Industrial vehicle intelligent instrument control system and method based on multi-mode biological recognition

Abstract

The invention discloses an industrial vehicle intelligent instrument control system and method based on multi-mode biological recognition, and the method comprises the following steps: S1, collecting fingerprint data of a driver, and carrying out the preprocessing; s2, fingerprint texture features are extracted, and weighted coding is carried out on local texture distribution and a global topological structure; s3, constructing an identity verification window, and performing bidirectional attention comparison on the fingerprint feature vector and the fingerprint template; s4, judging whether the identity verification is passed or not according to the similarity score, and if the verification is passed, triggering a buzzer or a voice broadcast device to generate control feedback; s5, starting the vehicle, awakening display, loading an instrument information page, and writing an identity verification result into a system buffer area; and S6, synchronizing an identity verification result to a mobile terminal APP or an applet platform through the Bluetooth MCU chip. According to the invention, industrial vehicle intelligent control based on fingerprint identification is realized, and the identity verification accuracy, the operation safety and the management efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent human-computer interaction and biometrics, and in particular to an industrial vehicle intelligent instrument control system and method based on multimodal biometrics. Background Art

[0002] With the continuous development of industrial automation and smart logistics, industrial vehicles, as critical operational equipment, are widely used in high-frequency, high-intensity scenarios such as factory workshops, warehousing and logistics, and port terminals. Traditional industrial vehicles rely on mechanical keys or IC card swiping for vehicle activation and identity management. While these methods offer a certain degree of security and convenience, they have exposed a series of problems in long-term practical use. Due to their simple structure, mechanical keys are easily copied or counterfeited, posing a risk of theft or illegal activation. Lost keys require lock cylinder replacement or rekeying, increasing operating and maintenance costs. Meanwhile, while card unlocking has introduced radio frequency identification technology, improving management efficiency and access control, it still carries numerous risks, such as easy card loss, difficulty tracing access rights, high management costs, and difficulty verifying authorized persons.

[0003] Currently, some intelligent control solutions attempt to achieve identity authentication and vehicle control through methods such as facial recognition, IC card binding, and QR code scanning. However, these methods are limited by factors such as lighting, device configuration, and operator habits, resulting in high false positive rates, significant verification delays, and poor environmental adaptability. These methods struggle to meet the requirements of industrial-grade scenarios requiring high security, high response speeds, and strong environmental adaptability. Furthermore, traditional methods inherently lack the ability to record user behavior and bind identities. This makes it impossible to effectively correlate driver behavior with device status, posing challenges to safety oversight, responsibility allocation, and efficiency assessment.

[0004] Fingerprint recognition, a mature and highly reliable biometric technology, boasts advantages such as strong uniqueness, ease of use, and strong environmental adaptability. It has been widely deployed in finance, security, access control, and other fields. However, its application in industrial vehicles is still in its infancy, particularly in areas such as embedded control, Bluetooth communication, multimodal fusion, and real-time log transmission, where systematic integration and scenario-adaptive intelligent control methods are lacking. Most existing solutions remain at the stage of single-point authentication, failing to establish a closed-loop control mechanism from fingerprint recognition, identity comparison, control feedback, unlock response, information synchronization, and permission management, making it difficult to achieve end-to-end security control and intelligent management.

[0005] Furthermore, existing fingerprint recognition applications often rely on static template matching for identity verification, lacking dynamic feature modeling capabilities. This significantly reduces recognition accuracy when faced with complex textures, incomplete fingerprints, or contaminated images, and the system is unable to adjust matching strategies in real time based on user behavior. Furthermore, many fingerprint recognition systems lack deep integration with vehicle instrumentation, preventing simultaneous startup, interface wakeup, and operational data loading after fingerprint triggering. This limits the overall system interoperability and interactive experience.

[0006] Furthermore, the lack of an effective linkage mechanism between mobile devices and the local control units of industrial vehicles makes it difficult to remotely issue commands, manage identity registration, synchronize data, and trace driving behavior. This results in management's limited control over vehicle operations, severe data fragmentation, and inefficient system maintenance and scheduling. While some existing solutions incorporate Bluetooth communication interfaces, they remain incomplete in areas such as hierarchical permission management, encrypted data transmission, and identity binding logging, leading to security vulnerabilities and data synchronization delays.

[0007] Therefore, how to provide an industrial vehicle intelligent instrument control system and method based on multimodal biometrics is a problem that those skilled in the art urgently need to solve. Summary of the Invention

[0008] One purpose of the present invention is to propose an industrial vehicle intelligent instrument control system and method based on multimodal biometrics. The present invention fully integrates multi-layer convolution extraction of fingerprint images, attention mechanism weighted coding, Bluetooth MCU identity comparison, CAN protocol signal linkage and mobile terminal information synchronization technology, and describes in detail the complete closed-loop process from fingerprint collection and identity authentication to vehicle control and data feedback. It has the advantages of high recognition accuracy, strong security level, fast control response and high management convenience.

[0009] The method for controlling an industrial vehicle intelligent instrument based on multimodal biometrics according to an embodiment of the present invention includes the following steps: S1. Collect the driver's fingerprint data through the sensor and pre-process it; S2. Based on the preprocessed fingerprint data, multi-layer convolution operations are used to extract fingerprint texture features, and the local texture distribution and global topological structure are weightedly encoded by the fusion attention mechanism to generate a fingerprint feature vector; S3. Use the Bluetooth MCU chip to build an authentication window, perform a two-way attention comparison between the fingerprint feature vector and the registered fingerprint template in the ferroelectric memory, and output a similarity score; S4. Determine whether the identity authentication is passed based on the similarity score. If the authentication is passed, trigger the buzzer or voice broadcast device to generate control feedback, and send an unlock signal to the industrial vehicle instrument terminal via the CAN communication protocol; S5. After receiving the unlock signal, the industrial vehicle instrument terminal starts the vehicle, wakes up the display, loads the instrument information page, and writes the authentication result into the system buffer; S6. Synchronize the authentication results to the mobile APP or mini-program platform via the Bluetooth MCU chip, update the vehicle status information and generate a driver behavior log, realizing a closed-loop multimodal biometric control process.

[0010] Optionally, the fingerprint data includes ridge line, valley line, endpoint and bifurcation point information.

[0011] Optionally, the preprocessing includes denoising, normalization, image enhancement and region segmentation.

[0012] Optionally, the fingerprint template includes corresponding fingerprint index information and is written into the ferroelectric memory via the SPI bus, along with a record of each identity authentication process.

[0013] Optionally, the mobile APP or mini program platform displays corresponding interface content according to the driver's authority identification, and allows the vehicle administrator to remotely issue fingerprint registration instructions, fingerprint deletion instructions, RTC setting instructions and vehicle usage record reading instructions.

[0014] Optionally, the S2 specifically includes: S21, inputting the preprocessed fingerprint data into a convolutional neural network, performing a first-layer convolution operation, extracting basic texture edge information and generating a first-dimensional feature map; S22, performing a second-layer convolution operation and a maximum pooling operation on the first-dimensional feature map, extracting the middle-layer texture change pattern and compressing feature redundant information, to generate a second-dimensional feature map; S23. Based on the second-dimensional feature map, a multi-head attention mechanism is introduced to calculate the correlation weight between each feature region and the global texture distribution to form a local texture attention map; S24. Based on the local texture attention map, a topological structure association matrix is ​​constructed, and the multi-scale texture features and spatial position relationships are integrated through weighted convolution coding operations to generate a fingerprint feature vector.

[0015] Optionally, the S3 specifically includes: S31, transmit the fingerprint feature vector to the Bluetooth MCU chip through serial communication, initialize the authentication window in the chip and activate the authentication status flag; S32, retrieve the registered fingerprint template from the ferroelectric memory, load it into the local fingerprint comparison buffer, and perform dimension alignment and structural normalization processing on the fingerprint feature vector and the fingerprint template; S33. Construct a bidirectional attention mechanism in parallel in the Bluetooth MCU chip. In a first direction, an attention weight distribution pointing to the template feature is generated based on the input fingerprint feature vector. In a second direction, an attention weight distribution pointing to the input fingerprint feature vector is generated based on the template feature. S34, fusing the bidirectional attention weights, constructing a cross attention matrix, and performing feature similarity calculation based on it to extract local matching patterns and global relative difference features; S35. Input the feature similarity calculation result into the scoring function to generate a similarity score representing identity consistency, and output it to the control process for identity authentication judgment.

[0016] Optionally, the S4 specifically includes: S41, receiving the similarity score output by the Bluetooth MCU chip, and comparing the value with the set threshold value to generate an identity authentication result identifier; S42. When the identity verification result is marked as passed, the Bluetooth MCU chip is controlled to output a trigger signal, driving the buzzer to generate continuous short sounds, or driving the voice broadcast device to play a voice message indicating that the identity verification is successful; S43, constructing a data frame including the identity authentication result identifier, the fingerprint index number and the time tag, encapsulating the instruction through the CAN communication protocol and setting the sending priority; S44. Upload the encapsulated data frame to the sending queue, which is represented as an unlocking signal, and send it to the industrial vehicle instrument terminal through the bus to implement the control instruction issuance and instruction confirmation process.

[0017] Optionally, the S5 specifically includes: S51, the industrial vehicle instrument terminal receives the unlocking signal sent by the Bluetooth MCU chip through the bus, parses the signal content, and extracts the identity verification result identifier, fingerprint index number, and time information; S52. Under the instruction of the unlocking signal, the power management circuit is controlled to output a start relay control instruction to complete the main power connection and control the vehicle start process; S53, activating the LCD display drive circuit, completing the screen wake-up operation, and synchronously loading the instrument information page containing the vehicle's operating status, power, voltage, and real-time parameters; S54. Write the identity authentication result identifier into the meter terminal system buffer as structured data storage content of behavior identification and usage records.

[0018] The AGI-based multimodal holographic light field interaction system according to an embodiment of the present invention includes: A data processing module, used to collect the driver's fingerprint data through the sensor and perform pre-processing; The feature extraction module is used to extract fingerprint texture features using multi-layer convolution operations and perform weighted encoding of local texture distribution and global topological structure through the fusion attention mechanism to generate fingerprint feature vectors; The comparison calculation module is used to build an authentication window using the Bluetooth MCU chip, perform a two-way attention comparison between the fingerprint feature vector and the registered fingerprint template in the ferroelectric memory, and output a similarity score; The authentication module is used to determine whether the authentication is passed based on the similarity score. If the authentication is passed, it triggers the buzzer or voice broadcast device to generate control feedback and sends an unlock signal to the industrial vehicle instrument terminal through the CAN communication protocol; The unlock execution module is used to start the vehicle, wake up the display, load the instrument information page, and write the authentication result into the system buffer after the industrial vehicle instrument terminal receives the unlock signal; The update control module is used to synchronize the authentication results to the mobile app or mini-program platform via the Bluetooth MCU chip, update the vehicle status information and generate a driver behavior log, realizing a closed-loop multimodal biometric control process.

[0019] The beneficial effects of the present invention are: First, the present invention significantly improves the accuracy and robustness of fingerprint recognition by introducing multi-layer convolution operations and attention mechanisms to perform deep feature extraction and weighted encoding of fingerprint images. In particular, it can still achieve stable and reliable identity recognition in industrial environments such as complex lighting, unclear images or incomplete fingerprints, breaking through the limitations of traditional template matching methods in texture modeling and spatial expression.

[0020] Secondly, the authentication mechanism built by combining the Bluetooth MCU chip and ferroelectric memory realizes two-way attention comparison and similarity scoring, ensuring that the authentication process is both real-time and secure. It cooperates with the CAN communication protocol to link the vehicle instrument terminal, effectively opening up the "identification-verification-control" link, and improving the closed-loop efficiency and response speed of the control process.

[0021] Finally, by closely linking the identity authentication results with vehicle startup, instrument wake-up, driving behavior recording and other operations, and synchronizing the results to the mobile mini-program or APP through Bluetooth communication, the driver's behavior can be traced, the authority can be controlled, and the vehicle status can be visually managed, which significantly improves the safety, intelligence and management convenience of industrial vehicles, and meets the core demands of modern industrial scenarios for efficient, precise and safe control. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] 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 This is a flow chart of the industrial vehicle intelligent instrument control method based on multimodal biometrics proposed by the present invention; Figure 2 This is a schematic diagram of the identity authentication process of the industrial vehicle intelligent instrument control method based on multimodal biometrics proposed in the present invention; Figure 3 This is a schematic diagram of the fingerprint registration and verification process of the industrial vehicle intelligent instrument control method based on multimodal biometrics proposed in the present invention; Figure 4 This is a schematic diagram of the fingerprint controller principle of the industrial vehicle intelligent instrument control method based on multimodal biometrics proposed in the present invention; Figure 5 This is a system architecture diagram of the industrial vehicle intelligent instrument control method based on multimodal biometrics proposed in the present invention; Figure 6 This is a module structure diagram of the industrial vehicle intelligent instrument control system based on multimodal biometrics proposed in the present invention. DETAILED DESCRIPTION

[0023] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0024] refer to Figure 1-4 , an industrial vehicle intelligent instrument control method based on multimodal biometrics, comprising the following steps: S1. Collect the driver's fingerprint data through the sensor and pre-process it; S2. Based on the preprocessed fingerprint data, multi-layer convolution operations are used to extract fingerprint texture features, and the local texture distribution and global topological structure are weightedly encoded by the fusion attention mechanism to generate a fingerprint feature vector; S3. Use the Bluetooth MCU chip to build an authentication window, perform a two-way attention comparison between the fingerprint feature vector and the registered fingerprint template in the ferroelectric memory, and output a similarity score; S4. Determine whether the identity authentication is passed based on the similarity score. If the authentication is passed, trigger the buzzer or voice broadcast device to generate control feedback, and send an unlock signal to the industrial vehicle instrument terminal via the CAN communication protocol; S5. After receiving the unlock signal, the industrial vehicle instrument terminal starts the vehicle, wakes up the display, loads the instrument information page, and writes the authentication result into the system buffer; S6. Synchronize the authentication results to the mobile APP or mini-program platform via the Bluetooth MCU chip, update the vehicle status information and generate a driver behavior log, realizing a closed-loop multimodal biometric control process.

[0025] By constructing a complete multimodal biometric control process, the present invention realizes an end-to-end control closed loop from fingerprint collection, feature extraction, identity verification to vehicle unlocking and status update, thereby improving the safety, response speed and identity management capabilities of industrial vehicle operations.

[0026] In this embodiment, the fingerprint data is formed by pre-processing the grayscale image collected by the sensor, specifically including the ridge trend map, valley line spacing map, feature point distribution map, and so on. Figure 3 Class structure data; among them, ridge information is used to describe the direction and thickness characteristics of continuous raised areas in the fingerprint image, valley information is used to extract the spacing pattern and topological characteristics of the concave area, endpoints are used to locate the natural interruption position of the ridge line, and bifurcation points are used to identify the split nodes of the ridge line direction. By extracting the above key structural points, a unique feature coordinate set is formed, and the feature coordinate set is used for subsequent multi-layer convolution operations.

[0027] The present invention utilizes rich feature information such as ridges, valleys, endpoints and bifurcation points to model texture structures, thereby enhancing the expression dimension of fingerprint images and individual recognition accuracy, and significantly improving the robustness and accuracy of fingerprint recognition.

[0028] In this embodiment, the preprocessing process is executed immediately after the fingerprint data collection is completed, and the fingerprint data is input into the preprocessing engine in the form of a serial data stream, including denoising, normalization, image enhancement and region segmentation. The denoising process adopts a joint strategy based on bilateral filtering and median filtering to filter out local impulse noise and high-frequency pseudo-edge structure. The normalization process concentrates the image grayscale distribution in a stable range through pixel value stretching and image histogram equalization. The image enhancement process adopts a directional enhancement filter based on frequency domain transformation to improve the clarity of the ridge direction texture, and performs local contrast enhancement in the spatial domain to highlight the key detail area. The region segmentation adopts a threshold layering strategy and an edge adaptive blocking mechanism to perform structural stripping processing on the central feature area and the edge noise area.

[0029] The present invention effectively improves the quality and structural clarity of fingerprint images through preprocessing methods such as denoising, normalization, image enhancement and region segmentation, provides a stable data basis for subsequent feature extraction and recognition, and ensures the overall performance of the system.

[0030] In this embodiment, the fingerprint template is a structured coding file, which is stored in the ferroelectric memory and is bound one-to-one with the fingerprint index information. The fingerprint template is encoded by the Bluetooth MCU chip during the user fingerprint registration process. First, a fingerprint embedding vector is generated through multi-layer convolution and attention weighted strategy, and then a unique index identifier is generated according to the registration sequence number, and the fingerprint embedding vector and the unique index identifier are data encapsulated; the encapsulated template data is written to the ferroelectric memory through the SPI bus, and the template data is called to perform comparison with the input feature vector during each identity authentication process; at the same time, after each verification is completed, the Bluetooth MCU chip will encapsulate and record the comparison time, similarity score, operation behavior identifier and verification status, and append them to the extended record segment corresponding to the fingerprint template to establish a complete historical operation track.

[0031] The present invention writes the fingerprint template and the identity authentication record into the ferroelectric memory together, thereby realizing the binding storage of identity information and behavior trajectory, enhancing the traceability and auditing capability of operation behavior, and supporting subsequent security review and behavior analysis.

[0032] In this embodiment, the mobile APP or mini-program platform establishes a pairing communication channel with the Bluetooth MCU chip to synchronize the vehicle status and fingerprint recognition results in real time, and loads the interface logic according to the permission identifier in the driver's identity information. The permission identifier is represented by a binary permission code and is divided into three levels: ordinary driver, vehicle administrator and operation and maintenance personnel; when the driver logs in to the mobile terminal, the platform automatically loads different function pages according to his or her permissions. The ordinary driver interface only supports viewing vehicle status and task information. The vehicle administrator interface loads the fingerprint registration management interface, fingerprint deletion entry, RTC calibration setting interface and vehicle usage log reading window. The operation and maintenance personnel interface supports temporary command issuance and equipment operation status inspection; the fingerprint registration and deletion instructions generate and issue control frames through the platform, which are transmitted to the Bluetooth MCU chip through the Bluetooth channel. After parsing, template storage, template erasure and other operations are executed, and the results are transmitted back to the mobile terminal for status confirmation, building an efficient permission control and operation and maintenance response system.

[0033] This invention supports administrators to remotely issue registration and deletion instructions and read operation records through APP or mini-programs, builds a permission classification and remote collaborative management mechanism, significantly simplifies the operation process and improves management efficiency and flexibility.

[0034] In this embodiment, S2 specifically includes: S21, inputting the preprocessed fingerprint data into a convolutional neural network, performing a first-layer convolution operation, extracting basic texture edge information and generating a first-dimensional feature map; S22, performing a second-layer convolution operation and a maximum pooling operation on the first-dimensional feature map, extracting the middle-layer texture change pattern and compressing feature redundant information, to generate a second-dimensional feature map; S23. Based on the second-dimensional feature map, a multi-head attention mechanism is introduced to calculate the correlation weight between each feature region and the global texture distribution to form a local texture attention map; S24. Based on the local texture attention map, a topological structure association matrix is ​​constructed, and the multi-scale texture features and spatial position relationships are integrated through weighted convolution coding operations to generate a fingerprint feature vector.

[0035] The present invention adopts a multi-layer convolutional network combined with an attention mechanism to perform multi-scale texture feature modeling on fingerprint images, realizes the fusion expression of local and global textures, and enhances the discrimination ability and feature expression integrity of the recognition system.

[0036] In this embodiment, S3 specifically includes: S31, transmit the fingerprint feature vector to the Bluetooth MCU chip through serial communication, initialize the authentication window in the chip and activate the authentication status flag; S32, retrieve the registered fingerprint template from the ferroelectric memory, load it into the local fingerprint comparison buffer, and perform dimension alignment and structural normalization processing on the fingerprint feature vector and the fingerprint template; S33. Construct a bidirectional attention mechanism in parallel in the Bluetooth MCU chip. In a first direction, an attention weight distribution pointing to the template feature is generated based on the input fingerprint feature vector. In a second direction, an attention weight distribution pointing to the input fingerprint feature vector is generated based on the template feature. S34, fusing the bidirectional attention weights, constructing a cross attention matrix, and performing feature similarity calculation based on it to extract local matching patterns and global relative difference features; S35. Input the feature similarity calculation result into the scoring function to generate a similarity score representing identity consistency, and output it to the control process for identity authentication judgment.

[0037] The present invention designs a bidirectional attention mechanism and a cross-attention matrix to realize high-dimensional, multi-angle dynamic comparison between fingerprint feature vectors and templates, significantly improving the accuracy, adaptability and intelligence of identity authentication.

[0038] In this embodiment, the S4 specifically includes: S41, receiving the similarity score output by the Bluetooth MCU chip, and comparing the value with the set threshold value to generate an identity authentication result identifier; S42. When the identity verification result is marked as passed, the Bluetooth MCU chip is controlled to output a trigger signal, driving the buzzer to generate continuous short sounds, or driving the voice broadcast device to play a voice message indicating that the identity verification is successful; S43, constructing a data frame including the identity authentication result identifier, the fingerprint index number and the time tag, encapsulating the instruction through the CAN communication protocol and setting the sending priority; S44. Upload the encapsulated data frame to the sending queue, which is represented as an unlocking signal, and send it to the industrial vehicle instrument terminal through the bus to implement the control instruction issuance and instruction confirmation process.

[0039] The present invention performs identity authentication judgment by setting a threshold, and executes unlocking instructions in combination with buzzer or voice feedback and CAN communication protocol, thereby realizing a high-security, low-latency physical linkage control mechanism and enhancing system responsiveness and user interaction experience.

[0040] In this embodiment, the S5 specifically includes: S51, the industrial vehicle instrument terminal receives the unlocking signal sent by the Bluetooth MCU chip through the bus, parses the signal content, and extracts the identity verification result identifier, fingerprint index number, and time information; S52. Under the instruction of the unlocking signal, the power management circuit is controlled to output a start relay control instruction to complete the main power connection and control the vehicle start process; S53, activating the LCD display drive circuit, completing the screen wake-up operation, and synchronously loading the instrument information page containing the vehicle's operating status, power, voltage, and real-time parameters; S54. Write the identity authentication result identifier into the meter terminal system buffer as structured data storage content of behavior identification and usage records.

[0041] The present invention enables the instrument terminal to automatically complete vehicle startup, information loading and status writing operations after receiving the unlocking signal, realizes the deep integration of fingerprint recognition and vehicle control process, and improves startup efficiency and information management consistency.

[0042] refer to Figure 5-6 , an intelligent instrument control system for industrial vehicles based on multimodal biometrics, including: A data processing module, used to collect the driver's fingerprint data through the sensor and perform pre-processing; The feature extraction module is used to extract fingerprint texture features using multi-layer convolution operations and perform weighted encoding of local texture distribution and global topological structure through the fusion attention mechanism to generate fingerprint feature vectors; The comparison calculation module is used to build an authentication window using the Bluetooth MCU chip, perform a two-way attention comparison between the fingerprint feature vector and the registered fingerprint template in the ferroelectric memory, and output a similarity score; The authentication module is used to determine whether the authentication is passed based on the similarity score. If the authentication is passed, it triggers the buzzer or voice broadcast device to generate control feedback and sends an unlock signal to the industrial vehicle instrument terminal through the CAN communication protocol; The unlock execution module is used to start the vehicle, wake up the display, load the instrument information page, and write the authentication result into the system buffer after the industrial vehicle instrument terminal receives the unlock signal; The update control module is used to synchronize the authentication results to the mobile app or mini-program platform via the Bluetooth MCU chip, update the vehicle status information and generate a driver behavior log, realizing a closed-loop multimodal biometric control process.

[0043] The present invention constructs a multi-module collaborative system that integrates data acquisition, feature extraction, identity authentication, control execution and mobile terminal synchronization, comprehensively improving the intelligence level, data closed-loop capability and overall security of industrial vehicle control.

[0044] Example 1: To verify the feasibility of this invention, we applied it to the control of industrial vehicles used for material handling and transfer operations within a smart manufacturing enterprise. In this scenario, vehicles are frequently used, personnel turnover is high, operating areas are dispersed, and some work environments are subject to complex conditions such as noise interference and insufficient light. Traditional mechanical key control methods suffer from issues such as duplication, difficulty in unified management, and high loss rates. Card swipe control also faces drawbacks such as easy card borrowing, difficulty in tracking permissions, and cumbersome management processes, making it unable to meet the security control requirements of high-intensity, high-frequency industrial operations.

[0045] To address these issues, the multimodal biometrics-based industrial vehicle intelligent instrument control method of the present invention was deployed on six electric traction industrial vehicles. Each vehicle's instrument is equipped with an integrated Bluetooth MCU authentication unit, fingerprint recognition module, buzzer feedback unit, and CAN communication interface, and is remotely managed and operated via a supporting app. All drivers are required to complete fingerprint registration. The system verifies the driver's identity through fingerprint recognition when the vehicle is first started. Once the verification is successful, the unlock command is automatically executed, the instrument terminal wakes up in real time, and the identity data is transmitted back to the management platform, generating a log record.

[0046] During actual operation, the system operated stably, with an average fingerprint verification response time of 0.87 seconds and a fingerprint recognition accuracy rate of 98.2%. It also maintained good recognition capabilities even when the operating fingers were slightly oily or sweaty. Compared with the 3.5-second verification delay of the original card swiping system, this method significantly improved the response speed. During the high-frequency operation period, the number of fingerprint recognition calls reached 1,064 times per day, with no false or missed recognitions. The system automatically uploads the identity authentication results to the management terminal and generates driver behavior log information daily, including identity index, start time, operating time, usage period and task type, providing management personnel with complete and traceable analytical data.

[0047] Furthermore, the system's remote fingerprint authorization issuance and deletion operations are completed within 3 seconds. Vehicle administrators can flexibly configure driving permissions based on their shift schedules, significantly reducing manual operation and management costs. During a 3-week verification period, approximately 27 manual repair reports due to lost keys, damaged cards, and other issues were reduced, saving over 43 hours of management and repair processing time. The instrument panel quickly loads the operating information page, and the real-time vehicle status monitoring refresh rate reaches 2Hz. The entire driver-initiated operation process averages less than 5 seconds, significantly improving the user experience and operational efficiency.

[0048] Table 1 Performance comparison data table ; It can be seen from the above comparative data that the present invention is superior to traditional control methods in terms of identity recognition accuracy, control response speed, data synchronization efficiency, management operation flexibility and system stability. It is suitable for high-frequency, complex, and multi-personnel collaborative industrial operation scenarios, and can effectively solve the problems of low recognition efficiency, high management costs, and insufficient security in traditional solutions. It has significant promotion value and engineering application significance.

[0049] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. An industrial vehicle intelligent instrument control method based on multimodal biometrics, characterized in that: The steps include: S1. Collecting the driver's fingerprint data through a sensor and preprocessing it. The fingerprint data includes ridge, valley, endpoint, and bifurcation point information. The preprocessing includes denoising, normalization, image enhancement, and region segmentation. S2. Based on the preprocessed fingerprint data, multi-layer convolution operations are used to extract fingerprint texture features, and the local texture distribution and global topological structure are weightedly encoded by the fusion attention mechanism to generate a fingerprint feature vector; S3. Use the Bluetooth MCU chip to build an authentication window, perform a two-way attention comparison between the fingerprint feature vector and the registered fingerprint template in the ferroelectric memory, and output a similarity score. The fingerprint template contains the corresponding fingerprint index information and is written to the ferroelectric memory via the SPI bus, along with a record of each authentication process. S4. Determine whether the identity authentication is passed based on the similarity score. If the authentication is passed, trigger the buzzer or voice broadcast device to generate control feedback, and send an unlock signal to the industrial vehicle instrument terminal via the CAN communication protocol; S5. After receiving the unlock signal, the industrial vehicle instrument terminal starts the vehicle, wakes up the display, loads the instrument information page, and writes the authentication result into the system buffer; S6. Synchronize the identity authentication result to the mobile APP or mini-program platform via the Bluetooth MCU chip, update the vehicle status information and generate a driver behavior log, realizing a closed-loop multimodal biometric control process. The mobile APP or mini-program platform displays the corresponding interface content according to the driver's authority identifier, and allows the vehicle administrator to remotely issue fingerprint registration instructions, fingerprint deletion instructions, RTC setting instructions and vehicle usage record reading instructions.

2. The industrial vehicle intelligent instrument control method based on multimodal biometrics according to claim 1 is characterized in that: The S2 specifically includes: S21, inputting the preprocessed fingerprint data into a convolutional neural network, performing a first-layer convolution operation, extracting basic texture edge information and generating a first-dimensional feature map; S22, performing a second-layer convolution operation and a maximum pooling operation on the first-dimensional feature map, extracting the middle-layer texture change pattern and compressing feature redundant information, to generate a second-dimensional feature map; S23. Based on the second-dimensional feature map, a multi-head attention mechanism is introduced to calculate the correlation weight between each feature region and the global texture distribution to form a local texture attention map; S24. Based on the local texture attention map, a topological structure association matrix is ​​constructed, and the multi-scale texture features and spatial position relationships are integrated through weighted convolution coding operations to generate a fingerprint feature vector.

3. The industrial vehicle intelligent instrument control method based on multimodal biometrics according to claim 1 is characterized in that: The S3 specifically includes: S31, transmit the fingerprint feature vector to the Bluetooth MCU chip through serial communication, initialize the authentication window in the chip and activate the authentication status flag; S32, retrieve the registered fingerprint template from the ferroelectric memory, load it into the local fingerprint comparison buffer, and perform dimension alignment and structural normalization processing on the fingerprint feature vector and the fingerprint template; S33. Construct a bidirectional attention mechanism in parallel in the Bluetooth MCU chip. In a first direction, an attention weight distribution pointing to the template feature is generated based on the input fingerprint feature vector. In a second direction, an attention weight distribution pointing to the input fingerprint feature vector is generated based on the template feature. S34, fusing the bidirectional attention weights, constructing a cross attention matrix, and performing feature similarity calculation based on it to extract local matching patterns and global relative difference features; S35. Input the feature similarity calculation result into the scoring function to generate a similarity score representing identity consistency, and output it to the control process for identity authentication judgment.

4. The industrial vehicle intelligent instrument control method based on multimodal biometrics according to claim 1 is characterized in that: The S4 specifically includes: S41, receiving the similarity score output by the Bluetooth MCU chip, and comparing the value with the set threshold value to generate an identity authentication result identifier; S42. When the identity verification result is marked as passed, the Bluetooth MCU chip is controlled to output a trigger signal, driving the buzzer to generate continuous short sounds, or driving the voice broadcast device to play a voice message indicating that the identity verification is successful; S43, constructing a data frame including the identity authentication result identifier, the fingerprint index number and the time tag, encapsulating the instruction through the CAN communication protocol and setting the sending priority; S44. Upload the encapsulated data frame to the sending queue, which is represented as an unlocking signal, and send it to the industrial vehicle instrument terminal through the bus to implement the control instruction issuance and instruction confirmation process.

5. The industrial vehicle intelligent instrument control method based on multimodal biometrics according to claim 1 is characterized in that: The S5 specifically includes: S51, the industrial vehicle instrument terminal receives the unlocking signal sent by the Bluetooth MCU chip through the bus, parses the signal content, and extracts the identity verification result identifier, fingerprint index number, and time information; S52. Under the instruction of the unlocking signal, the power management circuit is controlled to output a start relay control instruction to complete the main power connection and control the vehicle start process; S53, activating the LCD display drive circuit, completing the screen wake-up operation, and synchronously loading the instrument information page containing the vehicle's operating status, power, voltage, and real-time parameters; S54. Write the identity authentication result identifier into the meter terminal system buffer as structured data storage content of behavior identification and usage records.

6. An industrial vehicle intelligent instrument control system based on multimodal biometrics, implementing an industrial vehicle intelligent instrument control method based on multimodal biometrics according to any one of claims 1 to 5, characterized in that: include: A data processing module, used to collect the driver's fingerprint data through the sensor and perform pre-processing; The feature extraction module is used to extract fingerprint texture features using multi-layer convolution operations and perform weighted encoding of local texture distribution and global topological structure through the fusion attention mechanism to generate fingerprint feature vectors; The comparison calculation module is used to build an authentication window using the Bluetooth MCU chip, perform a two-way attention comparison between the fingerprint feature vector and the registered fingerprint template in the ferroelectric memory, and output a similarity score; The authentication module is used to determine whether the authentication is passed based on the similarity score. If the authentication is passed, it triggers the buzzer or voice broadcast device to generate control feedback and sends an unlock signal to the industrial vehicle instrument terminal through the CAN communication protocol; The unlock execution module is used to start the vehicle, wake up the display, load the instrument information page, and write the authentication result into the system buffer after the industrial vehicle instrument terminal receives the unlock signal; The update control module is used to synchronize the authentication results to the mobile app or mini-program platform via the Bluetooth MCU chip, update the vehicle status information and generate a driver behavior log, realizing a closed-loop multimodal biometric control process.

Images