EEG-based train emergency braking control system and method

Abstract

This invention discloses a train emergency braking control system and method based on electroencephalograms (EEGs). It collects real-time EEG signals from the driver's left frontal lobe to assist in emergency braking. An EEG-enabled smart helmet integrating an EEG module, microprocessor, and Bluetooth module is used to collect driver stress-induced EEG signals during train operation. Wavelet decomposition is employed to process these signals, and based on wavelet decomposition coefficients and combined with time, frequency, and entropy domain characteristics, the system extracts the characteristic signals that indicate the driver's decision to press the emergency brake button. Using online training and cloud-edge collaboration, the system updates the feature extraction parameters and classification criteria used in the microprocessor based on the driver's real-time EEG signals and historical data from the cloud. Different users can set different parameters to address individual differences in the characteristic signals during driver stress responses. Based on these characteristic signals, the system determines the driver's decision-making intent in real time, thereby improving the reaction speed and safety in handling emergencies.

Description

Technical Field

[0001] This invention relates to the field of auxiliary train braking technology, specifically to a train emergency braking control system and method based on electroencephalogram (EEG). Background Technology

[0002] As train speeds continue to increase, the train driver's reaction time in emergency situations has become a key factor affecting train safety. In sudden situations such as strong winds, heavy rain, blizzards, or obstructions, train drivers need to react quickly and take emergency braking measures to avoid accidents. However, a driver's reaction speed and judgment are affected by various factors, including physiological and psychological states. Especially under stress or fatigue, the reaction time to trigger the emergency braking button increases, thereby raising the risk of a collision. Electroencephalogram (EEG) signals, as a direct manifestation of brain activity, can reflect a driver's emotions, cognitive state, and stress response. Real-time analysis and processing of EEG signals can predict a train driver's intentions before they are completed, thus providing a faster response time for emergency braking.

[0003] One of the prior art solutions (CN104802800B), the closest prior art to this invention application, discloses a driver state recognition system based on steering wheel detection. This system includes a steering wheel motion sensor, an EEG detection helmet, a wireless communication interface, and a Freescale IMX6 processor. The steering wheel motion sensor detects the motion state of the car steering wheel, and the EEG detection helmet detects the driver's brainwave characteristics. The IMX6 processor is connected to the steering wheel motion sensor, the EEG detection helmet, and the wireless communication interface. Based on the motion state, it determines whether to initiate the EEG detection helmet's detection operation. After initiating the detection operation, it determines the driver's current driving state based on the analysis results of the brainwave characteristics and sends the current driving state to the wireless communication interface. This invention enables the analysis of the driver's state based on the determination of the car steering wheel's motion state, thereby reducing unnecessary system power consumption while ensuring detection efficiency.

[0004] One existing technical solution (CN107015489B) provides a vehicle braking control method and system. The method includes the following steps: acquiring the driver's brainwave signals, processing and recognizing the brainwave signals to obtain the driver's braking intention signal; receiving the braking intention signal and processing it to obtain a vehicle braking command; and controlling the vehicle's braking according to the braking command. This invention's method can quickly control the vehicle's braking based on the driver's brainwave signals, effectively reducing the driver's manual reaction time, shortening the operation execution time, and effectively solving the problem of mistakenly pressing the accelerator instead of the brake, thus improving driving safety.

[0005] A patent (patent number: 202311403600.8) integrates automatic braking into smart glasses, proposing a car braking system based on brainwave smart glasses. This system provides a car braking system based on brainwave smart glasses, which includes a brainwave TGAM module, an STM32F103C8T6 microprocessor module, and a Bluetooth module. The TGAM module collects brainwave data and transmits it to the STM32F103C8T6 microprocessor. The STM32F103C8T6 microprocessor processes and monitors the brainwave data; when abnormal brainwave data is detected, the microprocessor transmits this status to the vehicle's infotainment system via Bluetooth. This invention combines human brainwave data with automotive braking technology, which can prevent delayed braking and driving errors in many emergency scenarios.

[0006] Currently, some systems can monitor a driver's cognitive load and emotional state by collecting EEG signals. However, most of these systems are limited to simple signal monitoring and cannot accurately capture the driver's decision-making intentions under stress in real time under complex train operation scenarios. In addition, existing EEG data processing methods usually only focus on the single-dimensional features of EEG signals, failing to fully utilize the multi-dimensional features of the signals to improve recognition accuracy.

[0007] The limitations of existing technologies are: 1. There is currently no technology that links brainwaves to train braking. 2. Most existing technologies are limited to simple signal monitoring and cannot capture the train driver's decision-making intentions under stress in real time and accurately.

[0008] In view of the above-mentioned shortcomings of the existing technology, the present invention is proposed. Summary of the Invention

[0009] To address the shortcomings of existing technologies, the present invention aims to provide a train emergency braking control method and system based on brainwaves, a brainwave-based smart safety helmet structure, and a host computer. This method combines human brainwave data with train emergency braking to prevent train drivers from being unable to trigger the emergency braking button in a timely and effective manner during sudden situations such as strong winds, heavy rain, blizzards, and foreign object intrusion.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] A train emergency braking control system based on brainwaves includes a smart safety helmet worn by the train driver that monitors the driver's brainwaves in real time, and a main unit that is wiredly connected to the train traction braking control system.

[0012] The smart EEG cap includes a circuit board, monitoring electrodes, reference electrodes, and adjustment rods;

[0013] The circuit board includes a TGAM (Transcranial Doppler) module, an STM32F103C8T6 microprocessor, and a JDY31 Bluetooth module.

[0014] The EEG TGAM module is used to collect the train driver's EEG data in real time and transmit the preprocessed data to the microprocessor STM32F103C8T6 via Bluetooth module.

[0015] The STM32F103C8T6 microprocessor is used to receive, process and monitor EEG data in real time. When the processed feature signal is successfully compared with the braking command feature signal stored in the database, the STM32F103C8T6 microprocessor sends a traction resection command to the host via the JDY31 Bluetooth module.

[0016] The JDY31 Bluetooth module is used to enable bidirectional data transmission between the EEG TGAM module and the STM32F103C8T6 microprocessor, as well as between the STM32F103C8T6 microprocessor and the host.

[0017] The monitoring electrodes are connected to the main circuit board. By adjusting the deformation of the adjustment rod, the position and tightness of the monitoring electrodes against the left frontal pole of the train driver's brain can be adjusted to monitor the train driver's brainwave data in real time.

[0018] Reference electrodes are located on the left and right sides of the smart safety helmet to measure the potential of the left and right ears respectively, in order to remove interference from power frequency and physiological noise;

[0019] The host computer displays the train driver's brainwaves in real time and is wirelessly connected to the circuit motherboard via Bluetooth to receive real-time brainwave data and traction resection commands.

[0020] The host computer transmits the received traction cut-off command to the train traction and braking control system to realize automatic braking of the train in emergency situations.

[0021] Preferably, the host also has cloud-edge collaboration capabilities, enabling it to upload real-time collected EEG data to the cloud for analysis and optimization;

[0022] More preferably, the host computer dynamically adjusts the personalized settings of EEG characteristic signal parameters through the analysis of historical data in the cloud, thereby optimizing the generation and judgment process of traction commands, ensuring that the system can adapt to the individual differences in the driver's EEG, and improving the accuracy of traction removal in response to emergencies.

[0023] This invention also provides a train emergency braking control method based on electroencephalograms (EEGs), comprising the following steps:

[0024] Step S1: The train starts and the detection system begins operation;

[0025] The circuit board in the smart safety helmet has internal data interconnection and external Bluetooth connection with the host; the host is wired to the train traction and braking control system.

[0026] After the train starts, the driver puts on a safety helmet and adjusts the adjustment lever to adjust the position and tightness of the monitoring electrode of the main circuit board at the driver's forehead; the reference electrode of the main circuit board is connected to the driver's ears to ensure stable signal acquisition;

[0027] Step S2: The safety helmet collects brainwave data;

[0028] The smart helmet starts working, and the EEG TGAM module collects the driver's brainwave signals in real time and performs preliminary noise reduction processing.

[0029] Step S3: Bluetooth data transmission;

[0030] EEG data is transmitted to the STM32F103C8T6 microprocessor in real time via the JDY31 Bluetooth module.

[0031] Step S4: The microcontroller processes the EEG data;

[0032] The STM32F103C8T6 microprocessor performs wavelet decomposition on the received EEG signal and extracts the wavelet coefficients of the β and γ frequency bands.

[0033] The STM32F103C8T6 microcontroller extracts signal features from EEG signals, trains braking command data based on machine learning, and performs real-time EEG data processing and monitoring using online training and cloud-edge collaboration technologies. Specifically, this includes:

[0034] Step S41: Extract wavelet coefficients of the β and γ frequency bands of EEG in different scenarios from the EEG database through wavelet decomposition, and combine them with the time domain, frequency domain and entropy domain features of the original EEG data as input to the machine learning model;

[0035] Step S42: Use a random forest-based machine learning classification algorithm to perform semi-supervised learning on the data in the EEG signal database to extract the characteristic signals of EEG under stress scenarios, and convert the machine learning model trained by TensorFlow into a model suitable for the microprocessor STM32F103C8T6.

[0036] Step S5: Input the features in the time domain, frequency domain, wavelet domain, and entropy domain into the TensorFlow Lite model in the STM32F103C8T6 microprocessor to classify the EEG feature signals in the real-time scene.

[0037] Step 6: Based on the results of the model run in Step 5, determine whether the monitored EEG signal data is abnormal; compare it with the braking command feature signals in the database to determine whether the driver is in an emergency braking state.

[0038] If the data is abnormal, proceed to steps S7 and S8.

[0039] If the data is normal, return to step S5 to continue data monitoring;

[0040] Step S7: When an abnormal EEG signal of the driver is detected, that is, the characteristic signal matches the braking command characteristic signal in the database, the microprocessor STM32F103C8T6 sends a traction cut-off command to the host via the JDY31 Bluetooth module. The host transmits the traction cut-off command to the train traction and braking control system via wired transmission and immediately executes emergency braking.

[0041] Step S8: After the train driver restarts the equipment for monitoring, return to step 5 to re-monitor the data. The system will then resume receiving, processing, and monitoring the train driver's brainwaves.

[0042] in,

[0043] In step S42, the construction of the machine learning model trained by TensorFlow includes the following steps:

[0044] Step S421: Use the EEG TGAM module to collect EEG data of train drivers under normal driving and different stress scenarios, and construct an EEG signal database.

[0045] Step S422, data preprocessing stage: Wavelet decomposition method is used to extract wavelet domain features of EEG signals, and time domain, frequency domain and entropy domain features are combined to form a multidimensional feature vector;

[0046] Step S423: Train the EEG data using the random forest classification algorithm;

[0047] Step S424: By classifying the stress state and normal state data in the database, a model that can accurately identify the driver's stress state is trained.

[0048] Step S425: The model for identifying the driver's stress state in step S424 is performed in the TensorFlow framework. The final generated model is optimized and converted into TensorFlow Lite format suitable for the microprocessor STM32F103C8T6.

[0049] Step S426: The TensorFlow Lite format model generated in step S425 is synchronously updated online to the microprocessor STM32F103C8T6. The microprocessor STM32F103C8T6 stores the latest 1-second, 2-second, and 3-second EEG data in three circular buffers respectively, and uses wavelet decomposition to extract wavelet domain features of the three data segments with different time lengths.

[0050] In step S427, the system analyzes the real-time data based on the STM32F103C8T6 microprocessor and the cloud training results provided by the host, evaluates the three sets of features, and selects the data that best matches the user's EEG signal characteristics at that time as one of the inputs to the TensorFlow Lite format model in step S426.

[0051] Beneficial effects:

[0052] 1. Through real-time monitoring and analysis of brain waves, the system can automatically trigger the braking mechanism when the driver is in a state of stress, reducing human delay and ensuring that the train can quickly take emergency measures in dangerous situations;

[0053] 2. This invention utilizes wavelet decomposition technology, combining features from the frequency domain, time domain, and entropy domain, to accurately extract EEG feature signals of drivers under different stress scenarios. This multi-dimensional feature extraction method can effectively improve the system's accuracy in recognizing driver stress states and reduce the possibility of misjudgment.

[0054] 3. Through online training and cloud-edge collaboration technology, the system can dynamically update the model parameters in the microprocessor based on the individual characteristics of the driver after receiving real-time EEG data. This allows the system to better adapt to the differences in the original EEG data of different drivers and ensure a rapid and accurate response to stress situations. Attached Figure Description

[0055] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the train emergency braking control method and system based on brainwaves will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0056] Figure 1 This is a structural schematic diagram of the smart safety helmet provided in this embodiment;

[0057] Figure 2 This is a schematic diagram illustrating the information transmission of a train emergency braking control method and system based on electroencephalograms provided in this embodiment.

[0058] Figure 3This embodiment provides a system flowchart of a train emergency braking control method and system based on electroencephalograms.

[0059] Explanation of component symbols in the attached diagram:

[0060] 1. Train driver; 2. Smart safety helmet; 3. Circuit board.

[0061] 4. Monitoring electrode; 5. Reference electrode; 6. Adjustment rod.

[0062] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0063] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0064] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0065] To facilitate understanding of the embodiments of the present invention, two specific embodiments are provided below for further explanation, and these two embodiments do not constitute a limitation on the embodiments of the present invention.

[0066] Example 1:

[0067] This invention proposes a train emergency braking control method and system based on electroencephalogram (EEG) waves. Please refer to the following: Figure 1 , Figure 1This is a schematic diagram of the structure of a smart safety helmet provided in this embodiment. To achieve emergency braking control of a train, this invention provides a smart safety helmet for the train driver 1 that features real-time monitoring of the driver's brainwaves. The smart EEG helmet 2 includes a circuit board 3, monitoring electrodes 4, reference electrodes 5, and an adjustment rod 6. The circuit board 3 includes a TGAM (Transmission Thermoelectric Amplifier) ​​module, an STM32F103C8T6 microprocessor, and a JDY31 Bluetooth module. The TGAM module is used to collect the train driver's brainwave data in real time and transmit the preprocessed data to the STM32F103C8T6 microprocessor via Bluetooth. The STM32F103C8T6 microprocessor is used to receive, process, and monitor the brainwave data in real time. When the processed feature signal is successfully compared with the braking command feature signal stored in the database, the STM32F103C8T6 microprocessor transmits the data to the JDY31 Bluetooth module. The host sends traction resection commands; the JDY31 Bluetooth module is used to realize bidirectional data transmission between the EEG TGAM module and the STM32F103C8T6 microprocessor, and between the STM32F103C8T6 microprocessor and the host; the monitoring electrode 4 is connected to the circuit motherboard 3, and the position and tightness of the monitoring electrode 4 against the left frontal pole of the train driver 1 can be adjusted by adjusting the deformation of the adjusting rod 6, so as to monitor the EEG data of the train driver in real time; the reference electrode 5 is located on the left and right sides of the smart safety helmet, and measures the potential of the left and right ears respectively, so as to remove interference from power frequency and physiological noise.

[0068] Please refer to the following: Figure 2 , Figure 2 This is a schematic diagram illustrating the information transmission of a train emergency braking control method and system based on electroencephalograms (EEGs) provided in this embodiment. The main circuit board of the smart helmet transmits real-time EEG data and traction commands to the host computer via a JDY31 Bluetooth module. The host computer displays the working status of the smart EEG helmet and the train driver's EEG in real time. The host computer transmits the received traction cut-off commands to the train traction braking control system via a wired connection, enabling automatic braking of the train in emergency situations.

[0069] Please refer to the following: Figure 3 , Figure 3This embodiment provides a system flowchart of a train emergency braking control method and system based on brainwaves. After the train starts, the smart safety helmet begins to operate, and the TGAM module collects the driver's brainwave signals in real time and performs preliminary noise reduction. The TGAM module transmits the preprocessed data to the STM32F103C8T6 microprocessor module via the JDY31 Bluetooth module. The STM32F103C8T6 microprocessor performs wavelet decomposition on the received brainwave signals, extracting wavelet coefficients in the β and γ frequency bands. The time-domain and entropy-domain features are then input into the TensorFlow Lite model in the STM32F103C8T6 microprocessor. This model, based on the random forest classification algorithm, classifies brainwave feature signals under stress scenarios. By comparing these signals with braking command feature signals in the database, it is determined whether the driver is in an emergency braking state. When an abnormality in the driver's brainwave signal is detected—that is, when the characteristic signal successfully matches the braking command characteristic signal in the database—the STM32F103C8T6 microprocessor immediately sends a traction cut-off command to the host PC via the JDY31 Bluetooth module. The host PC then transmits the signal to the train traction and braking control system, which immediately executes emergency braking. After the train driver restarts the smart helmet device, the system resumes receiving, processing, and monitoring the driver's brainwaves.

[0070] Example 2:

[0071] This invention describes how to train a model using machine learning methods during the analysis of electroencephalogram (EEG) signals, and how to apply the model to an EEG-based smart safety helmet system.

[0072] The EEG module TGAm was used to collect EEG data from train drivers under normal driving conditions and different stress scenarios, constructing an EEG signal database. In the data preprocessing stage, wavelet decomposition was used to extract frequency domain features of the EEG signals, which were then combined with time domain and entropy domain features to form a multidimensional feature vector. A random forest classification algorithm was used to train the EEG data. By classifying stress and normal state data in the database, a model capable of accurately identifying driver stress states was trained. The model was run on the TensorFlow framework, and the final model, after optimization, was converted to TensorFlow Lite format suitable for the STM32F103C8T6 microprocessor. The model was synchronously updated online to the STM32F103C8T6 microprocessor. The STM32F103C8T6 microprocessor stored the latest 1-second, 2-second, and 3-second EEG data in three circular buffers, and used wavelet decomposition to extract wavelet domain features from the three data segments of different time lengths, combining time domain, frequency domain, and entropy domain features as feature inputs. The system, based on the analysis of real-time data by the STM32F103C8T6 microprocessor and the cloud training results provided by the host, evaluates these three sets of features, selects the data that best matches the user's brainwave signal characteristics at the time as the model input, and determines whether characteristic brainwave signals under stress are detected. In actual use, the STM32F103C8T6 microprocessor uses the received real-time brainwave data and the model to judge the driver's brainwave characteristics to determine whether to trigger an emergency braking command.

[0073] In summary, this invention, through real-time acquisition of train driver's electroencephalogram (EEG) signals and combining time-domain, frequency-domain, wavelet-domain, and entropy-domain features, utilizes a machine learning model to identify and judge stress states, constructing a mapping relationship between EEG signal features and emergency braking commands. Through data processing within the smart helmet and an external Bluetooth communication connection system, automatic traction abrogation and restoration are achieved. After multiple tests, this invention can accurately determine emergency braking needs based on the driver's EEG signals in real time, automatically sending traction abrogation commands, effectively improving the train's safety response speed and operational accuracy under stress conditions, and realizing intelligent safety control during train operation.

[0074] Furthermore, embodiments of this application also propose a computer-readable storage medium, which can be a non-volatile computer-readable storage medium, storing a behavior recognition program, which, when executed by a processor, implements the method of this application as described above.

[0075] The various embodiments of the electronic device and computer-readable storage medium of this application can be referred to the various embodiments of the behavior recognition method of this application, and will not be repeated here.

[0076] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0077] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0078] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause an electronic device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0079] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A train emergency braking control system based on electroencephalogram (EEG) waves, characterized in that, This includes a smart safety helmet worn by train drivers that monitors their brainwaves in real time, and a main unit that is wired to the train's traction and braking control system. The smart EEG cap includes a circuit board, monitoring electrodes, reference electrodes, and adjustment rods; The circuit board includes a TGAM (Transcranial Doppler) module, an STM32F103C8T6 microprocessor, and a JDY31 Bluetooth module. The EEG TGAM module is used to collect the train driver's EEG data in real time and transmit the preprocessed data to the microprocessor STM32F103C8T6 via Bluetooth module. The STM32F103C8T6 microprocessor is used for real-time reception, processing, and monitoring of electroencephalogram (EEG) data. Microcontrollers process electroencephalogram (EEG) data; The STM32F103C8T6 microprocessor performs wavelet decomposition on the received EEG signal and extracts the wavelet coefficients of the β and γ frequency bands. The STM32F103C8T6 microcontroller extracts signal features from EEG signals, trains braking command data based on machine learning, and performs real-time EEG data processing and monitoring using online training and cloud-edge collaboration technologies. Specifically, this includes: Wavelet coefficients of the β and γ bands of EEG in different scenarios are extracted from the EEG database by wavelet decomposition, and combined with the time domain, frequency domain and entropy domain features of the original EEG data, and used as input for machine learning models. Semi-supervised learning was performed on data from the EEG signal database using a random forest-based machine learning classification algorithm to extract characteristic signals of EEG under stress scenarios. The machine learning model trained by TensorFlow was then converted into a model suitable for the STM32F103C8T6 microprocessor. Features from the time domain, frequency domain, wavelet domain, and entropy domain are input into the TensorFlow Lite model in the STM32F103C8T6 microprocessor to classify EEG feature signals in real-time scenarios. Based on the results of the model operation, determine whether the monitored EEG signal data is abnormal; by comparing it with the braking command feature signals in the database, determine whether the driver is in an emergency braking state. When the processed feature signal is successfully compared with the braking command feature signal stored in the database, the STM32F103C8T6 microprocessor sends a traction cut-off command to the host via the JDY31 Bluetooth module. The JDY31 Bluetooth module is used to enable bidirectional data transmission between the EEG TGAM module and the STM32F103C8T6 microprocessor, as well as between the STM32F103C8T6 microprocessor and the host. The monitoring electrodes are connected to the main circuit board. By adjusting the deformation of the adjustment rod, the position and tightness of the monitoring electrodes against the left frontal pole of the train driver's brain can be adjusted to monitor the train driver's brainwave data in real time. Reference electrodes are located on the left and right sides of the smart safety helmet to measure the potential of the left and right ears respectively, in order to remove interference from power frequency and physiological noise; The host computer displays the train driver's brainwaves in real time and is wirelessly connected to the circuit motherboard via Bluetooth to receive real-time brainwave data and traction resection commands. The host computer transmits the received traction cut-off command to the train traction and braking control system to realize automatic braking of the train in emergency situations.

2. The train emergency braking control system based on electroencephalograms according to claim 1, characterized in that, The host also has cloud-edge collaboration capabilities, enabling it to upload real-time collected EEG data to the cloud for analysis and optimization.

3. A train emergency braking control system based on electroencephalograms according to claim 2, characterized in that, Specifically, the host computer dynamically adjusts the personalized settings of EEG characteristic signal parameters through the analysis of historical data in the cloud, thereby optimizing the generation and judgment process of traction commands, ensuring that the system can adapt to the individual differences in the driver's EEG, and improving the accuracy of traction removal in response to emergencies.

4. A control method for a train emergency braking control system based on electroencephalograms as described in any one of claims 1-3, characterized in that, Includes the following steps: Step S1: The train starts and the detection system begins operation; The circuit board in the smart safety helmet has internal data interconnection and external Bluetooth connection with the host; the host is wired to the train traction and braking control system. After the train starts, the driver puts on a safety helmet and adjusts the adjustment lever to adjust the position and tightness of the monitoring electrode of the main circuit board at the driver's forehead; the reference electrode of the main circuit board is connected to the driver's ears to ensure stable signal acquisition; Step S2: The safety helmet collects brainwave data; The smart helmet starts working, and the EEG TGAM module collects the driver's brainwave signals in real time and performs preliminary noise reduction processing. Step S3: Bluetooth data transmission; EEG data is transmitted to the STM32F103C8T6 microprocessor in real time via the JDY31 Bluetooth module. Step S4: The microcontroller processes the EEG data; The STM32F103C8T6 microprocessor performs wavelet decomposition on the received EEG signal and extracts the wavelet coefficients of the β and γ frequency bands. The STM32F103C8T6 microcontroller extracts signal features from EEG signals, trains braking command data based on machine learning, and performs real-time EEG data processing and monitoring using online training and cloud-edge collaboration technologies. Specifically, this includes: Step S41: Extract wavelet coefficients of the β and γ frequency bands of EEG in different scenarios from the EEG database through wavelet decomposition, and combine them with the time domain, frequency domain and entropy domain features of the original EEG data as input to the machine learning model; Step S42: Use a random forest-based machine learning classification algorithm to perform semi-supervised learning on the data in the EEG signal database to extract the characteristic signals of EEG under stress scenarios, and convert the machine learning model trained by TensorFlow into a model suitable for the microprocessor STM32F103C8T6. Step S5: Input the features in the time domain, frequency domain, wavelet domain, and entropy domain into the TensorFlow Lite model in the STM32F103C8T6 microprocessor to classify the EEG feature signals in the real-time scene. Step 6: Based on the results of the model run in Step 5, determine whether the monitored EEG signal data is abnormal; compare it with the braking command feature signals in the database to determine whether the driver is in an emergency braking state. If the data is abnormal, proceed to steps S7 and S8. If the data is normal, return to step S5 to continue data monitoring; Step S7: When an abnormal EEG signal of the driver is detected, that is, the characteristic signal matches the braking command characteristic signal in the database, the microprocessor STM32F103C8T6 sends a traction cut-off command to the host via the JDY31 Bluetooth module. The host transmits the traction cut-off command to the train traction and braking control system via wired transmission and immediately executes emergency braking. Step S8: After the train driver restarts the equipment for monitoring, return to step 5 to re-monitor the data. The system will then resume receiving, processing, and monitoring the train driver's brainwaves.

5. The train emergency braking control method based on electroencephalograms according to claim 4, characterized in that, Includes the following steps: In step S42, the construction of the machine learning model trained by TensorFlow includes the following steps: Step S421: Use the EEG TGAM module to collect EEG data of train drivers under normal driving and different stress scenarios, and construct an EEG signal database. Step S422, data preprocessing stage: Wavelet decomposition method is used to extract wavelet domain features of EEG signals, and time domain, frequency domain and entropy domain features are combined to form a multidimensional feature vector; Step S423: Train the EEG data using the random forest classification algorithm; Step S424: By classifying the stress state and normal state data in the database, a model that can accurately identify the driver's stress state is trained. Step S425: The model for identifying the driver's stress state in step S424 is performed in the TensorFlow framework. The final generated model is optimized and converted into TensorFlow Lite format suitable for the microprocessor STM32F103C8T6. Step S426: The TensorFlow Lite format model generated in step S425 is synchronously updated online to the microprocessor STM32F103C8T6. The microprocessor STM32F103C8T6 stores the latest 1-second, 2-second, and 3-second EEG data in three circular buffers respectively, and uses wavelet decomposition to extract wavelet domain features of the three data segments with different time lengths. In step S427, the system analyzes the real-time data based on the STM32F103C8T6 microprocessor and the cloud training results provided by the host, evaluates the three sets of features, and selects the data that best matches the user's EEG signal characteristics at that time as one of the inputs to the TensorFlow Lite format model in step S426.

Images