacy protection mechanism and authority management system, employees' biometric data is always safely protected, and no privacy data leakage has occurred. The wind farm's intelligent security system has become an industry benchmark, providing excellent practice and demonstration for the safe development of the wind power industry.
[0098] The present invention also discloses a wind farm personnel identity verification method based on face recognition and biometrics, comprising the following steps:
[0099] Obtaining biometrics of the verification person to generate a biometric vector;
[0100] Specifically, the biometric features of the verification personnel are collected by the collection device, that is, the face image, fingerprint image, palm print image and iris image of the verification personnel are collected, and the features of the face image, fingerprint image, palm print image and iris image are extracted respectively to obtain the face feature vector F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F pa1mprint and iris feature vector, merged with face feature vector F facefeature, fingerprint feature vector F fingerprint , palmprint feature vector F palmprint And the iris feature vector is used to obtain the biometric feature vector.
[0101] Generating a feature vector verification result and a threshold comparison result based on the biometric feature vector;
[0102] Specifically, the face feature vector F facefeature , fingerprint feature vector F fingerprint , palmprint feature vector F palmprint and extracting environmental features from the iris feature vector to obtain a first environmental feature, a second environmental feature, a third environmental feature, and a fourth environmental feature in sequence;
[0103] Acquire the current environment image, extract the current environment image, and obtain the current environment features;
[0104] The first environmental feature, the second environmental feature, the third environmental feature and the fourth environmental feature are respectively compared with the current environmental feature to obtain the face image discrimination threshold, the fingerprint image discrimination threshold, the palm print image discrimination threshold and the iris image discrimination threshold in turn.
[0105] The facial feature vector F facefeature Compare with the preset face feature vector to obtain the face feature vector verification result; the fingerprint feature vector F fingerprint Compare with the preset fingerprint feature vector to obtain the fingerprint feature vector verification result; convert the palm print feature vector F palmprint Compare with the preset palmprint feature vector to obtain the palmprint feature vector verification result; compare the iris feature vector with the preset iris feature vector to obtain the iris feature vector verification result, and collect the face feature vector verification result, fingerprint feature vector verification result, palmprint feature vector verification result and iris feature vector verification result to obtain the feature vector verification result.
[0106] The face image discrimination threshold is compared with the preset face threshold range to obtain the face threshold comparison result; the fingerprint image discrimination threshold is compared with the preset fingerprint threshold range to obtain the fingerprint threshold comparison result; the palmprint image discrimination threshold is compared with the preset palmprint threshold range to obtain the palmprint threshold comparison result; the iris image discrimination threshold is compared with the preset iris threshold range to obtain the iris threshold comparison result.
[0107] A modal verification result is formed based on the biometric feature vector, a feature verification strategy is established based on the modal verification result, the feature vector verification result and the threshold comparison result, the level area of the verification person is obtained and verified through the feature verification strategy to obtain the identity authentication result.
[0108] Specifically, for the face feature vector F facefeature, fingerprint feature vector F fingerprint , palmprint feature vector F palmprint and iris feature vectors are respectively subjected to deep feature extraction using DCNN to generate face modal features, fingerprint modal features, palmprint modal features and iris modal features, these modal features are mapped to a common feature subspace by a joint Bayesian method, and each modal feature preset in the common feature subspace is subjected to modal verification to obtain the confidence of each modal verification result, the confidence of each modal verification result is scored to obtain face modal verification score vector, fingerprint modal verification score vector, palmprint modal verification score vector and iris modal verification score vector, and the modal verification score vector is obtained by aggregation;
[0109] Processing the face threshold comparison result, the fingerprint threshold comparison result, the palm print threshold comparison result and the iris threshold comparison result, obtaining the face threshold verification score vector, the fingerprint threshold verification score vector, the palm print threshold verification score vector and the iris threshold verification score vector in turn, and collecting them to obtain the threshold verification score vector;
[0110] Verification score processing is performed on the face feature vector verification results, fingerprint feature vector verification results, palm print feature vector verification results and iris feature vector verification results, and the face feature vector verification score vector, fingerprint feature vector verification score vector, palm print feature vector verification score vector and iris feature vector verification score vector are obtained in turn, and the feature vector verification score vector is obtained by aggregation.
[0111] The modality verification score vector, feature vector verification score vector and threshold verification score vector are adaptively fused, and the fusion weight vector is dynamically generated. The final identity authentication result is obtained through weighted voting to obtain the entire feature verification strategy.
[0112] A final verification threshold is generated for the modal verification score vector, feature vector verification score vector and threshold verification score vector, and compared with the preset final threshold to generate a regular identity authentication result, obtain a regular verification strategy, and combine the regular verification strategy with all feature verification strategies to obtain a feature verification strategy.
[0113] Acquire regional images of wind farms, construct regional distribution maps of wind farms, mark regional security levels on the regional distribution maps of wind farms, and set biometric verification levels according to regional security levels; that is, for medium and high security areas of wind farms, such as the control center, it is marked as a high-level verification area, and a high security verification level is set for this area; low security areas, such as offices, are marked as ordinary verification areas, and a low security verification level is set for this area, and wind farm identity verification level distribution model data is generated.
[0114] Associating the low security verification level with the general verification policy in the feature verification policy, and associating the high security verification level with all feature verification policies in the feature verification policy;
[0115] Obtain the location of the verification personnel, determine the security verification level of the area where the verification personnel is located, determine the feature verification strategy based on the verification level, verify the verification personnel, and obtain the verification result.
[0116] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.
[0117] In addition, it should be understood that although this specification is described in accordance with the implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.
Claims
1. A wind farm personnel identity verification system based on face recognition and biometrics, characterized by: It includes a front-end acquisition terminal, the data of which is connected to an edge computing gateway, and the data of which is connected to a cloud management platform; The front-end acquisition terminal is used to collect multiple biometric information, pre-process and extract multiple biometric features, and issue an identity authentication request; The edge computing gateway is used to aggregate the biometric features after the preprocessing and feature extraction, as well as the identity authentication request and result; The cloud management platform is used to process the biometric features obtained in the edge computing gateway, generate identity authentication results, and then transmit them to the edge computing gateway.
2. The wind farm personnel identity verification system based on face recognition and biometrics according to claim 1 is characterized in that: The front-end acquisition terminal includes a face recognition camera, a fingerprint / palmprint collector, an iris collector and a liveness detection sensor. The face recognition camera, the fingerprint / palmprint collector, and the iris collector are respectively connected to the edge computing gateway data.
3. The wind farm personnel identity verification system based on face recognition and biometrics according to claim 2 is characterized in that: The liveness detection sensor includes an infrared camera and a structured light projector for face liveness detection, a multi-spectral imager, a skin color and texture sensor, and an iris sensor; The infrared camera and the structured light projector are respectively connected to the face recognition camera and the edge computing gateway; The multispectral imager is respectively connected to the face recognition camera and the edge computing gateway by data; The skin color and texture sensors are respectively connected to the face recognition camera and the edge computing gateway; The iris meter is data-connected to the face recognition camera and the edge computing gateway.
4. The wind farm personnel identity verification system based on face recognition and biometrics according to claim 3 is characterized in that: The front-end acquisition terminal also includes a biometric feature preprocessor, which is respectively connected to the face recognition camera, the fingerprint / palmprint acquisition device, and the iris acquisition device.
5. The wind farm personnel identity verification system based on face recognition and biometrics according to claim 1, characterized in that: The cloud management platform includes a biometric feature comparison module, a heterogeneous feature fusion module, an adaptive strategy management module, a data security management module and a wind farm business module; the biometric feature comparison module is respectively connected to the edge computing gateway and the heterogeneous feature fusion module; The heterogeneous feature fusion module is respectively connected with the biometric feature comparison module and the adaptive strategy management module; The adaptive policy management module is respectively connected to the edge computing gateway and the data security management module; The data security management module is data-connected to the wind farm business module, and the wind farm business module is communicatively connected to the edge computing gateway.
6. A wind farm personnel identity verification method based on face recognition and biometrics, characterized in that: The steps include: Obtaining biometrics of the verification person to generate a biometric vector; Forming a feature vector verification result and a threshold comparison result based on the biometric feature vector; A modal verification result is formed based on the biometric feature vector, a feature verification strategy is established based on the modal verification result, the feature vector verification result and the threshold comparison result, the level area of the verification person is obtained and verified through the feature verification strategy to obtain the identity authentication result.
7. The wind farm personnel identity verification method based on face recognition and biometrics according to claim 6 is characterized in that: The step of obtaining the biometrics of the verification personnel to generate a biometrics vector comprises: Collect the facial image of the verification personnel, extract the features of the facial image, and obtain the facial feature vector F facefeature ; Collect the fingerprint image and palm print image of the verification person, extract the features of the fingerprint image and the palm print image, and obtain the fingerprint feature vector F fingerprint And the palmprint feature vector F palmprint ; Collecting an iris image of a verification person, performing feature extraction on the iris image, and obtaining an iris feature vector; Collect the facial feature vector F facefeature , the fingerprint feature vector F fingerprint , the palmprint feature vector F palmprint , and the iris feature vector to obtain the biometric feature vector.
8. The wind farm personnel identity verification method based on face recognition and biometrics according to claim 6, characterized in that: The feature vector verification result and the threshold comparison result formed based on the biometric feature vector include: The face feature vector F facefeature , the fingerprint feature vector F fingerprint , the palmprint feature vector F palmprint Performing environmental feature extraction with the iris feature vector to sequentially obtain a first environmental feature, a second environmental feature, a third environmental feature, and a fourth environmental feature; Acquire a current environment image, extract the current environment image, and obtain current environment features; Compare the first environmental feature, the second environmental feature, the third environmental feature and the fourth environmental feature with the current environmental feature respectively, and obtain a face image discrimination threshold, a fingerprint image discrimination threshold, a palm print image discrimination threshold and an iris image discrimination threshold in sequence; Compare the face image discrimination threshold, the fingerprint image discrimination threshold, the palm print image discrimination threshold, and the iris image discrimination threshold with a preset discrimination threshold to obtain the threshold comparison result; The face feature vector F facefeature , the fingerprint feature vector F fingerprint , the palmprint feature vector F palmprint , the iris feature vector is compared with a preset feature vector to obtain the feature vector verification result.
9. The wind farm personnel identity verification method based on face recognition and biometrics according to claim 6, characterized in that: The forming of the modality verification result based on the biometric feature vector includes: Performing deep feature extraction on the biometric feature vector to obtain modal features; The modal features are mapped to a common feature subspace through a joint Bayesian method, and modal verification is performed on each modal feature preset in the common feature subspace to obtain a modal verification result.
10. The wind farm personnel identity verification method based on face recognition and biometrics according to claim 6, characterized in that: The feature verification strategy based on the modal verification result, the feature vector verification result and the threshold comparison result includes: Performing verification score processing on the feature vector verification result, the threshold comparison result and the modal verification result to obtain a feature vector verification score vector, a threshold verification score vector and a modal verification score vector in sequence; Based on the modality verification score vector, the feature vector verification score vector and the threshold verification score vector, adaptive fusion is performed to dynamically generate a fusion weight vector, obtain a final identity authentication result through weighted voting, and obtain a complete feature verification strategy; Generate a final verification threshold for the modal verification score vector, the feature vector verification score vector and the threshold verification score vector, compare it with the preset final threshold, generate a regular identity authentication result, and obtain a regular verification strategy; The conventional verification strategy and all feature verification strategies are combined to obtain the feature verification strategy.
Citation Information
Cited By
Multi-mode identity authentication method and system based on cloud side-end cooperation
CN120321053A
Long-term identity recognition method and system for multi-cat individuals
CN121074993A
A multi-cat individual-oriented long-term identity recognition method and system
CN121074993B
Personnel biological characteristic information acquisition method based on deep learning
CN121170917A
