Biological characteristic self-adaptive instrument interaction system for industrial vehicle driving safety

Abstract

The invention discloses a biological characteristic adaptive instrument interaction system for industrial vehicle driving safety, and the system comprises a key module which is used for receiving an input instruction of a driver, and serves as a preliminary interface of system interaction; the face acquisition module is used for acquiring a face image of a driver in real time and preprocessing the image; the face recognition module is used for realizing efficient feature extraction and comparison based on a VoVNet-ETSform model and carrying out driver identity verification; the communication module is used for realizing real-time and safe data transmission between each module in the system and an external control center or a background data platform; the storage module is used for storing the collected data; and the control module is used for integrating, coordinating and managing each functional module and optimizing the overall architecture of the system. According to the invention, efficient, safe and intelligent management of industrial vehicle identity recognition is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of biometric adaptive instrument interaction, and in particular to a biometric adaptive instrument interaction system for industrial vehicle driving safety. Background Art

[0002] In the intelligent and safe driving management of industrial vehicles, the application of driver identification and interaction systems is crucial. With the development of industrial automation and intelligent driving technology, traditional driver identity authentication methods face many challenges. For example, authentication methods such as keys, IC cards or passwords are easy to lose, forget or copy, and it is difficult to effectively prevent unauthorized operations. These authentication methods rely on physical media and have low security. Once lost or stolen, serious safety accidents may occur. In addition, traditional identity authentication methods cannot monitor the driver's status in real time, and lack effective response measures for fatigue driving, driver identity changes, etc.

[0003] Some current intelligent identity recognition systems have begun to introduce biometric recognition technology, especially face recognition technology, to improve the security and convenience of identity authentication. Face recognition technology uses deep learning models to compare identities through facial features, which can effectively prevent illegal use and reduce the safety hazards caused by human operation. However, most existing face recognition systems are used in static environments or fixed equipment, such as access control systems, attendance equipment, etc. In industrial vehicle driving environments, complex lighting conditions, dynamic posture changes, partial occlusion, vibration interference and other factors will reduce the accuracy of face recognition. In addition, the existing face recognition methods still have insufficient recognition accuracy and stability when dealing with complex backgrounds, low-light environments and high-speed dynamic changes, which may lead to misidentification or rejection of driver identity authentication, affecting the normal use of the vehicle.

[0004] In response to these problems, some existing technologies have begun to combine deep learning with embedded computing to improve the real-time and robustness of face recognition. For example, some systems use convolutional neural networks (CNNs) for feature extraction and combine attention mechanisms to optimize the feature expression of key areas. However, these methods still have limitations. First, the efficiency of feature extraction is low. The traditional CNN structure is difficult to balance between computational complexity and recognition accuracy, which makes it difficult for embedded systems to run efficiently. Secondly, most of the current face recognition systems are based on static database matching, lack the ability to adapt to real-time driving environments, and have difficulty handling lighting, angle changes, and personalized features of different drivers. In addition, existing recognition methods mainly rely on fixed feature templates for comparison, and cannot dynamically optimize feature selection, resulting in insufficient robustness in the face of environmental changes.

[0005] On the other hand, the interaction methods of existing industrial vehicle instrument systems are relatively traditional, mainly relying on physical buttons, touch screens or voice commands for operation, and lacking real-time monitoring and adaptive adjustment of driver identity information. These interaction methods usually do not have security verification functions, and drivers may make accidental touches or unauthorized operations during operation, increasing safety risks. In addition, most existing instrument interaction systems lack intelligent data analysis functions and cannot make adaptive adjustments based on factors such as driver identity and environmental conditions, making it difficult to efficiently manage the system in a multi-user environment. For example, in a factory or logistics warehousing environment, multiple drivers share the same industrial vehicle, and the system cannot be personalized according to the habits and permissions of different drivers, resulting in increased management complexity and may even affect operational safety.

[0006] In order to improve the driving safety of industrial vehicles, some studies in recent years have attempted to apply optimization algorithms to face recognition to improve the efficiency of feature extraction and identity verification. For example, some methods introduce ant colony optimization algorithms, particle swarm optimization algorithms and other optimized feature matching processes to improve recognition accuracy. However, traditional optimization algorithms have low search efficiency in high-dimensional feature spaces and are difficult to converge to the optimal solution quickly, affecting real-time performance. In addition, these methods usually rely on fixed feature mapping strategies and lack adaptive optimization capabilities for complex driving environments, resulting in reduced recognition accuracy in conditions such as lighting changes and partial occlusion.

[0007] Most existing face recognition optimization methods are limited to a single feature extraction strategy and cannot fully utilize the multi-scale information expression capabilities of deep learning networks. For example, some studies use standard convolutional neural networks (CNNs) for feature extraction, but the CNN structure has the problem of limited local feature expression and is difficult to capture global features. Although the self-attention mechanism (Transformer) can model long-distance feature dependencies, it has high computational complexity and is difficult to run efficiently on embedded systems. In addition, existing feature optimization methods mainly focus on the feature selection level, and lack end-to-end optimization for the feature extraction process, resulting in recognition results limited by the initial feature quality and unable to fully adapt to complex environmental changes.

[0008] In general, the existing industrial vehicle driver identification system has the following defects: first, the traditional identity authentication method has low security and can be easily forged or bypassed; second, the existing face recognition method is not robust enough in complex driving environments and has difficulty coping with changes in lighting, posture and dynamic interference; third, the existing deep learning model has low computational efficiency on embedded systems, making it difficult to achieve efficient and real-time identity recognition; finally, the existing instrument interaction system lacks the ability to link with the driver's identity information, making it difficult to achieve intelligent management.

[0009] In response to these problems, the present invention proposes a biometric adaptive instrument interaction system for industrial vehicle driving safety, which uses the VoVNet-ETSformer network for efficient feature extraction, and combines the improved artificial bee colony optimization algorithm (I-ABC) to optimize the feature selection process to enhance recognition accuracy and real-time performance. The system integrates facial acquisition, face recognition, data storage, communication and control modules in industrial vehicle instruments to achieve efficient verification of the driver's identity, and combines intelligent optimization algorithms to improve the system's adaptive capabilities. Through innovative optimization strategies, the present invention improves the stability of face recognition in complex driving environments, while enhancing the security and intelligence level of the instrument interaction system, and solves many key problems in the prior art.

[0010] Therefore, how to provide a biometric adaptive instrument interaction system for industrial vehicle driving safety is an urgent problem that technicians in this field need to solve. Summary of the invention

[0011] One object of the present invention is to propose a biometric adaptive instrument interaction system for industrial vehicle driving safety. One object of the present invention is to propose a biometric adaptive instrument interaction system for industrial vehicle driving safety. The present invention makes full use of the VoVNet-ETSformer deep learning model, the improved artificial bee colony optimization algorithm and the intelligent instrument interaction system, and describes in detail the optimization method for efficient and accurate driver identity recognition and interactive control in complex driving environments. It has the advantages of high identity authentication security, high recognition accuracy, strong real-time performance and strong system intelligent management capability.

[0012] A biometric adaptive instrument interaction system for industrial vehicle driving safety according to an embodiment of the present invention includes: A key module is used to receive input commands from the driver and serves as a preliminary interface for system interaction; A facial acquisition module is used to acquire the driver's facial image in real time and pre-process the image; The face recognition module implements efficient feature extraction and comparison based on the VoVNet-ETSformer model to perform driver identity verification; Communication module, used to realize real-time and secure data transmission between modules within the system and with the external control center or background data platform; A storage module, used for storing collected data; The control module is used to integrate, coordinate and manage various functional modules and optimize the overall system architecture.

[0013] Optionally, modules can be connected via the following methods: S1, receiving the driver's input command through the key module, using the facial acquisition module to collect the driver's facial image in real time, and pre-processing the facial image; S2, using the VoVNet-ETSformer model to normalize the features of the preprocessed facial images, and using the multi-scale alignment method for geometric transformation; S3, using the single aggregation structure of the VoVNet network to extract multi-scale features from the geometrically transformed image, combined with the self-attention mechanism of the ETSformer module to enhance the feature expression ability and obtain the extracted features; S4. Based on the improved artificial bee colony optimization algorithm, the extracted features are optimized, the bee colony is initialized, the population size is set, and the candidate feature subsets are randomly generated. The employed bees evaluate the features extracted by VoVNet-ETSformer, and the weights are adaptively assigned. The scout bees use chaotic mapping to enhance the global search capability and eliminate local optimality. The follower bees use the improved LévyFlight search strategy to optimize feature selection. The adaptive mutation mechanism adjusts the feature subset, selects the optimal feature subset, and generates the identity recognition result. S5, transmitting the generated identity recognition result to the storage module through the communication module, and comparing it with the pre-stored authorization information; S6. Generate operation permission information in the control module according to the comparison result, display it in real time through the embedded instrument, encrypt the data transmitted by the communication module, monitor the system status in real time, schedule system resources, and record and store identity authentication and interaction data.

[0014] Optionally, the S3 specifically includes: S31. Perform multi-scale feature extraction on the face image after geometric transformation. Use the single aggregation structure of the VoVNet network to establish a cross-layer connection mechanism for feature information at different levels, extract local texture information through a deep convolution module, and fuse global features using a cross-scale aggregation strategy. S32. For the extracted multi-scale features, the dilated convolution technique is used to expand the features, and convolution kernels with different dilation rates are introduced to expand the receptive field. The model learns key facial features at different scales. S33, combined with the self-attention mechanism of the ETSformer module, uses multi-head self-attention to calculate the global correlation between facial feature points, calculates feature weights in different feature spaces, emphasizes the feature contribution of key areas, and maintains spatial structure information through position encoding technology; S34, performing nonlinear transformation on the features optimized by the ETSformer module, using normalization strategy to enhance feature contrast, and adjusting the distribution of feature vectors; S35. Perform feature mapping conversion on the finally extracted features, use the embedding space mapping method to project the high-dimensional features into the low-dimensional identity feature space, and output the optimized features.

[0015] Optionally, the S4 specifically includes: S41. Initialize parameters based on artificial bee colony optimization algorithm and set feature vector set ,in Represents the i-th feature vector, setting the initial candidate feature subset , Represents the jth candidate feature, defines the initial population size N, sets the number of iterations T and the fitness calculation function, sets the optimization objective function J(S), reflects the overall fitness of the entire feature subset S, and calculates the initial weight of each feature : ; in, represents the i-th candidate feature, Representative features The weight is not a fixed value and is constantly updated. The value of J(S) will change with change, Representation characteristics The similarity between the target feature set T; S42, the hired bee stage, through feature weight calculation, feature contribution calculation and feature distribution balance calculation, screens discriminative features, removes redundant features, and covers a variety of feature information; S43, the scout bee stage uses an improved chaotic mapping strategy for global search, combining feature subset fitness calculation and global feature distribution calculation to expand the search range and avoid local optimality; S44, follow the bee stage using improved The search strategy is locally optimized and combined with the local feature balance metric to make the feature subsets reasonably distributed in all feature dimensions; S45, calculating the fitness of the newly generated feature subset, screening the feature subset based on the fitness, normalizing the selected optimal feature subset, obtaining the final feature vector, and generating the identity recognition result.

[0016] Optionally, the S42 specifically includes: S421, calculate the feature weight of each individual after iterative optimization , introduce feature sparsity regularization: ; Among them, m is the total number of feature subsets, n is the total number of feature vectors, The initial weight of each feature Start calculating iterations, and the weight after final iteration optimization is recorded as Features In the subset The matching degree in is the feature sparsity control parameter; S422. Calculate the feature contribution and construct the feature screening objective function: ; in, Features The distribution density in the candidate subset, , To adjust parameters adaptively; S423. Calculate the feature distribution balance as the initial measure of global feature balance: ; in, Representation characteristics The probability distribution in the subset S, log is the logarithmic function.

[0017] Optionally, the S43 specifically includes: S431, introduce improved chaotic mapping search to generate new feature subsets: ; in, is the global search control parameter, is the disturbance factor, combined with the sine chaotic mapping, sin is the sine function; S432, calculate the fitness of the feature subset, for the newly generated feature subset , calculate its fitness: ; in, Select sensitivity parameters for features; S433, optimize the global feature distribution, for the new feature subset , calculate the global feature balance metric: ; in, is the characteristic equalization control parameter.

[0018] Optionally, the S44 specifically includes: S441, local optimization search strategy, introduce improvements Search and optimize feature subsets : ; in, is the optimized feature subset, is the step size factor, is the search control parameter, t is the current iteration number, use Distribution, the mathematical form is as follows: ; in, yes The exponential parameter of the distribution, is the gamma function, x represents The generated random variable, sin is the sine function; S442, reference feature subset fitness Select appropriate feature subsets for balance optimization and calculate local feature balance metrics: ; in, is the characteristic equalization control parameter.

[0019] Optionally, the S45 specifically includes: S451, calculate the final fitness , improve the fitness calculation method, adopt the weighted fusion model, and make the contribution of each metric more reasonable: ; in, , , , is the weighting coefficient; S452, calculate the newly generated feature subset The fitness of , defines the final screening target: ; in, represents the optimal feature subset finally screened out, Indicates that the return results in Get the feature subset with the maximum value; S453. Normalize the selected optimal feature subset to obtain a final feature vector and generate an identity recognition result.

[0020] The present invention combines the VoVNet-ETSformer deep learning model, the improved artificial bee colony optimization algorithm (I-ABC) and the intelligent instrument interaction system to achieve significant technical progress in industrial vehicle driver identification and interaction management, which is specifically reflected in the following three beneficial effects: (1) The present invention uses the VoVNet-ETSformer deep learning model for efficient feature extraction and identity recognition, combined with multi-scale feature extraction, dilated convolution and self-attention mechanism, and can still maintain high-precision face recognition capabilities under complex working conditions such as lighting changes, driver posture changes, partial occlusion, etc. At the same time, the multi-scale alignment method is used to perform geometric transformation on facial features to improve the robustness of the model, ensure that different drivers can quickly and accurately pass identity authentication, and reduce the probability of misidentification and rejection. Compared with traditional identity authentication methods based on IC cards or passwords, the biometric recognition system of the present invention is safer and difficult to forge or bypass, effectively preventing unauthorized personnel from driving vehicles and improving the security of vehicle management.

[0021] (2) The present invention introduces an improved artificial bee colony optimization algorithm (I-ABC) in the feature optimization process, including feature screening in the employed bee stage, global search in the scout bee stage, and local optimization in the follower bee stage, to ensure rapid convergence to the optimal feature subset in the high-dimensional feature space. The chaos mapping global search strategy is used to avoid the optimization from falling into the local optimum, and the LévyFlight search strategy is combined to further optimize the feature subset, thereby improving the real-time performance of identity recognition. At the same time, the adaptive mutation mechanism is used to adjust the feature subset, which improves the stability of feature selection, allowing the system to maintain efficient operation in different driving environments. Compared with the traditional CNN model, the computing efficiency is improved, and the consumption of computing resources is reduced, so that the system can run efficiently in embedded industrial instruments and meet the real-time requirements of vehicle intelligent interaction.

[0022] (3) The present invention integrates intelligent data storage, encrypted communication and control modules in the industrial vehicle instrument interaction system, dynamically adjusts the driver's authority through identity authentication results, and encrypts the communication data, thereby improving the security of the system. In addition, the system can adaptively adjust the interactive interface according to the driver's identity, operating habits and environmental conditions, provide personalized instrument display solutions for different drivers, and enhance user experience. At the same time, the system can monitor the driver's operating status in real time, prompt or restrict abnormal driving behaviors (such as fatigue driving, illegal operations), effectively reduce the safety risks of industrial vehicles, and improve the level of intelligent management. Compared with the existing industrial vehicle instrument system, the present invention realizes the deep integration of identity recognition and instrument interaction, significantly improving the intelligent and automated management capabilities of industrial vehicles.

[0023] In summary, the present invention not only improves the security and accuracy of industrial vehicle driver identity authentication, but also optimizes feature extraction and computing efficiency, while enhancing the intelligent management capabilities of the instrument interaction system, providing an innovative solution for safe driving and efficient management of industrial vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 A flow chart of the biometric adaptive instrument interaction method for industrial vehicle driving safety proposed by the present invention; Figure 2 This is a schematic diagram of the biometric adaptive instrument interaction system module for industrial vehicle driving safety proposed by the present invention. DETAILED DESCRIPTION

[0025] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0026] refer to Figure 1 and Figure 2 , a biometric adaptive instrument interaction system for industrial vehicle driving safety, including: A key module is used to receive input commands from the driver and serves as a preliminary interface for system interaction; A facial acquisition module is used to acquire the driver's facial image in real time and pre-process the image; The face recognition module implements efficient feature extraction and comparison based on the VoVNet-ETSformer model to perform driver identity verification; Communication module, used to realize real-time and secure data transmission between modules within the system and with the external control center or background data platform; A storage module, used for storing collected data; The control module is used to integrate, coordinate and manage various functional modules and optimize the overall system architecture.

[0027] In this implementation, the modules are implemented by the following method: S1, receiving the driver's input command through the key module, using the facial acquisition module to collect the driver's facial image in real time, and pre-processing the facial image; S2, using the VoVNet-ETSformer model to normalize the features of the preprocessed facial images, and using the multi-scale alignment method for geometric transformation; S3, using the single aggregation structure of the VoVNet network to extract multi-scale features from the geometrically transformed image, combined with the self-attention mechanism of the ETSformer module to enhance the feature expression ability and obtain the extracted features; S4. Based on the improved artificial bee colony optimization algorithm, the extracted features are optimized, the bee colony is initialized, the population size is set, and the candidate feature subsets are randomly generated. The employed bees evaluate the features extracted by VoVNet-ETSformer, and the weights are adaptively assigned. The scout bees use chaotic mapping to enhance the global search capability and eliminate local optimality. The follower bees use the improved LévyFlight search strategy to optimize feature selection. The adaptive mutation mechanism adjusts the feature subset, selects the optimal feature subset, and generates the identity recognition result. S5, transmitting the generated identity recognition result to the storage module through the communication module, and comparing it with the pre-stored authorization information; S6. Generate operation permission information in the control module according to the comparison result, display it in real time through the embedded instrument, encrypt the data transmitted by the communication module, monitor the system status in real time, schedule system resources, and record and store identity authentication and interaction data.

[0028] In this implementation, S3 specifically includes: S31. Perform multi-scale feature extraction on the face image after geometric transformation. Use the single aggregation structure of the VoVNet network to establish a cross-layer connection mechanism for feature information at different levels, extract local texture information through a deep convolution module, and fuse global features using a cross-scale aggregation strategy. S32. For the extracted multi-scale features, the dilated convolution technique is used to expand the features, and convolution kernels with different dilation rates are introduced to expand the receptive field. The model learns key facial features at different scales. S33, combined with the self-attention mechanism of the ETSformer module, uses multi-head self-attention to calculate the global correlation between facial feature points, calculates feature weights in different feature spaces, emphasizes the feature contribution of key areas, and maintains spatial structure information through position encoding technology; S34, performing nonlinear transformation on the features optimized by the ETSformer module, using normalization strategy to enhance feature contrast, and adjusting the distribution of feature vectors; S35. Perform feature mapping conversion on the finally extracted features, use the embedding space mapping method to project the high-dimensional features into the low-dimensional identity feature space, and output the optimized features.

[0029] In this implementation, S4 specifically includes: S41. Initialize parameters based on artificial bee colony optimization algorithm and set feature vector set ,in Represents the i-th feature vector, setting the initial candidate feature subset , Represents the jth candidate feature, defines the initial population size N, sets the number of iterations T and the fitness calculation function, sets the optimization objective function J(S), reflects the overall fitness of the entire feature subset S, and calculates the initial weight of each feature : ; in, represents the i-th candidate feature, Representative features The weight is not a fixed value and is constantly updated. The value of J(S) will change with change, Representation characteristics The similarity between the target feature set T; S42, the hired bee stage, through feature weight calculation, feature contribution calculation and feature distribution balance calculation, screens discriminative features, removes redundant features, and covers a variety of feature information; S43, the scout bee stage uses an improved chaotic mapping strategy for global search, combining feature subset fitness calculation and global feature distribution calculation to expand the search range and avoid local optimality; S44, follow the bee stage using improved The search strategy is locally optimized and combined with the local feature balance metric to make the feature subsets reasonably distributed in all feature dimensions; S45, calculating the fitness of the newly generated feature subset, screening the feature subset based on the fitness, normalizing the selected optimal feature subset, obtaining the final feature vector, and generating the identity recognition result.

[0030] In this implementation manner, the S42 specifically includes: S421, calculate the feature weight of each individual after iterative optimization , introduce feature sparsity regularization: ; Among them, m is the total number of feature subsets, n is the total number of feature vectors, The initial weight of each feature Start calculating iterations, and the weight after final iteration optimization is recorded as Features In the subset The matching degree in is the feature sparsity control parameter; S422. Calculate the feature contribution and construct the feature screening objective function: ; in, Features The distribution density in the candidate subset, , To adjust parameters adaptively; S423. Calculate the feature distribution balance as the initial measure of global feature balance: ; in, Representation characteristics The probability distribution in the subset S, log is the logarithmic function.

[0031] In this implementation manner, the S43 specifically includes: S431, introduce improved chaotic mapping search to generate new feature subsets: ; in, is the global search control parameter, is the disturbance factor, combined with the sine chaotic mapping, sin is the sine function; S432, calculate the fitness of the feature subset, for the newly generated feature subset , calculate its fitness: ; in, Select sensitivity parameters for features; S433, optimize the global feature distribution, for the new feature subset , calculate the global feature balance metric: ; in, is the characteristic equalization control parameter.

[0032] In this implementation manner, the S44 specifically includes: .S441, local optimization search strategy, introduce improvements Search and optimize feature subsets : ; in, is the optimized feature subset, is the step size factor, is the search control parameter, t is the current iteration number, use Distribution, the mathematical form is as follows: ; in, yes The exponential parameter of the distribution, is the gamma function, x represents The generated random variable, sin is the sine function; S442, reference feature subset fitness Select appropriate feature subsets for balance optimization and calculate local feature balance metrics: ; in, Controls the rate at which the characteristic's balance changes.

[0033] In this implementation manner, the S45 specifically includes: S451, calculate the final fitness , improve the fitness calculation method, adopt the weighted fusion model, and make the contribution of each metric more reasonable: ; in, , , , is the weighting coefficient; S452, calculate the newly generated feature subset The fitness of , defines the final screening target: ; in, represents the optimal feature subset finally screened out, Indicates that the return results in Get the feature subset with the maximum value; S453. Normalize the selected optimal feature subset to obtain a final feature vector and generate an identity recognition result.

[0034] S453. Normalize the selected optimal feature subset to obtain a final feature vector and generate an identity recognition result.

[0035] Embodiment 1: In order to verify the feasibility of the present invention in implementation, the present invention is applied to an intelligent logistics vehicle management system in a large industrial park, which includes multiple factories and storage areas. Hundreds of industrial vehicles (such as forklifts, tractors, and automatic guided vehicles AGVs) perform transportation, distribution, and loading and unloading tasks between different areas on a daily basis. Due to the large number of industrial vehicles in the park, the driver identity management is chaotic, and there are safety hazards such as unauthorized driving and misoperation. In addition, since drivers often need to change vehicles, traditional identity authentication methods based on IC cards or passwords are not only inefficient, but also easily lead to card loss or password leakage, affecting normal operations.

[0036] The biometric adaptive instrument interaction system for industrial vehicle driving safety of the present invention was tested and applied in this scenario. The system first integrates a key module, a facial acquisition module, a face recognition module, a storage module, a communication module and a control module on the dashboard of the industrial vehicle. After the driver enters the vehicle, the key module triggers the facial acquisition module to collect facial images, and performs feature extraction and comparison through the VoVNet-ETSformer network, and combines the improved artificial bee colony optimization algorithm (I-ABC) for feature optimization. The entire recognition process takes less than 1.2 seconds, and the driver can quickly complete identity authentication without using a key or IC card. After the identity authentication is passed, the system will automatically load the driver's operating permissions and adjust the instrument interface according to his operating habits, such as adjusting the brightness of the dashboard, displaying the most commonly used control button layout, etc.

[0037] The following are the performance test data of the system in different environments and working conditions: Table 1: Performance test results of the biometric adaptive instrument interaction system for industrial vehicle driving safety ; During the specific test process, the system conducted face recognition tests in daytime, nighttime, strong light, low light, and complex environments (such as changes in lighting in the warehouse, drivers wearing helmets, etc.). The experimental results show that the face recognition system of the present invention maintains a high recognition accuracy rate under various lighting conditions, with an average accuracy rate of 99.2%, which is about 3.5% higher than the traditional CNN recognition model. In addition, in dynamic environments (such as the driver walking into the vehicle from a distance, approaching the camera at different angles, etc.), the system can still complete identity authentication within 1.5 seconds, and the rejection rate is less than 0.8%. Under extreme lighting changes (such as direct sunlight, backlight environment), the recognition accuracy rate is still maintained at more than 97.5%, showing strong environmental adaptability.

[0038] To further verify the security of the intelligent instrument interaction system of the present invention, the system simulated a variety of unauthorized driving scenarios, such as: unregistered drivers trying to start the vehicle, abnormal driver identity information (such as wearing a mask, covering part of the face), etc. The results show that the system successfully intercepted 98.7% of illegal driving attempts, provided safety tips on the dashboard, and sent abnormal driving alerts to the background management system through the communication module. In addition, the encrypted communication mechanism adopted by the present invention effectively prevents identity data from being stolen or tampered with, and improves data security compared to traditional IC card-based identity authentication methods.

[0039] During the peak dispatch period of industrial vehicles (such as 9:00-11:00 am and 14:00-16:00 pm every day), the traditional IC card or password verification method usually leads to long waiting time, with an average verification time of about 8.5 seconds per vehicle, while this system only takes 1.2 seconds through face recognition, greatly improving the efficiency of vehicle use. After 3 months of test operation, the average identity verification time of industrial vehicles in the park was shortened by 85.9%, the efficiency of driver identity management was improved by 67.3%, the misuse of vehicles was reduced by 92.5%, and the overall security and intelligence level of the industrial vehicle management system was significantly improved.

[0040] The present invention also tests the system's operating performance during the driver identity authentication process. The system can be recognized normally in extreme industrial environments such as low temperature (-10°C), high temperature (45°C), high humidity (90%RH), and strong vibration, and there is no recognition failure or misjudgment. In terms of computing resource usage, when the optimized VoVNet-ETSformer network of the present invention runs on an embedded device (NVIDIA Jetson AGX Xavier), the average CPU occupancy rate is 38.6%, which is 17.2% lower than the traditional CNN recognition solution, and the energy efficiency ratio is improved by 22.8%, indicating the high efficiency of the system in an embedded environment.

[0041] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A biometric adaptive instrument interaction system for industrial vehicle driving safety, characterized by: include: A key module is used to receive input commands from the driver and serves as a preliminary interface for system interaction; A facial acquisition module is used to acquire the driver's facial image in real time and pre-process the image; The face recognition module implements efficient feature extraction and comparison based on the VoVNet-ETSformer model to perform driver identity verification; Communication module, used to realize real-time and secure data transmission between modules within the system and with the external control center or background data platform; A storage module, used for storing collected data; The control module is used to integrate, coordinate and manage various functional modules and optimize the overall system architecture.

2. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 1 is characterized in that: The modules are implemented in the following ways: S1, receiving the driver's input command through the key module, using the facial acquisition module to collect the driver's facial image in real time, and pre-processing the facial image; S2, using the VoVNet-ETSformer model to normalize the features of the preprocessed facial images, and using the multi-scale alignment method for geometric transformation; S3, using the single aggregation structure of the VoVNet network to extract multi-scale features from the geometrically transformed image, combined with the self-attention mechanism of the ETSformer module to enhance the feature expression ability and obtain the extracted features; S4. Based on the improved artificial bee colony optimization algorithm, the extracted features are optimized, the bee colony is initialized, the population size is set, and the candidate feature subsets are randomly generated. The employed bees evaluate the features extracted by VoVNet-ETSformer, and the weights are adaptively assigned. The scout bees use chaotic mapping to enhance the global search capability and eliminate local optimality. The follower bees use the improved LévyFlight search strategy to optimize feature selection. The adaptive mutation mechanism adjusts the feature subset, selects the optimal feature subset, and generates the identity recognition result. S5, transmitting the generated identity recognition result to the storage module through the communication module, and comparing it with the pre-stored authorization information; S6. Generate operation permission information in the control module according to the comparison result, display it in real time through the embedded instrument, encrypt the data transmitted by the communication module, monitor the system status in real time, schedule system resources, and record and store identity authentication and interaction data.

3. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 2 is characterized in that: The S3 specifically includes: S31. Perform multi-scale feature extraction on the face image after geometric transformation. Use the single aggregation structure of the VoVNet network to establish a cross-layer connection mechanism for feature information at different levels, extract local texture information through a deep convolution module, and fuse global features using a cross-scale aggregation strategy. S32. For the extracted multi-scale features, the dilated convolution technique is used to expand the features, and convolution kernels with different dilation rates are introduced to expand the receptive field. The model learns key facial features at different scales. S33, combined with the self-attention mechanism of the ETSformer module, uses multi-head self-attention to calculate the global correlation between facial feature points, calculates feature weights in different feature spaces, emphasizes the feature contribution of key areas, and maintains spatial structure information through position encoding technology; S34, performing nonlinear transformation on the features optimized by the ETSformer module, using normalization strategy to enhance feature contrast, and adjusting the distribution of feature vectors; S35. Perform feature mapping conversion on the finally extracted features, use the embedding space mapping method to project the high-dimensional features into the low-dimensional identity feature space, and output the optimized features.

4. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 2 is characterized in that: The S4 specifically includes: S41. Initialize parameters based on artificial bee colony optimization algorithm and set feature vector set ,in Represents the i-th feature vector, setting the initial candidate feature subset , Represents the jth candidate feature, defines the initial population size N, sets the number of iterations T and the fitness calculation function, sets the optimization objective function J(S), reflects the overall fitness of the entire feature subset S, and calculates the initial weight of each feature : ; in, represents the i-th candidate feature, Representative features The weight is not a fixed value and is constantly updated. The value of J(S) will change with change, Representation characteristics The similarity between the target feature set T; S42, the hired bee stage, through feature weight calculation, feature contribution calculation and feature distribution balance calculation, screens discriminative features, removes redundant features, and covers a variety of feature information; S43, the scout bee stage uses an improved chaotic mapping strategy for global search, combining feature subset fitness calculation and global feature distribution calculation to expand the search range and avoid local optimality; S44, follow the bee stage using improved The search strategy is locally optimized and combined with the local feature balance metric to make the feature subsets reasonably distributed in all feature dimensions; S45, calculating the fitness of the newly generated feature subset, screening the feature subset based on the fitness, normalizing the selected optimal feature subset, obtaining the final feature vector, and generating the identity recognition result.

5. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 4 is characterized in that: The S42 specifically includes: S421, calculate the feature weight of each individual after iterative optimization , introduce feature sparsity regularization: ; Among them, m is the total number of feature subsets, n is the total number of feature vectors, The initial weight of each feature Start calculating iterations, and the weight after final iteration optimization is recorded as Features In the subset The matching degree in is the feature sparsity control parameter; S422. Calculate the feature contribution and construct the feature screening objective function: ; in, Features The distribution density in the candidate subset, , To adjust parameters adaptively; S423. Calculate the feature distribution balance as the initial measure of global feature balance: ； in, Representation characteristics The probability distribution in the subset S, log is the logarithmic function.

6. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 4 is characterized in that: The S43 specifically includes: S431, introduce improved chaotic mapping search to generate new feature subsets: ; in, is the global search control parameter, is the disturbance factor, combined with the sine chaotic mapping, sin is the sine function; S432, calculate the fitness of the feature subset, for the newly generated feature subset , calculate its fitness: ; in, Select sensitivity parameters for features; S433, optimize the global feature distribution, for the new feature subset , calculate the global feature balance metric: ; in, is the characteristic equalization control parameter.

7. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 4 is characterized in that: The S44 specifically includes: S441, local optimization search strategy, introduce improvements Search and optimize feature subsets : ; in, is the optimized feature subset, is the step size factor, is the search control parameter, t is the current iteration number, use Distribution, the mathematical form is as follows: ; in, yes The exponential parameter of the distribution, is the gamma function, x represents The generated random variable, sin is the sine function; S442, reference feature subset fitness Select appropriate feature subsets for balance optimization and calculate local feature balance metrics: ; in, is the characteristic equalization control parameter.

8. The biometric adaptive instrument interaction system for industrial vehicle driving safety according to claim 4 is characterized in that: The S45 specifically includes: S451, calculate the final fitness , improve the fitness calculation method, adopt the weighted fusion model, and make the contribution of each metric more reasonable: ; in, , , , is the weighting coefficient; S452, calculate the newly generated feature subset The fitness of , defines the final screening target: ； in, represents the optimal feature subset finally screened out, Indicates that the return results in Get the feature subset with the maximum value; S453. Normalize the selected optimal feature subset to obtain a final feature vector and generate an identity recognition result.

Images