Intelligent sound management system and method based on physiological state of driver

Abstract

The invention discloses an intelligent sound management system and method based on the physiological state of a driver, and belongs to the technical field of automobile intelligent cabins. The system comprises a multi-source state sensing module, an audio control center, a sound execution system and a feedback learning module. The multi-source state sensing module is used for acquiring physiological, behavior and context data of a driver; the audio control center integrates the data to identify the driver state, and matches an audio scene strategy packet; the sound execution system plays audio according to the strategy packet, adjusts a three-dimensional sound field and manages active sound; and the feedback learning module optimizes the strategy according to the physiological state change. The method comprises the steps of obtaining multi-source data; identifying a state tag; matching a strategy; executing audio adjustment; and real-time feedback optimization is carried out. Through multi-dimensional physiological perception, three-dimensional sound field dynamic adjustment and closed-loop learning, the mental state of a driver is actively intervened, fatigue and pressure are relieved, driving safety and comfort are improved, and personalized learning ability is achieved.

Description

Technical Field

[0001] This invention relates to the field of automotive intelligent cockpit technology, specifically to an intelligent audio management system and method based on the driver's physiological state, which can actively adjust the driver's mental state through audio means to improve driving safety and comfort. Background Technology

[0002] Modern car audio systems have evolved from simple radio and music playback to advanced features such as multiple speakers, active noise cancellation, and independent sound field zoning. Meanwhile, Driver Monitoring Systems (DMS) are becoming increasingly common, used to detect fatigue and distraction. Currently, these two technologies operate largely independently: DMS typically issues warnings via visual or tactile feedback when it detects an anomaly; the audio system responds based on the user's manual selection of playback content or provides simple speed-sensitive volume adjustment. In recent years, the industry has begun to explore combining the two, automatically adjusting audio output based on driver status to enhance driving safety and comfort.

[0003] Publication number (CN117302083A) discloses a control system, control method, and vehicle for an intelligent cockpit. This solution uses a DMS camera to acquire images of the driver's body movements, determines the driver's mental state, and then controls the operation of the music amplifier module and air conditioning system to provide a comfortable driving environment and improve safety. However, this existing technology still has the following shortcomings: intervention methods are mainly based on music content switching, the response mode is relatively simple, it does not consider individual differences and subtle variations in the driver's state, and lacks in-depth utilization of the core capability of the audio system—three-dimensional sound field shaping; it cannot dynamically adjust spatial perception attributes such as sound field width, height, and focus position according to the driver's physiological state; the system mostly performs one-off actions, lacking the ability to continuously optimize based on feedback from the driver's physiological state after adjustment; and it has low personalization, unable to learn the driver's audio preferences and adaptively adjust strategies. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of single response mode, lack of sound field coordination adjustment, isolated system without closed loop and insufficient personalization in the linkage of car audio system and driver status monitoring system in the prior art. It provides an intelligent audio management system and method that can dynamically and finely adjust the audio content and three-dimensional sound field characteristics according to the driver's real-time physiological state.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: An intelligent audio management system based on the driver's physiological state includes: a multi-source state perception module 1, an audio control center 2, and an audio execution system 3; the output of the multi-source state perception module 1 is connected to the input of the audio control center 2; the output of the audio control center 2 is connected to the input of the audio execution system 3.

[0006] Furthermore, the multi-source state perception module 1 collects the driver's physiological data, behavioral data, and contextual data; the audio control center 2 integrates an intelligent audio management engine, which receives the data collected by the multi-source state perception module 1, performs data fusion processing to identify the driver's current state, generates corresponding state labels, and matches preset audio scenario strategy packages according to the state labels; the audio execution system 3 includes a multi-channel power amplifier module, a speaker array module, an active noise cancellation module, and an engine sound simulation module, and the audio execution system 3 performs audio playback, three-dimensional sound field adjustment, and active sound management according to the audio scenario strategy package.

[0007] Furthermore, the multi-source state perception module 1 includes: physiological sensors, behavioral sensors, and context sensors; the physiological sensors include a heart rate sensor and a heart rate variability sensor, installed on the surface of the steering wheel rim or inside the driver's seat back. The context sensors include vehicle sensors and a navigation system installed on the steering wheel or seat to collect the driver's heart rate and heart rate variability in real time; the behavioral sensors include a driver monitoring system (DMS) camera and microphone to collect the driver's facial expressions, eyelid opening and closing, gaze direction, and yawn and sigh sounds, respectively; the context sensors include vehicle sensors and a navigation system to collect vehicle speed, steering angle, time information, navigation path complexity, and traffic conditions, respectively.

[0008] Furthermore, the audio control center 2 loads the driver's historical preference model from the storage unit based on the driver's identity recognition, and makes personalized modifications to the audio scenario strategy package based on the historical preference model.

[0009] Furthermore, the audio execution system 3 also includes headrest speakers, which are used to achieve sound field focusing or cabin zone volume control.

[0010] Furthermore, the audio scenario strategy includes: an audio content selection strategy, a three-dimensional sound field adjustment strategy, and an active sound management strategy. The audio content selection strategy involves the system intelligently recommending or playing one or more of the following from audio content sources: music, white noise, natural sounds, and podcasts, based on status tags and user historical preferences. The three-dimensional sound field adjustment strategy involves the system dynamically adjusting one or more of the following: sound field focus position, sound field width and height, cabin area volume balance, and frequency equalization curve. The active sound management strategy involves the system adjusting the noise reduction intensity of the active noise cancellation module and the activation or deactivation or simulation intensity of the engine sound simulation module.

[0011] Furthermore, the system also includes a feedback learning module 4, the input of which is connected to the output of the multi-source state perception module 1, and the output of which is connected to the input of the audio control center 2. The feedback learning module 4 evaluates the effectiveness of the audio scenario strategy package based on the trend of the driver's physiological state changes after the audio execution system 3 is adjusted, and optimizes the weights or parameters of the audio scenario strategy package based on the evaluation results.

[0012] This invention also provides an intelligent audio management method based on the driver's physiological state, comprising the following steps: S1: Multi-source state perception module 1 acquires multi-source state data of the driver, including physiological data, behavioral data and contextual data; S2: Audio control center 2 performs fusion analysis on multi-source state data, identifies and quantifies the driver's current state label, which includes one or more of fatigue, stress, focus, and monotony; S3: Match the corresponding audio scenario strategy package based on the status label. The audio scenario strategy package includes at least the audio content selection strategy, the three-dimensional sound field adjustment strategy, and the active sound management strategy. S4: Based on the audio scenario strategy package, control the audio execution system 3 to play audio content, dynamically adjust sound field parameters, and perform active noise cancellation or engine sound simulation; S5: Monitors changes in the driver's physiological state in real time after adjustment, and optimizes and updates the audio scenario strategy package based on feedback.

[0013] Furthermore, the process of identifying status tags includes: S201: Input heart rate, heart rate variability, facial features, and driving behavior data into a pre-trained state recognition model; S202: Output the quantized state label and its confidence level; S203: Modify or weight the status label by combining context information to generate the final status label.

[0014] Compared with traditional solutions, the present invention has the following advantages: (1) This invention effectively relieves fatigue, stress and monotony through precise and flexible audio intervention, helps drivers maintain a better mental state, and reduces the risk of accidents from the perspective of human factors engineering.

[0015] (2) The present invention can provide an unprecedented intelligent acoustic environment that changes dynamically with the state, and upgrade the audio system from a functional component to a core intelligent cockpit experience that is emotional and caring.

[0016] (3) The system of the present invention has learning ability and can customize exclusive audio playback solutions for different drivers, which greatly improves user satisfaction.

[0017] (4) This invention deeply explores and integrates the potential of existing hardware such as DMS, HR sensor and high-end audio, and realizes high added value functions through software and algorithm innovation, with controllable cost increment. Attached Figure Description

[0018] This manual includes the following figures, which illustrate the following: Figure 1 This is a schematic diagram of the intelligent audio management system based on the driver's physiological state according to the present invention.

[0019] Figure 2 This is a flowchart illustrating the intelligent audio management method based on the driver's physiological state according to the present invention.

[0020] The system includes: 1. Multi-source state perception module; 2. Audio control center; 3. Audio execution system; 4. Feedback learning module. Detailed Implementation

[0021] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, in order to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention, and to facilitate its implementation.

[0022] like Figure 1 As shown, the intelligent audio management system based on the driver's physiological state of the present invention includes: a multi-source state perception module 1, an audio control center 2, an audio execution system 3, and a feedback learning module 4.

[0023] The multi-source state perception module 1 is used to collect the driver's physiological data, behavioral data, and contextual data. Specifically, it includes physiological sensors, behavioral sensors, and contextual sensors. The physiological sensors include a heart rate sensor and a heart rate variability sensor, installed on the surface of the steering wheel rim or inside the driver's seat back. The heart rate sensor uses the photoplethysmography (PPG) principle, emitting green or infrared light to irradiate the skin and detecting changes in blood volume to measure the driver's heart rate (HR) in real time. Heart rate variability (HRV) is calculated by analyzing the time changes between consecutive heartbeats and reflects the activity of the autonomic nervous system, serving as an important indicator of fatigue and stress. The physiological sensors transmit the collected data to the audio control center 2 via a CAN bus or in-vehicle Ethernet at a certain frequency, such as once per second.

[0024] The behavioral sensors include a Driver Monitoring System (DMS) camera and a microphone. The DMS camera is mounted near the rearview mirror or A-pillar, with its lens facing the driver's face to continuously capture facial images. Through a built-in image processing chip or an algorithm run by the audio control center 2, features such as facial expressions, eyelid opening and closing, gaze direction, and head posture can be extracted. The microphone is mounted on the ceiling or A-pillar in the cockpit to capture sounds emitted by the driver, such as yawns, sighs, and voices, and uses voice recognition algorithms to detect acoustic features related to fatigue or emotions.

[0025] Context sensors include vehicle sensors and the navigation system. Vehicle sensors acquire data such as vehicle speed, steering angle, braking, accelerator pedal opening, and lateral acceleration via the CAN bus. The navigation system provides information such as navigation route complexity, real-time traffic conditions, remaining travel time, and current time. This data helps determine the driving scenario and driver status.

[0026] The audio control center 2 is the core processing unit of this system, integrated into the intelligent cockpit controller, and includes a processor, memory, and communication interface. It integrates an intelligent audio management engine, which exists as a software module. This engine is responsible for receiving data transmitted from the multi-source state perception module 1, performing data fusion, state recognition, strategy matching, and generating control commands. The input of the audio control center 2 is connected to the output of the multi-source state perception module 1 via a CAN bus or vehicle Ethernet; its output is connected to the input of the audio execution system 3 via an audio bus or low-voltage differential signal interface.

[0027] The audio system 3 includes a multi-channel amplifier module, a speaker array module, an active noise cancellation module, and an engine sound simulation module. The multi-channel amplifier module has multiple independent amplification channels, each capable of driving different speaker units. The speaker array module includes multiple speakers distributed in different locations within the vehicle cabin, such as front door subwoofers, rear door full-range speakers, a center speaker, surround speakers, and ceiling speakers. The system also includes headrest speakers installed in the driver's seat headrest for sound field focusing or cabin zone volume control. The active noise cancellation module collects in-vehicle noise through microphones and uses the amplifier to emit inverse sound waves through the speakers to cancel out noise in specific frequency bands. The engine sound simulation module simulates the sound of a traditional engine in electric vehicle or energy-saving mode. The audio system 3 receives commands from the audio control center 2, selects audio content from sources such as the local music library, online streaming media, and radio, and controls the speaker array module to play the audio via the multi-channel amplifier, while simultaneously adjusting the operating status of the active noise cancellation and engine sound simulation modules.

[0028] The feedback learning module 4 can be integrated into the audio control center 2 or used as a separate hardware module. Its input is connected to the output of the multi-source state perception module 1, and its output is connected to the optimization input of the audio control center 2. The feedback learning module 4 is responsible for continuously monitoring the driver's physiological state changes after the audio adjustment is executed, evaluating the adjustment effect, and optimizing the weights or parameters of the audio scenario strategy package accordingly to achieve closed-loop learning.

[0029] Figure 2 The flowchart of the intelligent audio management method based on the driver's physiological state according to the present invention includes the following steps: S1: After the system starts, the multi-source state sensing module 1 continuously collects the following data at a fixed sampling frequency of 50Hz: Physiological data: including heart rate (HR) and heart rate variability (HRV); Behavioral data includes facial image sequences captured by the DMS camera, which are processed to obtain PERCLOS (eyelid opening and closing), blinking frequency, yawning frequency, head posture, and facial expressions. The emotion recognition model outputs pleasure level and activation level. Audio signals captured by the microphone are used to identify yawning sounds, sighing sounds, and talking sounds through the sound event detection model. Contextual data includes vehicle speed, steering angle, brake / accelerator pedal status, navigation route complexity, real-time traffic conditions, current time, and driving duration.

[0030] After preprocessing, all data is packaged and sent to the intelligent audio management engine of the audio control center 2.

[0031] S2: The intelligent audio management engine fuses and analyzes the received multi-source data to identify and quantify the driver's current state label. The specific process is as follows: S201: Extract HRV frequency domain features such as LF / HF and time domain features such as RMSSD from physiological data; extract PERCLOS mean, blink frequency, yawn frequency, head posture entropy, etc. from behavioral data; extract vehicle speed variance, steering angle entropy, driving duration, etc. from contextual data. Combine the above features into a multi-dimensional feature vector and input it into a pre-trained state recognition model.

[0032] S202: The state recognition model can employ deep learning classifiers such as Convolutional Neural Networks (CNNs) combined with Long Short-Term Memory (LSTM) networks, or traditional machine learning classifiers such as Random Forests and Support Vector Machines. The model outputs the probability or confidence score for each possible state label, for example: fatigue 0.85, stress 0.12, focus 0.03, monotony 0.00. In this embodiment, state labels include, but are not limited to: fatigue, stress, focus, and monotony.

[0033] S203: Adjust or weight the confidence score of the model output by incorporating contextual information. For example, in scenarios involving long-distance driving at night with empty roads, the weight of fatigue states can be increased even if fatigue characteristics are not obvious; in traffic congestion, the weight of stress states can be appropriately increased; and in high-speed cruising with simple routes, the weight of monotonic states can be increased. The adjusted state labels and their confidence scores serve as the final recognition results for subsequent policy matching.

[0034] S3: The audio control center 2 has multiple pre-installed audio scenario policy packages in its memory. Each policy package targets one or more combinations of status tags and contains a complete set of audio adjustment instructions. The system also supports user-defined policy packages. The policy packages mainly include three types of policies: audio content selection policies, three-dimensional sound field adjustment policies, and active sound management policies.

[0035] The audio content selection strategy intelligently recommends or plays specific types of audio content from audio content sources based on status tags and user history preferences. When fatigued, it plays upbeat music with a tempo of 120-140 BPM, or audiobooks, news broadcasts, and other audio that requires cognitive engagement; when stressed, it plays soothing classical music, piano pieces, natural sounds, or white noise; when focused, it plays wordless background music or ambient music to avoid distractions; and when bored, it plays podcasts, talk shows, audiobooks, or periodically switches music genres to maintain a sense of novelty.

[0036] The three-dimensional sound field adjustment strategy can dynamically adjust the spatial perception attributes of the sound field, including the sound field focus position, sound field width and height, cabin area volume balance, and frequency equalization curve. By adjusting the delay and gain of each speaker channel, the perceived position of the sound source can be moved horizontally and vertically, thereby adjusting the sound field focus position. For example, the sound field can be focused in front of, behind, to the left and right of the driver's head, or slowly moved within a certain range. By adjusting the gain and phase of the surround channels, the sound field width and height can be adjusted, making the sound field feel wider or more focused. For example, increasing the gain of the side surround and rear surround channels can expand the sound field width; increasing the gain of the ceiling speakers or the speakers at the windshield can improve the perceived height of the sound field. By independently controlling the speaker volume in different seating areas, zoned sound fields can be achieved. For example, music can be played only in the driver's area, while the passenger area remains silent or at a low volume. The gain of each frequency band can be adjusted according to the state. For example, when fatigued, the mid-high frequencies can be appropriately boosted to refresh the mind; when stressed, the low frequencies can be attenuated to reduce the feeling of pressure; and a flat response can be maintained when focused.

[0037] The active sound management strategy can adjust the noise reduction intensity of the active noise cancellation module and the activation or deactivation of the engine sound simulation module. When under high pressure or when a quiet environment is needed, the active noise cancellation intensity is increased and the engine sound simulation is reduced; when driving aggressively or needing to stimulate the driver's energy, the active noise cancellation is weakened, the engine sound simulation is increased, and a sporty sound is selected.

[0038] When matching strategies, the audio control center 2 first identifies the driver, for example through facial recognition, key ID, or Bluetooth connection to a mobile phone, and loads the driver's historical preference model from the storage unit. The historical preference model records the user's feedback on audio content and sound field adjustment under different states. Based on the historical preference model, the system personalizes the general strategy package and generates the final execution command. For example, even if a user is fatigued, but their historical preferences show that they dislike fast-paced music and prefer audiobooks, the system will prioritize audiobooks over music.

[0039] The construction and maintenance process of the historical preference model is as follows: Upon initial system use or user's first login, the audio control center 2 establishes the user's initial preference model through initialization settings or a brief preference guidance questionnaire. The initial model can be built based on the user's selected music style, sound field preference, and feedback on preset scenario strategies. As the driving process continues, the system continuously enriches and corrects the model by combining physiological and behavioral data collected by the multi-source state perception module 1 with the user's feedback after audio adjustment. The data of the historical preference model includes at least the following: driver identification, state label, audio content type, three-dimensional sound field parameters, active sound management parameters, and the user's rating or feedback weight for the strategy combination. The model update mechanism adopts an incremental learning approach. After each driving cycle, the feedback learning module 4 uses the user's physiological change trends and feedback collected during the adjustment process as training samples, and optimizes the weights of the historical preference model through collaborative filtering or deep reinforcement learning algorithms, enabling the model to dynamically adapt to the evolution of user preferences. In addition, the system supports multi-user configuration. When different drivers are identified, the corresponding historical preference model is automatically loaded to achieve seamless switching of personalized strategies.

[0040] S4: The audio control center 2 sends the generated execution commands to the audio execution system 3. The multi-channel power amplifier reads the audio stream from the designated audio source according to the commands, performs decoding and digital-to-analog conversion, and then drives the speaker array to produce sound through each channel. Simultaneously, the power amplifier's internal digital signal processor (DSP) performs delay, gain, and equalization processing on each channel according to the commands, achieving three-dimensional sound field adjustment. The active noise cancellation module adjusts the parameters of the noise cancellation filter according to the commands, changing the noise reduction depth and frequency band; the engine sound simulation module selects the sound type and volume according to the commands.

[0041] S5: During and after audio adjustment, the system continues to monitor the driver's physiological state changes through the multi-source state perception module 1. The feedback learning module 4 collects physiological data over a period of time after adjustment and calculates state improvement indicators, such as the percentage decrease in stress index and the degree of improvement in fatigue level. If the improvement indicators exceed a preset threshold, the current strategy is considered effective. The feedback learning module 4 strengthens the association between the strategy package and the current state label, increases the weight of the strategy package, and makes it easier to select in the future. If the improvement is not significant or the state deteriorates, the system attempts to fine-tune the strategy parameters, such as changing the music type, adjusting the sound field position, or switching to an alternative strategy package, and records the effect after the switch. In addition, users can provide explicit feedback on the currently playing content and sound field effects through voice commands or the central control screen, indicating whether they "like" or "dislike" them. The feedback learning module 4 also uses this feedback as a learning basis, continuously updating the user's personalized preference model through collaborative filtering or deep learning algorithms.

[0042] In this invention, the headrest speaker plays a special role. In scenarios requiring sound field focusing, the headrest speaker can cooperate with other speakers to achieve more precise sound image localization. For example, when the system wants to deliver sound only to the driver without disturbing passengers, the audio control center 2 can control the headrest speaker to play the main audio, while other speakers play only low-volume ambient sounds or remain completely silent. In fatigue scenarios, when the headrest speaker is used in conjunction with a sound field forward-shifting strategy, the directionality of warning sounds can be enhanced to avoid startling passengers. In a sound field oscillation strategy, the headrest speaker can cooperate with the left and right front door speakers to achieve a smooth movement of the sound image between the left ear, the center of the head, and the right ear, creating a unique spatial auditory experience.

[0043] This embodiment uses physiological sensors to collect heart rate and heart rate variability, behavioral sensors to collect facial expressions, eyelid opening and closing, gaze direction, yawn sounds, and other features, and context sensors to collect environmental information such as vehicle speed, steering, navigation path complexity, and traffic conditions, achieving a comprehensive perception of the driver's state. The fusion analysis of multi-source data overcomes the shortcomings of single sensors, such as susceptibility to interference and low recognition accuracy, and can accurately distinguish between different states such as fatigue, stress, focus, and monotony, providing a reliable basis for subsequent precise intervention. For example, by combining multi-dimensional features such as decreased HRV, increased PERCLOS, and driving time exceeding 2 hours, the system can reliably identify that the driver has entered a state of fatigue, rather than misjudging it as a brief distraction.

[0044] This embodiment utilizes a digital signal processor (DSP) to precisely control the delay, gain, and equalization of each speaker channel. The system can dynamically adjust the sound field focus position, width, and height, as well as the volume balance and frequency equalization curves within the cabin area, directly influencing the driver's psychological experience from a spatial auditory perspective. For example, when fatigued, the sound field is shifted forward and narrowed while boosting mid-to-high frequencies, making the sound more impactful and invigorating; when stressed, the sound field is expanded and surround sound is enhanced with active noise cancellation, creating a soothing and quiet environment. This spatial auditory intervention is impossible to achieve with traditional single-content switching, significantly enhancing the effect of audio adjustment.

[0045] The feedback learning module continuously monitors the driver's physiological state changes after intervention, quantitatively assesses the intervention's effectiveness, and dynamically adjusts strategy weights and parameters based on the assessment results. If the intervention is effective, the strategy is reinforced; if it is ineffective, parameters are automatically fine-tuned or alternative strategies are switched. This mechanism enables the system to have adaptive learning capabilities, allowing it to continuously optimize based on different drivers, scenarios, and times, avoiding the limitations of one-off actions. For example, if a driver responds well to rock music, the system will increase the frequency of rock music use when fatigued; if a driver is not comfortable with sound field fluctuations, the system will gradually reduce the application of that strategy.

[0046] This embodiment achieves deeply personalized services through driver identification and historical preference models. The system records user feedback and preferences in different states. With the accumulation of usage time, the system can customize exclusive plans for different drivers, making audio intervention more tailored to individual needs and avoiding the single-response mode of existing technologies, which can cause user resentment.

[0047] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution; or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.

Claims

1. An intelligent audio management system based on the driver's physiological state, characterized in that, include: The system comprises a multi-source state perception module (1), an audio control center (2), and an audio execution system (3); the output of the multi-source state perception module (1) is connected to the input of the audio control center (2); and the output of the audio control center (2) is connected to the input of the audio execution system (3).

2. The intelligent audio management system based on driver's physiological state according to claim 1, characterized in that, The multi-source state perception module (1) collects the driver's physiological data, behavioral data, and contextual data; the audio control center (2) integrates an intelligent audio management engine, which receives the data collected by the multi-source state perception module (1), performs fusion processing on the data to identify the driver's current state, generates corresponding state labels, and matches preset audio scenario strategy packages according to the state labels; the audio execution system (3) includes a multi-channel power amplifier module, a speaker array module, an active noise cancellation module, and an engine sound simulation module, and the audio execution system (3) performs audio playback, three-dimensional sound field adjustment, and active sound management according to the audio scenario strategy package.

3. The intelligent audio management system based on the driver's physiological state according to claim 1, characterized in that, The multi-source state perception module (1) includes: a physiological sensor, a behavioral sensor, and a context sensor; the physiological sensor includes a heart rate sensor and a heart rate variability sensor, installed on the surface of the steering wheel rim or inside the driver's seat back. The context sensor includes a vehicle sensor and a navigation system installed on the steering wheel or seat to collect the driver's heart rate and heart rate variability in real time; the behavioral sensor includes a driver monitoring system camera and microphone to collect the driver's facial expressions, eyelid opening and closing, gaze direction, and yawning and sighing sounds, respectively; the context sensor includes a vehicle sensor and a navigation system to collect vehicle speed, steering angle, time information, navigation path complexity, and traffic conditions, respectively.

4. The intelligent audio management system based on the driver's physiological state according to claim 1, characterized in that, The audio control center (2) loads the driver's historical preference model from the storage unit based on the driver's identity recognition, and makes personalized modifications to the audio scenario strategy package based on the historical preference model.

5. The intelligent audio management system based on driver's physiological state according to claim 1, characterized in that, The audio execution system (3) also includes a headrest speaker, which is used to achieve sound field focusing or cabin zone volume control.

6. The intelligent audio management system based on driver physiological state according to claim 1, characterized in that, The audio scenario strategy includes: an audio content selection strategy, a three-dimensional sound field adjustment strategy, and an active sound management strategy. The audio content selection strategy involves the system intelligently recommending or playing one or more of the following from audio content sources: music, white noise, natural sounds, and podcasts, based on the status tags and user historical preferences. The three-dimensional sound field adjustment strategy involves the system dynamically adjusting one or more of the following: sound field focus position, sound field width and height, cabin area volume balance, and frequency equalization curve. The active sound management strategy involves the system adjusting the noise reduction intensity of the active noise cancellation module and the activation or deactivation or simulation intensity of the engine sound simulation module.

7. The intelligent audio management system based on the driver's physiological state according to claim 1, characterized in that, The system also includes a feedback learning module (4), the input of which is connected to the output of the multi-source state perception module (1), and the output of which is connected to the input of the audio control center (2). The feedback learning module (4) evaluates the effectiveness of the audio scenario strategy package based on the trend of the driver's physiological state change after the audio execution system (3) is adjusted, and optimizes the weights or parameters of the audio scenario strategy package based on the evaluation results.

8. A management method for an intelligent audio management system based on driver physiological state according to any one of claims 1-7, characterized in that, Includes the following steps: S1: The multi-source state perception module (1) acquires the driver's multi-source state data, which includes physiological data, behavioral data and contextual data; S2: The audio control center (2) performs fusion analysis on the multi-source state data, identifies and quantifies the driver's current state label, which includes one or more of fatigue, stress, focus, and monotony; S3: Match the corresponding audio scenario strategy package according to the status label. The audio scenario strategy package includes at least an audio content selection strategy, a three-dimensional sound field adjustment strategy, and an active sound management strategy. S4: According to the audio scenario strategy package, control the audio execution system (3) to play audio content, dynamically adjust sound field parameters, and perform active noise reduction or engine sound simulation; S5: Monitor the driver's physiological state changes in real time after adjustment, and optimize and update the audio scenario strategy package based on feedback.

9. The intelligent audio management method based on the driver's physiological state according to claim 8, characterized in that, The process of identifying the status label includes: S201: Input heart rate, heart rate variability, facial features, and driving behavior data into a pre-trained state recognition model; S202: Output the quantized state label and its confidence level; S203: Modify or weight the status label by combining context information to generate the final status label.

Images